I would like to apologise to Shaun @ F-stoppers - I said in this video that they agreed with Tony's findings, however I now realise that isn't the case. I misunderstood their interpretation of what Tony was suggesting and believed they were fully agreeing with him, having watched the video again since I realise they are actually more in line with what I am suggesting here. Sorry F-stoppers
I was going to raise that as the F-Stopper video did report visible differences between the underexposing 5 stops then boosting by 5 stops, even if they didn't know why.
I look forward to seeing that. I am also aiming to put together a more comprehensive video looking at how the different kind of ISO works, hopefully we can all settle this debate 😁
@@FStoppers I do wonder if there's a difference in how the image is processed. For example, I know the Sony Alpha series does noise reduction even on RAW files.
The biggest problem with this topic when trying to simplify it into a RUclips video is how it disregards just how camera-dependent this topic is. Tony went hard in one direction and I feel you might have gone a little too hard in other. For some cameras ISO is almost entirely digital gain and might as well be a metatag, because you can change it in post all day. For other cameras like many Canon models, changing ISO is not reversible in post and has serious consequences. For other cameras still, it's a hybrid where it's analog to a certain ISO and then a larger digital component at higher ISOs, and then fully digital in extended ISOs. The list goes on and on. The a7 III is a fun example because it does demonstrate invariance but at different steps like you mentioned. There's a lower stage from 100-640ish, and then from 800 onward. My point is it's all too camera-dependent to be making any claims about ISO in general, unless the claim is look how funky ISO is. And so, I didn't mind Tony's video because I felt like that was his key point. I also enjoyed your video because your counterpoints further demonstrated that it's not as simple as most would think. Cheers! 👍😃
Thanks Gerald, that means a lot. I know it's a complex, camera dependent topic, and yes I probably did go a bit far in the opposite direction to try and balance out what Tony was saying. But the can of worms has been opened now 🤣 so think I might go for a more in depth video looking at the numerous different options and how they work
@@DaveMcKeegan Here's a resource that might help you in your journey: www.photonstophotos.net/Charts/PDR_Shadow.htm#Sony%20ILCE-7M3 This one's for the a7 III, but they've charted a ton of cameras. Compare that to a Canon and you'll see what's going on.
@@DaveMcKeegan Please do! I really dig this information and like Gerald mentioned it seems like ISO is generally all over the place - These videos help us all at least try to understand what we're messing with. Cheers
Coming from an audio background, I think iso is best understood as "gain". The lower the signal, the more the noise will be visible/audible when amping it. The lower your iso, the higher your signal to noise ratio. In audio, you try to get your signal input as high as possible without peaking. Basically with iso it's exactly the same. Only it's things like camera shake you're trying to avoid as well.
Most video cameras used gain. Now, with the ISO craze, they have the option to use gain or ISO values. This bothers me a bit because every camera has a different "native ISO", so knowing X camera has the least grain at ISO Z, doesn't mean Y camera will have the lowest grain at ISO Z. But, they'll both have the lowest grain at 0db gain...
In olden days, you could set the ISO setting, shutter and aperture based on a light meter based on film sensors and get a reasonable result. So, I'm not really sure what's changed. Although, it is worth noting that it never quite worked because digital sensors tend to blow out highlights in colors of red, yellow or magenta more so than film would. And you'd need to manually account for that if you've got any of those colors being directly lit, or you could wind up with unexpectedly blow areas in an image.
@@SmallSpoonBrigade the difference is: on film, the ISO is the actual light sensitivity on the film. Changing the ISO just told the camera what film is loaded. All you could do was push or pull with a "wrong" setting. Now, the ISO setting boosts the Signal from the base iso, kind of like the gain on an amplifier. You still really only have your base ISO unless you have a camera with multiple base values (forgotten the term). But that's the reason why you have noise: the lower the light the higher the signal to noise ratio on your sensor, the more noise you get by bumping it up.
@@odemata87 I don't think that most of the youtube photographers actually consider themselves to be great photographers. Jared Polin doesn't, the F-Stoppers don't, Matt Granger doesn't and the Northrup family consider themselves to be generalists without a particular expertise. So it's all good.
No one really cares about these topic except the tech geeks and influencers who need to have a "topic" to talk about and generate view count. In short - each camera is different and each camera has different technology and different tweaking made to them.
Many exceptional photos are taken on fully automatic settings too. If you just remember to use the lowest gain/ISO settings and expose using shutter speed and aperture as much as possible, you will be fine.
5 лет назад+4
@@clifftotten7609: Your post is confusing. "... expose using shutter speed and aperture as much as possible..." does not match with "Many exceptional photos are taken on fully automatic settings too."
Im pretty sure its not just the exposure the camera changes. Look at mobile cameras today. Software has already come a long way. How hard would it be to make one for a bigger camera?
@@melvinch Except attacking, trolling, being nasty and disrespectful - what is your purpose in this? I never ever read any explanation, argument, proper grown-up discussion fro your side. You are just nasty, petty wanna be "photographer" (even though I doubt it), who is seeking his 5 seconds of glory on forums. Your work and contribution are "fabulous" here and I will definitely try to find a way to get you of this social media.
This topic is something astrophotographers are well aware of since we are almost exclusively dealing with very low levels of signal and have to be very aware of what effect ISO has on noise. I think you explained the concepts very well for technically-minded beginners. What you are discussing starting at 04:10 is what astrophotographers call "read noise". Changing the ISO has typically been a balancing game between not increasing noise by setting the ISO too low, but maintaining enough dynamic range that one doesn't cripple color in brighter areas (for astrophotographers the stars). ISO-invariant cameras have been amazing because it means the ability to shoot at ISO 400 (more dynamic range) with still very low added read noise. I don't know of any serious astrophotographers shooting at ISO100 though for the reason you mentioned: no DSLR is truly "ISO-less" as that would mean zero read noise. My 2 cents.
"Out and out wrong" is too strong. After I saw Tony's video, I took some pictures and tested his results. As far as I was concerned, there was no noticeable difference and it made a huge difference to my thinking, which was treating ISO as thought I had a film camera. Even though you are right, you should be praising Tony for highlighting this and many more things like the true meaning of f1.8 stuck on the front of a camera. Yes I fell into that trap. f1.8 wasn't f1.8 and the camera manufactures should be shot for misleading people. Lying.
The problem is that not every iso step applies the same formula for analog gain/digital gain and this is different for every camera. Sometimes going from 200 to 400 applies digital gain and going from 400 to 800 dials back the digital gain and applies analog gain, thus the noise pattern in the shadows changes. Ideally you should test every iso in your camera to find the ones without digital gain and stick to those. Not only will it preserve dynamic range, but it'll force you to light your shots and/or choose an aperture that allows your camera to perform at it's best..
This is what I am curious about, thank you. so do you have a testing method for this? Or is there one provided some where and or do the camera manufacturers state their ideal iso setting? (should i set up a colour chart and photograph it with a metered setting to start and then go through my iso setting then apply post and see what setting needed the least amount of adjustment?
I'm an electrical engineer and I do signal processing. Your video is very correct. Amplifying a signal before it's processed (known in the business as "sensor adaptation") is indeed going to prevent having to amplify the noise added by the whole processing chain. What Tony stated is equivalent to saying a digital zoom is just as good as an optical zoom : it's not, because the optical zoom scales up the picture before any processing, whereas the digital zoom must create pixels for which the original signal doesn't contain any data. Tony is a good source of a advice, but this is RUclips, world HQ of self-proclaimed experts. No matter how good someone seems to be, you had better double-check anything they say. Feel free to include me in the lot.
Hi Jean. In any precision electronic readout, you would like to be limited by the sensor's noise and get the most dynamic range out of it with the least amount of noise. So let's get paint this picture, the sensor generates some signal and some noise, and by default, we get a 8-bit signal out in the final picture. Which option is better: -Building a front end 2x amplifier that an input referred noise of half of the sensor and an 8 ENOB AD converter. -Build a 9 ENOB AD converter with an input referred noise of half of the sensor and decimate one of the bits for a 2x amplification in the digital domain. In both cases, we would end up with the same image in terms of SNR, and if we do smart choices for the AD design you can save significant area and power. If you check, manufacturers use 12-14bit SNR a/d converters, but each pixel generates a 8 bit information... that is just digital amplification, and if the noise level is sensor limited you are able to digitally "amplify" that information without losing any information. And that is pretty in accord to the image that the guy show, from ISO 100 to 12800 is around 16x but the SNR is not nearly 16x worse... one can quantize it but it is barely 2x worse, wich means you are 1 bit off in you adc... The digital zoom analogy is a bit off... if you happened to have a digital capture that is 2x better than what your eyes can even imagine to ever get, a 2x zoom would not result in any loss of information for you. And that is one point that the manufacturers play around, but with sensor noise. In the end manufacturers do play around a lot with pre amplification-dynamic range-digital quantization to get what they look for with conversion time/power/area/flexibility/SNR.
@@jethro606 you are splitting hairs my friend. I understand what you wrote, but who else does ? My comment wasn't aim at fellow engineers and in the strictest sense, you have to agree I'm correct. Taking into account parameters like the limitations of the human eye, for example, isn't really what we're discussing here 😉
I'm an electronics designer, specializing in mixed signal design and data acquisition. When I watched Tonys' Video I thought exactly the same... naaah there is no way that's true... I said to myself either there is some biasing on the sensor and the bias level is tuned for the ISO setting or there is a PGA in front of the ADC to sort of "auto adjust" the signal or set based on the ISO settings... Then i looked it up ... at least from the info i could find nope... nothing.. nada... of the cameras i could find some sort of block diagram or schematic there is no such stage... it is possible that Tony is correct... at least for some cameras
@@TheNefastor Hey, I know what I commented was not for everyone to understand, but I guess you would and I am fine with that. Also I would like to attract more of other engineers attention so they could correct me in the case. But if you got my point you would see that a pre amplifier is not always necessary. And like the other fellow said, you don't always find a front end PGA in such sensors because the ADC already bring noise levels so low that doing post digital amplification or analog amplification has nearly 0 difference. And that is what both guys showed in their videos, amplifying a signal 16x and not seeing 16x extra noise sort of means that.
In other words: For best results, know your camera. Thanks for the great explanation. The most important part of the video is one some people seem to be missing. Yes, ISO is applied gain, after the light hits the sensor (as Ken Wheeler might say) but it's applied gain BEFORE it goes to the converters and then the processor,, thereby amplifying it at it's purest state, without amplifying any of the noise associated with said converters and processor. Is that correct?
That is true. But when the noise associated with converter and processor gets very very minimal, amplifying the signal before or after becomes almost irrelevant. That's the whole idea of ISO invariance.
I think Tony’s point still stands. FStoppers did actually see the colour shifts that you mention, but came to the conclusion (which I think is right) that for the difference seen the results are basically the same after post processing. I think Tony’s point was to educate people here and show them something that they may never have realised. It certainly opened my eyes on this. I don’t think Tony suddenly expects everyone to suddenly go out and shoot at ISO100 for a wedding reception without flash. His point is theoretical, and interesting.
Except ISO and post processing are fundamentally different because one works on an analogue signal while one works on a digital ISO is like the preamps in audio equipment which will always yield better results than turning up the volume afterwards
The ISO values should be adjusted to reflect the sensor’s light transmission factor. This is what the ISO measurement is supposed to measure. ISO is supposed to be a standard for how much light the sensor captures, AFTER TRANSMISSION FACTOR. Also, the point of the RAW format is to capture exactly the readout from the sensor, so the AD converter output should be identical between different ISO values, if the camera circuitry isn’t applying electrical amplification. If they are, it would seem of critical importance that the amps are very high quality. If they are not high quality, bumping up the exposure in software will be similar in noise level, but waveform-clipping artifacts would still be a problem.
This is where I'm a bit skeptical. If the ISO we get in the camera is just in relation to the baseline for that particular sensor, then it won't be possible to use external meters without manually adjusting the readings, but I don't think that's the case. When I used to use an incidence light meter, I don't recall needing to adjust the output to compensate for the fact that I was shooting on a digital sensor.
I’m a beginner and what I got out of his video is DONT TRUST THE ISO SPECS ON A CAMERA SO MUCH !! THE STANDARD IS NO LONGER A STANDARD!! And I believe he’s talking to people like me the video wasn’t intended to school professionals !!! Of course it’s more technical in depth! Don’t get so defensive. He didn’t insist the earth is flat !
You missed the main point (while making a few good points). Tony is *primarily* pointing out that camera manufacturers exploit the inaccuracy (and non-standard lack of uniformity) of ISO as a unit of sensitivity measurement; thereby allowing for more flattering ISO specification declarations. It is largely misleading, and that's what he's pointing out.
Of course, but how many people know this, esp. beginners? The discussion here generated by Tony's video enlightened me about some cameras using a different tpye of amplifications at different ISO settings; before I thought it was a straight linear amplification. The manufacturers should be more open and tell us more about how they produce their raw files and JEPG's.
@@DaveMcKeegan Yes! Also stated by Tony (re: "always been like this"). But I still think reviewers such as yourself keep peeps honest by monitoring the fringes of speech - so keep up the good work! :-)
Kind off you to speak for Tony. I am of the opinion that if Tony had wanted to say that he would have. But he didn't. He deserves full credit for obfuscation, inaccuracy and ignorance.
Also cameras of same make and model can vary. They are not 100% equally due to sensors are made of organic materials, if I don’t remember wrong. Can also differ when they where made.
ISO don’t care! Do we really think that camera manufacturers put a settings on the camera just so we could mess with one more knob? Think about it. This whole conversation is really out there. This is what I do know; Set the ISO as low as you can and shoot for correctly exposed image; Know your camera and know it’s limitations; Settings between cameras and settings even between models will vary; If it’s dark increase your shutter speed and open your f-stop and when you run out increase the ISO; You get what the light gives you and what the manufacturer provides. Nikon, Canon, Fuji and Sony all want you to be able to take the best picture possible. If the best solution was post software then they would provide that as a proprietary means of pushing your image. And they would drop the ISO setting. I don’t mind a good technology discussion, but this one is just absurd... to me!
Below I have two articles. I have summarized information from the Stanford article so nearly all credit should be given for the below summary to them. I’ve added my own explanations to some parts however. The number of photons arriving on a pixel for any given exposure is governed by the Poisson Distribution. From this we get a mean pixel value and a standard deviation pixel value. The signal to noise ratio or SNR = (mean pixel value)/(standard deviation of pixel value) SNR(dB) = 20*log10*(mean /sigma ) Therefore, doubling the width and height of a pixel increases its area by 4×, hence # of photons by 4×, hence SNR by 2× or +6 dB Not all photons will produce an electron. Quantum Efficiency is a number we can calculate as Q.E = #electrons/#of photons For a typical digital camera this will be less than 50%. For back side illuminated sensors this will be greater since the wiring doesn’t obstruct photons. Larger pixels allow for longer measurements of small currents to be made and therefore a more accurate measurement. It is reasonable to say that a full frame sensor with let’s say 24 megapixels will generate a larger signal to noise ratio than that of a crop sensor with 24 megapixels and thus less noise. This is assuming that the tech in the signal chain after the sensor i.e Amplifier, ADC etc is the same and therefore the noise contributions from the rest of the signal chain are the same. This is of course assuming the photo sites on the 24 megapixel camera are larger which would naturally be the case since they would be expanded to fit the larger surface area of a full frame sensor vs the crop sensor. Pre gain stage before the ADC is typically an analog gain amplifier. If I assumed an inverting amplifier the gain equation is: Vout=-Rf/Rin*Vin Vin would be the signal of interest plus the inherent noise at the input of the amplifier pins (1/f noise, broadband noise etc). This would be where the base iso would come from and the noise would vary depending on the amplifier of choice. The ratio of resistors chosen would determine the gain of the signal Vin and maybe be a “standard gain” whereas the quality of the amplifier would determine the noise level. One other option is for them to use a PGA (programmable gain amplifier) with values 10,100, 1000 (pre ADC). Post ADC the signal can now be infinitely gained up digitally therefore there’s no limit as you said. Citations and References: pdfs.semanticscholar.org/e0af/181fd692ef06b7a114948164c399b8869be0.pdf graphics.stanford.edu/courses/cs448a-10/sensors-noise-14jan10-opt.pdf
Tony's main point seems to be that different cameras implement different standards for ISO. This doesn't make ISO fake. It really doesn't matter. Once you become familiar with your camera's ISO standard, you can use it to produce properly exposed pictures. Most modern cameras have an auto ISO that will nail the exposure.
The real lesson here is never substitute gain for light, whether through ISO or in post. Choose slower shutter over higher ISO if the situation allows it. No?
What few realize is that in the film days, ISO was changed by using smaller or larger photosensitive particles. Higher ISO would be "more sensitive" because larger grains would have a higher chance at being hit by photons than smaller grains. This is also what gave higher ISO film that "grainy" look. Back then, the equivalent to what digital ISO does would be to develop your negatives and then overexpose or underexpose when producing your positives. If we want to recreate film ISO in the digital world, the closest would be what is called "binning", where the camera groups adjacent photosites to be treated as a single pixel by adding the charge of all those before going to the ADC. As the ADC gets the added charge of 2 (1x2 binning) or 4 (2x2 binning), and lumps that charge into a single pixel, the pixels (not the photosites) is really more sensitive to light, because ot has the light from all those photosites. This will cost you resolution (and aspect ratio, but that is another topic), which would parallel the grainy look of higher ISO film.
I was saying we need to standardize the lens (use the same lens adapted for each test camera) and test for shutter speed consistency. Good call on the sensor glass. Someone should do a scientific test.
I agree with you. Also Fstoppers used for their test an old APC sensor camera in the form of a Nikon 1DH and were comparing it to the same ISO setting performance in 2 modern day Full frame sensor cameras. This in part would account for a difference. I do not think people always realise or appreciate that the size of the sensor is another crop factor that one has to consider. Other than Aperture and focal length multiplication. So an APC ISO equivalence of an ISO 400 setting on a full frame sensor camera is more akin to an ISO of approximately 156 (Canon) 177 (Nikon) APC sensor camera. Tony Northrup gives the formulae: ISO setting of FF divided by the crop factor (squared).
You guys are splitting hairs. ISO is a marketing scheme. I am a machinist at several ISO companies and some of those companies manufacture poor quality products. The rating standard means nothing. As far as photography is concerned ISO really doesn’t matter. Just shoot away and whatever looks good to the photographer and or customer will suffice. Sometimes good enough is good enough. Perfection is only found in nature.
Cornelius Fischer states: "Haven't read the article, but it is right that high ISO does not mean, a camera sensor becomes more sensitiv for light. This could only be done by applying a higher voltage to the CMOS sensor. Without active cooling (as for example done on CMOS/CCD sensors on fluorescence microscopes) the sensor would just burn on a higher voltage. That's the reason why regular camera CMOS sensors have a native ISO. This is the baseline sensitivity based on the applied voltage. So how is a higher "sensitivity" achieved. Simply by applying a signal gain factor. This is applied on the CMOS output signal voltage with an analog circuit and prior the ADC unit. Sensor and ADC have their specifiy/fixed voltages, resulting in the native dynamic range. When I now using the gain (analog amplification of my sensor signal --> higher ISO), I still have a maximum limit for digitation on my ADC. Therefore, increasing ISO/gain results in a drop of dynamic range and at the same time an increase of signal noise because of a reduction of signal-noise-ratio. The better all components work together and the better the signal-to-noise ratio of the baseline signal detection is, the less noise you will see on higher ISO values. So, higher ISO will not mean more voltage/sensitivity on a sensor, but you will get a analog signal amplification. This always results in a different data, then by just pushing RAW file. PS: I have worked as lab technician for about 15 years in the field of flow cytometry/microscopy and met some cool guys from Olympus/Sony/BD/Leica during that time. These guys are awesome with their knowledge about the technical things behind sensor systems. 😳" "Jamie Moffat - technically not really the same. Overclocking really changes the clock frequency of the CPU. An analog signal amplification increases the voltage of an analog signal. E.G. using a gain factor of two, input voltage is 1V, output will be 2V. The signal still has the same frequency. I have worked on flow cytometer microscope which was using two CCD cameras which had a variable readout frequency. The system could measure the speed of a particle in a liquid flow and could adjust the line based CCD readout. At the end a particle passed the CCD chip while the signal get passed line by line downwards with the same speed as the particle. That results in a true signal amplification right on the chip. This is actually the same technique as we photographers use, then we do a longer exposure on a fast moving object while moving the camera with the object. The machine is just adjusting the line based readout to the speed ob the moving object.😎"
that's great are you making compelling content???? That is the main and will always be the main question otherwise we have a perfectly exposed perfectly grainless character list the dull picture.....
@@nelsonmoreno3680 I admit that my comment is sloppy, in that it does not explain detail. Further I forgot to mention that it is as well a factor for the highlights and to gain the full benefits might require a controlled overexposure in this area. The effect is, that you have more color information available in the range you want to expose, than if you had when you exposed in a lower iso range (by practical means having a higher dynamic range). Further down you find a video explaining what I mean. You can see there how clipping is happening, resulting in an image that looks completely dull, lacking color information. Tony Northrup on the other hand is talking about noise, what creates a somewhat grainy picture in result. It can be fixed at the price of a more blurry picture. Dave is arguing that by technical means, before converting the image into digital, you can reduce this noise and he is correct with that. And yes, you are right in what you mean usually the noise in the darker areas is stronger, since: there is less light information, it is more noticable to amplifying the blacks and with that the noise it has. But that is not what I mean. Check out the video to understand the details: ruclips.net/video/2sshGdMgJxQ/видео.html
Tony’s right. 1) On most digital cameras, ISO is just digital gain. 2) And one reason the pictures lifted in post are noisier is because the camera is applying some processing even to RAW files. 3) He’s also right about ISOs being rather arbitrary.
From what I've read most Canon cameras are not ISO invariant. I did my own test on my 6Dii and there was way more noise on a lower ISO with the same shutter and aperture when conditions were dark. Maybe his statement is true for some, many, or maybe even most cameras. However, it seems like it isn't true for most Canon models and that isn't exactly a no name manufacturer.
Except arguably ISO isn't digital gain because it's the analogue signal that is being increased. ISO in cameras is like audio recorders, it's the preamps that boost the audio before it's saved, not just turning up the digital volume later
Even if the signal boost is a digital one - it performs differently than the lightroom/photshop/whatever algorithm performs like, thus giving different results. The in camera boost is "tuned" to the specific sensor and processor.
@@DaveMcKeegan So, ISO values are equal to the amount of recording level on recorders? Or is it equivalent to the "High Gain" setting most (hand) wave recorders have. I have the Roland R-05 and I use always the Hi gain setting instead of turning it off and boost the sound levels in post during editting for example. Is it also better to record in the lowest recording level setting (in normal) and boost in in post anyway in order to get the cleanest sound since High gain settings is only a boosted analog signal (real world sounds hitting the microphone and gets amplified before getting it transformed to a digital signal)?
To add a little, ISO speed is not rated for raw at all. Here is from current standard, ISO 12232:2019: "ISO speed and ISO speed latitude values shall not be reported for raw images, however, because with raw images processing that affects the values has not been performed".
Tony's video itself was the reaction to somebody else's video published one or two days before. I watched that also, but don't remember who it was. It was about higher DR with higher ISO at highlights and that ISO actually only is a parameter metadata at raw photos.
When I pick up a camera I am mostly concerned about getting the shot. I need it for myself (to prove that I know what I'm doing). My client needs it(I'm under pressure). That gain(iso) is just another tool in the tool bag. I just need it to do what it is universally understood to do). I am so thankful to live in a world where I have many choices when it comes to photography. (Camera brands, third party lens, third party products, RUclips channels, and competition that makes this industry sizzle). I appreciate everyone's feedback. It is an incredible journey to learn and discover new things. Anyone who decides to make a RUclips channel based on photo/video is very much welcomed.
To clarify what Dave said, on a CMOS sensor, which is now use on pretty much every camera, before the AD converter ( which is an independent componant of the sensor ), there's actually a charge converter included on each individual photosite of the sensor, which produce a voltage depending of the charge of the photosite, which can be then send to the AD converter. It's a little confusing, but without the charge converter for each photosite, it's not CMOS, it's a CCD sensor. ( IT CCD utilize lines of buffer photosite between normal one, and just serve as emptying active photosite, and transfering those charges to the charge converter, and then the AD converter. there is olso the FT version which use a second "sensor" that just serve as buffer, so that the active sensor have a better fill factor. )
exactly, here is some old site about that from 2002 (the days CCD was the king of the picture and CMOS an unwanted baby), and light and color. very nice read. micro.magnet.fsu.edu/primer/digitalimaging/index.html
The comment toward the end about shutter speed is the first thing that came to my mind. If you're putting your camera on something like aperture priority and you're taking a low-light photo without a tripod, increasing the ISO is very useful since that allows for a faster shutter. So, not only are you getting a better photo than using a low ISO+post production, you're also getting a better photo since you're not getting motion blur from unsteady hands.
Capturing images work a little bit more complicated. But I agree with Dave in most of the things he said. People tend to forget (or maybe do not understand the basics of physics) how devices they use work. To conclude: there are many factors which influence the final results (a photo-cell size, it's capability of producing the current from some amount of photons, etc.). However, the ISO is adopted to our needs because we can somehow understand where we are when capturing photos.
I think you got caught up in the details... Though Tony did use those examples I believe he also expressed the difference was minimal. Not in those exact words... The gist was ISO was a standard in film and a ISO100 from mfg to mfg would result in very similar exposure settings without regard to camera mfg. However, digital world there is no standard. (Which completely defaults the use of ISO... Which means "equal") So though you are correct... I also believe Tony is as well. ISO in the digital camera is a completely false statement as each mfg, and even different cameras from the same mfg will require different camera settings in the same lighting condition.
@@chrisholmesphotography2456 I disagree. I will concede ISO logo is used for the International Organazation of Standardization. Which initials would be IOS... ISO = Equal.
@@chrisholmesphotography2456 still no... yes you look it up it will lead you to exactly what you say... except the acronym for International Organizations of Standardization would be IOS... therefore.. ISO still = equal. Quote "According to ISO, ISO is not an abbreviation. It is a word, derived from the Greek isos, meaning "equal," which is the root for the prefix iso- that occurs in a host of terms, such as isometric (of equal measure or dimensions) and isonomy (equality of laws, or of people before the law). The name ISO is used around the world to denote the organization, thus avoiding the assortment of abbreviations that would result from the translation of "International Organization for Standardization" into the different national languages of members. Whatever the country, the short form of the organization's name is always ISO." Link -> searchdatacenter.techtarget.com/definition/ISO
Sometimes its not always about absolute right or wrong, but I appreciate a good counterpoint every now and then. That's the type of thing that keeps you guys on your toes, and honest. BTW, I tired fixing it post with my MK III and it didn't work. Not even close.
Tony's point about ISO 100 is that it isn't set to a standard level. ISO 100 doesn't give consistent results across manufacturers or even cameras. You've tackled him on a different point.
@@DaveMcKeegan I wonder how big the difference in ISO was with film compared with the differences in ISO within digital camera's today. If the difference in film is a lot smaller then digital could have been a standard within tolerances. Wondering when is something consistent. What makes the difference between consistent en not being consistent. Where to draw the line. In my opinion film was consistent enough, within manufacturing tolerance, to be able to say that there was a standard. ISO was the parameter that told you how long to develop the film. It must have been close enough to give good consistent results. Digital might be off a lot more than film ever was.
@@primefotoNL Yes, back in the day, film had to meet a certain standard within allowable tolerances so diferent brands of film were very close. Digital camera have a wider variance in that standard so ISO numbers aren't as closely accurate between brands as film was. I was shooting film from the 70's till the 2000's.
agree. ISO100 in different brands have different approach to the ISO100. It is not standardized anymore. tsk tsk tsk! i could not even finish this guy's vid.
@@DaveMcKeegan Not just from brand to brand, but also from developer to developer - the same film can have significantly different actual ISO when developed in, say, D76 and Rodinal
After ADC signal is already digital. Passing around digital signal does not introduce any additional noise. Only possible explanation why ISO can't be applied in post, is either that gain is applied with analog amplifier before ADC. Or if gain is applied after ADC digitally, but ADC resolution is greater than amount of bits used per pixel in RAW format. Most probably it's first one.
I think the point you make, Dave, that raising the ISO means that the camera boosts the analog signal before conversion to digital, is central to your argument. It totally makes sense, and would account for the increased noise (and also notice color shift as well) if one merely shoots at native ISO, then raising values during editing. Digitizing an analog signal always results in data loss: a curve is converted to a series of discrete steps, and that loss results in noise. Increasing exposure of a native ISO image during editing magnifies that noise; by boosting the signal (shooting higher ISO) before digitizing, you are retaining as much information as possible before the image is digitized: you are magnifying the analog curve, not the discrete steps. So the bits of noise are smaller and fewer. Fstoppers even remarked that they noticed more noise in the greatly underexposed native ISO image, but then blithely went on to ignore this observation, saying "Tony is right." I'm not a Tony fanboy, but like T&C, and Fstoppers too, but in this case, I think they're wrong, and I believe you explained why. Thanks, and good job.
Boosting the analogue signal before quantisation is indeed the central point. Dave has mentioned, though not stressed that quantisation is not a lossless process, it actually removes data from the analogue signal to be able to represent it in a digital form, so all the amplification is better to be done before that happens. Dave said "the interference" which is, in my eyes too soft of a term, because parts of data actually do get thrown away, not just interfered with. Also, and it has never been mentioned before, amplification of the analogue signal may, and probably does involve multiple stages like filtering, compression etc. Actually, the analogue part is exactly the place where the miracle of developing new sensors with better ISO range happens every year. Only that we, as a broad public are not aware of that fact. Well, just to summarize of what I've just said - Tony had represented a pixel as an empty bucket for collecting photons, connected to the input of a ADC, while in the reality it is, of course, much more complex, hence the mistake.
Apart from quantization noise (which isn't that big of a deal when you do the math) the biggest source of noise is the inaccuracy of the ADC itself, the so called "read noise". The Z6 perform great at high ISO because it has an extremely low read noise and good quantum efficiency being a BSI sensor, however at low ISO if you try to boost in post you will notice the different read noise between the normal pixels and those used for the AF, which results in the banding you can see when trying to open shadows 5-6 stops. At high ISO the difference in read noise is unnoticeable so the banding is completely swamped by other sources of noise and becomes completely irrelevant.
Yeah, it's like with digital audio. Before digitisation you want to get as much analog gain as possible without any clipping. If you recorded everything with 36 dB of headroom and then boosted it in post, you'd be wasting a whole lot of bit depth.
The point is, that modern ISO invariant sensors do NOT boost the analog signal anymore. The analog/digtial converter is right behind the sensel wich brings other advantages too. This is the major flaw in this video. Analog amplification was performed on "older" sensors. This is also one reason why not every camera is ISO invariant.
Based on the description of the circuitry, an Analog signal (voltages) from the sensor passes through an Analog amplifier before being converted by an ADC to a Digital facsimile (numbers) of the original signal. The amplifier boosts the Analog signal and the thermal noise of the sensor plus introduces it's own thermal noise. A voltage has an infinite range of possible values but an ADC must select a discrete number as a best fit to represent it, imposing quantization distortion, another source of noise. An ADC also has a limited number of numerical representations available to it (dynamic range) so converting a low range (under exposed) signal will vastly under utilizes the limited range available to the ADC. Dave rightly states that amplifying the Analog signal as soon as possible minimizes the potential to pick up additional line noise and as the amplifier is in circuit at all times, using it or not using it introduces the same amount of additional noise. Post processing simply scales (multiplies) the numbers of the Digital facsimile and potentially adds it's own noise (rounding errors). Processing a low range digital signal (all small numbers relative to the available range) into a full range one will involve a lot of rounding error. To my mind, the less post processing you have to do to a wide dynamic range source, the less noise you will introduce in the system as a whole. Early amplification to get the best use of the available dynamic range of the ADC followed by minimal post processing to avoid injecting and enhancing noise just sounds sensible. Think of ISO as conditioning the Analog signal to get the best performance from the ADC and minimize the potential of adding to or enhancing the unavoidable noise inherent at every stage. PS: I'm a telecoms guy.
Finally, thanks for clearing this all up! I've read up on it all after Tony posted it the way he did, as it sounded really fishy. Your explanation now was definitely the easeiest to follow.
Easy to follow but his main point is wrong. ISO invariant cameras do not boost an analog signal. It gets converted to digital pretty much right after light hits the sensel and the rest of the processing (including ISO) is on the digital part, so it can be altered later.
Great job describing how ISO works, as an engineer "sensor sensitivity" felt wrong but that's how everybody describes ISO and if it wasn't for Tony's boldness & ignorance many people would still be clueless. Should we flame Tony for not knowing what a DAC is and how signals work in electronics? Hell no, I pretty much like his content and we are all evolving here. It's actually very nice of youtubers to step up and put each others concepts in the test, that's how science works; you publish an idea and expect others to (dis)prove it.
@@henri.witteveen the point is none cares except the engineers and frankly I wouldn't build my own camera with so many well engineered options on the market. Just enjoy photography!
One of the issues I have with Tony is that he's not always consistent about what he does and doesn't test. Some of the stuff, is right despite making no sense. My Canon 70-200mm f2.8L IS was never sharp on my 10D or 7D, but on my 5D mk IV, it's significantly sharper than it was on either of those bodies, even if I crop the resulting image. Without Tony, I wouldn't have consider that to be possible. But, in this case, the take home should be to apply moderate increases to the gain ahead of capturing it, knowing that depending upon the RAW format, you can probably get at least a stop or two of relatively clean signal from that point. Not to mention apply noise reduction with a much more powerful device and with smarter masking.
In your experiment, did you have ISO noise reduction On or Off? Most digital cameras employ digital noise reduction to the higher ISO images, and some do it whether the photographer wants it or not. You can't compare the noise and grain of images without stating what camera sensor and noise reduction setting was used. It could be that the images are noisier simpy because in-camera noise reduction was used in the higher ISO images but not in the lower to native ISO images. All of which can be mostly duplicated in post.
@@DaveMcKeegan Tony is correct when he says ISO is "fake" as there is no standard to compare the value to, and not even the same manufacturers use the same definition of ISO between different cameras. Roger Clark gives base ISO as the "relative level of light compared to the maximum signal a pixel is capable of recording" which is dependent of the size/area of the pixel used. Tony is also correct when he states that a higher ISO is the digital amplification of the sensor readout. Again, R. Clark, "ISO is simply a post sensor gain and digitization range." and that at a certain level you can simply take a picture and make adjustments in post (as long as you have a good algorithm for managing "gain"). Tony is simplifying the math to make his point. You are (correctly) getting a bit more into the details. All digital cameras have a range where they are ISO invariant. This is the range where images become sensor read noise limited. There is no IQ/dynamic range improvment from using a higher ISO value. For the Canon 5D Mark II it is around ISO 1600. The lower ISO values are limited by electronics downstream from the sensor. This is the area being debated. I believe what you are noticing is the differences in the algorithms used between the camera manufacturer and Adobe to increase gain as digital gain is digital gain, ie.e, a simple multiplication As R. Clark states in his article on "What is ISO on a digital camera? When is a camera ISOless? ISO Myths and Digital Cameras", ...Future cameras will have 16-bit A/D converters thus reducing the need for ISO gain. Actually 18-bit A/D converters are needed for the top cameras, but we are not likely to see that anytime soon. With 16-bit A/Ds, we'll only need broad categories of ISO, like low, medium and high, and choose precise gain (ISO) with raw conversion. "As camera electronics become better, we will no longer need an ISO setting on the camera, as we can effectively choose it later." So, who is wrong on ISO? Neither. If Adobe had the same algorithms that Sony or Canon uses, would you see the same image using the same true exposure? Most likely. see: www.clarkvision.com/articles/iso/
you both have a point (tony and you). but i dont think there could be a much difference between boosting the singnal before and after a.d. i think the cameras put in noise reduction even if you disable this option
The difference is, as you pass a signal round you introduce more noise, if you amplify the signal later then you amplify the noise as well. As a basic example if you get 4db of light signal (not the correct units) and you're going to boost it by 2 stops up to 16db, if the circuit introduces 1db of noise in the process then boosting the signal straight away will give you 16:1 signal to noise, however if you boost it afterwards you end up also boosting the noise by 2 stops so you'll end up with only a 4:1 ratio
So much depends on the bit depth of the AD converter, the quality and profile of the signal amplifiers, the correction of rounding errors, the noise cancelling algorithms, etc. Technically there should be no noise introduced after the AD conversion as the values are all fixed into memory. If noise could be introduced after the AD conversion then computers wouldn't work as no values would stay the same. Most of the difference comes in the bit depth of the raw file. For a 15 bit raw file, a correctly exposed image will have values from 0 to 32767 in steps of 1. A dark image taken at low iso may only have values from 0 to 1023 in steps of 1. When the dark image is pushed 5 stops in post the values will go to 0 to 32736 but in steps of 32 making any gradations in tone (and noise) very noticeable. Try this with JPEGS and the result is even more noticeable. So Yes - Always get your ISO correct in camera. - I think!
Base gain, (transformed to base iso) is the gain for witch you match the pixel saturation (the full well) and the amplification saturation. For instance if your full well is 60ke- the corresponding digital number will be about 2^14 = 16384 for a 14bits camera.
Interesting...Your vid made me think of something: DPReview did an article recently about the Nikon Z6 banding issue, and to demonstrate they pushed a 100 ISO file up +6 EV in post. The banding is definitely present at ISO 100 pushed EV +6. However, if you look at their scene tool, and push the camera up +6 EV to 6400 ISO (which is an IN CAMERA setting), you'll notice that not only does the banding vanish, but additionally the noise levels are a bit better too. This seems to show that no, just using software to push your EV up is not the same as adjusting it in camera. The DP Review article is entitled "Nikon Z6 image quality and dynamic range impress, but not without caveats" for those that want to take a look.
What you’re saying makes more sense to me based on photo processing experience. I tried taking pictures at ISO 100 in low light and boosting exposure in post processing. Results were not good. Signal to noise ratio is the reason a recording artist uses a preamp to boost a quiet audio signal before analogue to digital conversion.
I agree on your your take on ISO invariance. But I don‘t on your point on ISO100. ISO 100 should be same for every camera and I am sure the manufacturers try. The base ISO in my world is what the sensor captures when only A/D conversion without boosting or reducing. Some cameras have base of ISO100, some ISO200 or even ISO64. Of course these values are not exact, but not wrong by a full stop. An additional variant of. ourse is the light transmission of the lens. This is what t-stops are for as apposed to f-stops. But practically 99% of images are shot within camera light metering through the lens, so nobody cares about the actual numbers of t-stop or ISO (in the sense of a comparable, standardized number). Also a partial stop is irrelevant when using post processing.
signal boosting level never changes by changing iso except for the cameras which have dual native iso, about all of photographic cameras have only one signal boosting level which we call it native iso ,other iso levels are applying digitally
A always like viewing Tony Northrup's video's. I don't have to agree with everything he says but I do learn when viewing because his videos make you think. He also brings out "thinking" among rival video bloggers too, judging by the recent rebuttals his videos have taken. Obviously this ISO discussion has led to multiple rebuttals yet there is nothing wrong with that. Discussion is a learning tool for many. Another topic video Tony put out recently (?) was a discussion of JPG versus Raw. That video drew the ire from at least one video blogger in froknowsphoto who took exception to the video. Again, no harm comes from opposing points of view. I don't see the point of getting so emotional in opposing points of view. YMMV
I liked your passion about this topic which is steeped in technicalities. My criticism of various RUclipss is advertising. Are you sponsored? That is my first question. The real issue with ISO (vs ASA) is S/N ratio. That is the heart of the issue. Increasing signal over noise. That is the electronic way of viewing this. I came to the conclusion that it is not noise that we are trying to suppress but higher ISO has less range. Lower ISO increases range or how f-stops we can capture. But I do like the push back here. It is good for the community.
You should be aware you're making a statement logically wrong You're comparing your Sony a7III results with F-stoppers \ Tonys Nikon D850 results None of you is able to make statements about the majority of digital cameras of different manufactors Note:. - Most sensors nowerdays are produced by Sony (Fuji X-Trans, even latest Nikons) - The digital images we potographers actually recieve (let it be RAW or JPEG) are heavily processed Sensor signals (selective denoising (referring to "edge noise"), vignetting correction depending on the lens mounted, and most obviously: converting the Bayer/X-Trans Filter Colourpatterns to sRGB \ Adobe RGB Colourroom and sampling in missing information\holes created by phase-detect pixels. All of this is done well before the RAW image is processed in the camera, is NOT affected by any settings in the camera menu (Sidenote: genious tec: Sony a7II colourchannel correction for manually compensating for CAs of attached vintage lenses (don't know where in the processing chain this correction is made as it's a well kept secret and removed in later cameras)), AND note: ALL processing (-software) between photocell charge (from sensor readout to the RAW file) is a manufactors biggest secret and a result of insane development. Talking about ISO: When using 35mm film, which is a physical part beeing interchangeable between film-cameras standardisation of the films sensitivity was a must have to propperly expose an image. ISO could be determined by actual testing and was a physical property of film (reaction time of the chemical ingredients) Nowerdays ISO is one of many imput parameters used to influence the image-processing and is therefore depending on software/programming of an individual body and even the lens used (f. ex.: Fujifilm camera and lens firmware versions) As a matter of fact: by shooting the same image (with the same settings) trough the SAME LENS, wich is UNABLE to communicate with the camera body you could actually make out differences in ISO between camera manufactors, and even bodys of the SAME manufactor. Mainly ISO is a matter of what a camera manufactor is LETTING you use and the manufactor is deciding which compromises can be made reguarding image-quality, low-light (high-ISO) -capability, clean-image (low-ISO) -capability Examples: - Nikon D850 extended Low ISO of 32 (producing incredibly clean landscape Images) - Sony a7III extended High ISO of up to 102400 (great low light images reguarding noise) Low- and High-ISO capabilitys are quiet different topics: High ISO means plenty of work to do in denoising (typical grain, also colour-noise, without creating processing artefacts) while maintaining good colour-resolution, hitting correct Whitebalance with auto WB, while accurate and quick focusing with mirrorless cameras is still possible, etc. To summ it up: mostly a Software issue (AF with contrast AF included; AF with the use of phase detect on Sensor is also depending on the engineering of the sensor), all affected by processing power Low ISO really comes down to physics and engineering: how much charge a photocell of the sensor can handle while maintaining valid information, inter-photocell isolating, readout process, etc. CONCLUSION: on modern Digital cameras ISO is mainly a parameter you dial in to influence processing the Digital Image, Each camera manufactor is deciding which values for ISO he's letting you use (which is also heavily influenced by marketing), ISO is mostly a software issue, and sceems to (with digital) no longer describe a "standard", but a setting not consistent between camera brands or even different bodys of the same brand (compare Fujifilm X-T3 ISO 200 to that of an Sony a7III with same shutterspeed and lens transmission "t-stop") I hope that did / does help you understand the fascinating world of photography a tiny bit better. But reguardless of your camera / lens/ or even ISO setting: Go outside and take photographs of whatever you like, AND: have fun !
That noise level is going to be different depending on what you used to process your RAW image (your camera or lightroom for example). Pixel count, signal to noise ratio and dynamic range are only real measurable values of the sensor. Iso is strictly software and image processor related feature. And yes you can get same or better results if you use native ISO and then crank it up in your computer but that would not be so much practical to do. In the film industry they like to find that native ISO that gives them best image that sensor can do and never even touch that thing again. And if you have to push or pull brightnes you do it in software that comes with camera by changing that ISO value. Even on real film stock things works more or less the same. You push or pull your stock in dark room to get brighter or darker image that also introduce more visible noice.
But, if iso really is arbitrary I should be over to overexpose in camera and bring it back. In my 80D after 2 stops adjustment the picture is useless. Taking an image that needs brightened works well. I agree about the noise and what device processes, I've been saying the same thing. If it didn't matter I should be able to recover the picture in both directions.
@@fixitrod4969 Thats because Canons sensors depends very much on image processor and their RAW files are not exactly the best ones. Go to dpreview and download image from Nikon Sony or Panasonic and push it few stops and then do same for Canon. Canon lucks dynamic range on their sensors and and highly rely on their processors and in camera software to fix things up even in raw. Thats why there is not so much use of Canon raws in terms of pushing or puling exposure.
no its not ... he made it very clear just you havent got it. Than you can say ISO is fake whatsoever even in film. because also film can be darker or brighter even the same ASA ISO or DIN (which ever measurement you use) because its dependable on developing as well as digital cameras on boosting signal. Its the same premise.
...actually, ASA was written on the Kodak box but it was exactly equivalent to ISO. ISO was on Ilford boxes. In the dark days of film we spoke in ASA or ISO interchangeably according to our preference.
@Puppet Master - If people set their camera to ISO 100 or ASA 100, (That is, if an ASA setting actually existed on digital cameras, using older analogue terms here), rather than employing older analogue terms, what difference does it really make? I for one consider ASA and ISO equivalent meanings for photographers, even if they're not so equivalent for more scientifically minded people.
@Puppet Master They weren't exactly the same to start with, but eventually the two standards organizations merged their two standards in various revisions.
Reading into this too much.. Tony was spot on.. When he said u could just adjust exposure in lightroom he didnt mean it as an exact science but its bloody close enough
Except he didn't say "they look close enough" , he said they looked "identical" which is incorrect, and that was on a more modern camera, on older cameras that difference would be even greater and could lead people into believing there is no difference when there actually is
As far as I know the noise amount with may camera stays almost the same but I get annoying color shifts if I underexpose a photo 1.5~2 stops or more where the blacks have too much red. I've also noticed that my camera (Sony A68) in general makes good photos except at ISO 800. That's like the sweet spot of the worst look I'd say.
I believe he was generally correct, the iso setting is fake because its based off film standard and this is digital. Like you have said the camera companies aren't held to a strict standard just an estimation so 100 might be 85 might be 105 or whatever, and that was the point. The 4 stop difference to just shoot at 100 will get you close, but you are correct it would be better to shoot at the correct exposure for the best result. I think he just mad that iso 100 should be iso 100 no matter what camera you are using and pointed this out by showing the results of different cameras. This is not a new problem, I have worked in the dark room before and NO film was exactly correct, they were close generally with the better brand films but they were Never 100% correct.
Precision is not the issue. Tony said you can fix it later in post and that's simply not true. Granted there is some amount of latitude but its never as good as native iso.
In the film days (some of us still shoot film) we would buy our film in batches and test the first roll for iso. The same can be done for raw files which are not subject to any iso standards. No need to test jpgs though as they are already controlled by the iso standard
Well done. In the olden days ISO was either ASA or Din and was supposed to represent a standard measure of the sensitivity of film and offer a guide line for exposure. Film is analagous to the sensor in today's digital cameras. In film days one changed the film sensor depending on expected light conditions. That's why I loaded some cartridges with 12 or fewer frames. In today's digital I take ISO as a point of departure not arrival. Despite what the camera is able to capture with all that intervening circuitry, I still have the challenge of what software and monitor will let me render and see; every software has it's own language. I think recently Northrup back peddled his opinions about ISO.
Absolutely! At the time of my reply Tony has over 232K views of his video while this guy only has nearly 73K views.... Views = $ …. This guy is getting Tony paid! Tony is laughing, alright- all the way to the bank!
Tony is NOT baiting anybody! There is a TON of real misunderstanding about what GAIN actually is. Actually, the ISO term was fine in film but it doesnt belong in analog/digital sensors! In fact, the ISO concept is screwing photographers up today. Gain/ISO has nothing to do with exposure of light to a photosite. Only shutter and apertute controll "exposure".....NOT gain/ISO. The faster people understand this, the better we all are.
I thought it sounded a bit off (doesn't help that he can come over as a bit smug). What I don't get is is there really that much interference in camera? Good to see your tests though to prove your point.
If they all have different ISO100, right, how can that work with the doubling and halving ISO being equivalent to opening or closing the aperture 1 stop, or doubling or halving the exposure time ?
Are these photos you tested RAW or JPEG ? It's correct to shoot RAW when comparing different ISO's since many camera process or interfere with the jpeg photo and it might be the reason why you're getting different noise or colors. How do we know for sure that a certain camera doesn't process and reduce the image noise over a certain ISO level?
At 2.30 you state that Tony is totally wrong. Others have tried this and confirmed that you can get virtually the same result post camera as in camera. In camera ISO = equals gain, nothing more. Doing that at different stages, I agree, introduce minor changes in noise. Tony was not advocating ISO 100, he only advocated leaving ISO on auto and worry about other aspect of controlling the photographic image that have a significant impact on image quality.
- If an A/D has infinite precision, there is _no quality loss_ using the base ISO and correcting in post. - If an A/D is very good, there is _no practical loss in quality._ - If an A/D is poor, there is a _slight loss in quality._ The _only_ reason adjusting in post causes quality loss is due to rounding errors. I think Tony should have been clear that there is a difference. But the point is you shouldn't be too concerned with getting the ISO perfect when using modern cameras. It's not easy to get everything perfectly correct in a video, and easy to criticize, as you probably noticed with your mistake regarding F-stoppers.
Karl A Every post processing software is different. On top, being completely off in ISO during shooting will deviate you from correct exposure... without signal (too short exposure) you will always end up with a crappy image, no matter if you‘re using high ISO in camera or pulling up the exposure in post.
I just have one question.... What did he say at the end? that is his conclusion? I have just made an exercise ISO 100, 400, 6400, and really there is not much difference that post cannot solve. I am more inclined to the shutter speed and Transmission of the lenses, than in the ISO. If shooting RAW there is not that much of a difference, If shooting JPEG because you are in a wedding and have to deliver immediately, then yes, use a correct balance between ISO, Sutter, and aperture.
I don’t know your channel and technical background on this topic, RUclips just recommended this video to me, but I would like to point out, that you indeed got something major wrong here. It is at 4:10. You claim, that the gain/amplification of the sensel voltages is performed before it gets through the ADC and that is NOT the case with current ISO invariant sensors. It used to be a dedicated ADC, seperate from the sensor for that. On these modern sensors, the sensel voltages go through an ADC directly on the sensor more or less directly after each sensel and that is the big thing to consider. This "short way" reduces the base noise/increases the SNR and shifts the ISO adjustment to the digital section. Most of these cameras still do have a dual native ISO or “dual-gain architecture”, so images with ISO 100 actually do have less noise and more dynamic range than at 1600. The two base ISO ranges of an A7 III are 100-500 and then 640+. You can’t change these values but the “rest” of the ISO is just a multiplication factor baked in the RAW file and this can be changed through RAW converters like Lightroom. But I’m totally with you in saying that shooting ISO 12800 and 1600 + exposure boost of the RAW file are not 100% the same. You actually do lose a little bit if dynamic range and get very slightly more noise since no Sensor is 100% invariant but the results of a modern sensor under these circumstances actually are VERY close. For my background on this: I work for one of the largest camera manufacturers (industrial, not consumer) and have a direct insight on current and upcoming sensor technology.
Thank you for the insights I appreciate not all sensors work in that way, however Dual ISO architecture is costly to make and not all sensors are ISO invariant, so there are still cameras on the market that do work in their manner. The point I was trying to highlight is that unless you have a true ISO invariant camera, then you will have difference results from increasing exposure in camera to in post. I am going to do another video that delves more in depth into the difference ways ISO is done
@@DaveMcKeegan Yes, im totally with you there. Most cameras do not use this swapped ADC/ISO adjustment. It is mainly Sony who implements this in their more modern cameras but since Sony also manufactures sensors for other companys, they are not the only ones on the market. I think the main point Tony was trying to make is, that in the "real world" it doesnt really matter if you correct your image brightness +3 stops in post (if you have the right camera that is). With Canon cameras for example you won't have such a good time trying this extensively.
Higher ISO as well means you need to down sample your resolution. Perform the "pixel binning". That is what a film did, faster film has larger silver halides for better light capture but so on lower detailing capability. With electronic sensor (it is not digital until converted) people do not want to give up with their "valuable pixels" that they pay for. In film era everyone learned that faster film meant you get less definition. In digital era most so not ever learn that they should lower the resolution and apply different demosaic method. In film era everyone knew that slower film have larger enlargement capability or possibility crop tighter. In digital era most believe they can print as large as ever regardless the ISO setting. Now try this. Take 24mm lens and use ISO 200 to photograph a scene. Then take 50mm lens and set ISO 1600 and capture same scene with four frames and stich them together. Down sample it to same resolution as sensor has with single frame. Again, you exchange speed to definition. Same thing can be done with single frame, you capture full frame with ISO 200 and then ISO 3200. Now you get to files, full resolution and 1/4-1/6th of the original size (in area). So 50-33% scale from original. Meaning, just like in film era you didn't use 135 film if you wanted 100x70 cm enlargement with ISO 64, but you used 120 film with ISO 64. Same speed, same sensitivity, same light gathering power, but just larger frame, that you exchanged to a definition in larger enlargements. When you didn't need more than A3 size prints, you didn't use medium format. When you used faster film, you knowledged that you do drop the possible enlargement size each stop faster shutter speed you want. With digital, that ain't automatic, as there ain't physical silver halides but a physical static pixels in grid, regardless what sensitivity (speed) you use. It is photographer tasks to knowledge that how the print size changes based ISO speed. And only image resources tests this! The give you their visual testing results from the print sizes based the ISO changes. They literally tell that example with ISO 200 you can make great 35" wide print, but at ISO 1600 a 16x11" and at ISO 12800 only anymore a 8x10". Because they understand that sensitivity and it's trade-off. That is as well why it is wrong to talk about "crop sensors" when people do not crop but they frame everything same. No, of you want to talk about crop sensor, then you perform every comparisons with same focal length, same F-stop, same shutter speed and same ISO, and then you keep same perspective to subject and take the photos and then crop the larger format area to same as smaller sensor and that way compare the crop qualities. What larger sensors offer is larger field of view with everything else same (except smaller sensors usually have better sensor technology and results a better quantum efficiency and so on better image quality at pixel level, that you elongate by enlarging files X times more with smaller sensors than larger sensors. Because people think wrongly that quality difference stays same regardless the output size (don't get it that larger ISO = smaller print with same quality). Because they do not understand the context why photography is made! 1) How many here does constantly prints larger than 100 x 75 cm? 2) how many does max 75 x 50 cm? 3) How about family albums like A6 and A5 size? If you belong to first group, then get a FF If you need speed (ISO 8000-51200) or get MF if you need quality from 20-40 cm viewing distance. Otherwise 4/3" does as good results in any other target audience. Prints smaller than 150 x 75 cm at ISO 200-800. Prints smaller than 80 x 55 cm at ISO 1600-3200. Prints smaller than A3 after ISO 6400. Digital form real world use (4-10" screens with less than 4 Mpix) you don't get difference to go larger formats.
I think T&C wanted to give an elementary answer for the benefit of those beginners out there that need it broken down in an elementary way. I can appreciate that even though I'm not a beginner. Sometimes we forget how difficult is was to grasp certain understandings back when we first started out. Dave Mckeegan, great job and great info too.
That's is exactly the danger of his channel. He is misleading beginners ! You will learn more photographic truth from the Angry Photographer, backing up facts with optical principles. Phony Northrup on the other hand, mixed truths with lies that sounds believable for beginners.
@@melvinch Oh please, the angry photographer is a hack. He talks and talks yet never backs up his word with any photo proof. At least T&C show their work when testing a camera or lens. All the angry photographer does is say trust me I know then does his weird laugh.
@@williamneuman7783 AP has photos on Flickr. They may not be artistic but are technically competent. Phony Northup's photos on the other hand, are amateurish at best. And I'm not the only one who think that way, you can google that. And most importantly, Phony Northrup has a fucking face that says "I'm a blatant Sony shill" written all over.
Well IF that's true I don't think it's intentional. Everyone is right in there own heart so I don't think it would be intentional. I personally haven't seen any issue where he was apsolutelly completelly wrong on an issue eather. With that said i know there's not a 100% of anything and so theres exceptions in every topic. Some people think that rair exceptions are the proof that someone is 180° wrong and that's just not the case even tho there is marriet in the accusation.
you made good points ...thanx for posting only thing clear to me about ISO is there is most camera shooters don't fully understand it. but the bottom line for me would be if they DID fully understand it, would their shots improve ? i have spent a LOT of time studying ISO and my images haven't improved by a significant amount.
I'm not a great specialist, but I'm sure that in the 80-90's we could have had the same debat about different film makers, (Kodak, Ilford,Fuji). The different negatve films rated 100 ISO surely gave different performances in different cameras with different lens. An interesting debat though kind a sort of tempete in a glass a water... I'm glad that I don't think all about that when I take pictures. But I think that the emphasis that Tony Northrup made which is the most important for me, as that the different manufactors of digital cameras today have all adapted their own ISO standards(and that they promote as being the most senstive). 100 ISO from one camera from another and the values of ISO we can reach is really based on the manufactor's own definition of 100 ISO. In another 10 years I'm sure that we will have 100 megapixel sensors shooting at 25600 ISO on the market without 1 grain of noise in the frame, what will 100 ISO mean then? Maybe that already exists...
Anthony Salema they did and changed emulsion to emulsion that’s why we bought a case and tested the film. Found the correct iso and used a Wratten filter to color correct.
The "boosted" image in post production looks more noise because of the way Lightroom handles the data, you would get different results with different softwares but basically is what the camera does just as Tony said. The thing is that every manufacturer applies their own noise reduction to the picture to hide the noise
Check out F-stoppers latest video, they tested the same thing but using several different pieces of software including manufacturers own processing computer software, and none of them could equal the result of the correct ISO shot. The thing with in-camera noise reduction is it works in one of two ways, either a filter like you see in lightroom that softens out the noise (but that softens the images), and true noise reduction which takes a dark frame with the same settings to measure the interference and then remove it from the original file (but that takes time for the camera to process and needs to switched on
Shout out to the Tony and Chelsea fanboys who are going to dislike this video because you called out their favorite RUclipsr for being misleading. I had a feeling that video (Tony’s) was going to be wild by the title. But i couldn’t explain why it like you did. Salute sir
Danwarb - if you're referring to the sample shots around 5:30 then there isn't more light The scene remained consistent, I just increased the ISO (the same as Tony's experiment) I was just showing strips of each file of which the higher ISO shot happened to include the light but the light was present in the 100 shot as well It's why @6:00 I compared the same part of the image for iso100 Vs iso1600 and I went with the sky rather than foreground so ambient light was guaranteed to be consistent
depends on the camera. I never shoot over iso 800 on my a7ii, as it looks utterly horrible. The internal processing is really bad. I got way better results using photoshop to push the image later.
ISO = signal gain in digital cameras. No "ISO" control in video cameras/camcorders, only signal boost/gain. No ISO in digital still cameras either. ISO really is just a holdover label from film, when ISO determined the size of the silver crystal halide particles in the film emulsion which determined proper exposure times. So technically, there is no ISO in any camera that doesn't shoot film. People know what it means. Of course the meaning is hitting that point where the average person learning photography will never shoot film. Technical, technical; does not matter. "ISO" adjustment is applied to the analog signal captured by the image sensor prior to digital conversion. Doing this in software would be the equivalent of applying changes post A/D conversion. Not the same, even if they can produce similar results. Case in point: Black Magic cameras (and other dual native ISO cameras) make use of this to full effect with two separate native ISO base lines. If the "magic" wasn't taking place at analog signal level prior to A/D conversion, there wouldn't be much point to dual native ISO cameras, because it wouldn't produce the results actually seen. Tony is technically correct about ISO, but incorrect with regards to what was stated with regards to signal gain applied to an analog signal being the equivalent of applying exposure correction in software to a converted digital image. No ifs and or buts about it.
@@DaveMcKeegan Yeah that's exactly what went through my mind - using a microphone preamp to adjust gain in your recording or just pushing the fader in your DAW. sounds different and you lose resolution.
@@DaveMcKeegan Well, with an ISO invariant sensor you can to a certain degree. See Dpreviews RAW ISO- Invariance test tool www.dpreview.com/reviews/nikon-d810/13
It's really a shame and misleading to label the sensitivity of the sensor "ISO". I appreciate why "ISO" was used for digital cameras--it was a term very familiar to film users and does relate to aperture and shutter speed in the exposure triangle the way film emulsion sensitivity to light does--but considering that most people taking pictures today have never shot a frame with a film camera in their lives it would be much clearer to call "ISO" in digital cameras something else. If it were instead called "gain" (I'm just throwing out a word) then camera manufacturers could assign values to it without reference to some historic standard value and without the temptation to compare one camera's base value to another's (which apparently is futile anyway). And, if I understand your video correctly, it would also be clearer--"0 gain" would then be the base value without any arithmetic boost between analog signal and digital value. That seems much more understandable to me.
David Jenkins there is no exposure triangle. Exposure is only determined by the aperture and the time. Nothing else. However I do agree with your overall point about ISO being used in digital cameras instead of gain. It is definitely more familiar to those people who came from film.
Just a comment with some math. Normal file (JPG) 8 bit covers 256 shades of grey each color. Each 1 bit increment doubles informationso - double f stop. So 14 bit RAW file covers 6 f-stop depth of sensitivity. Question is if this increment goes both sides (into shadows and into light)? If it goes both sides we should get information 3 f-stop below measured and 3 f-stop above. There is simply no numbers (information) to retrieve more than 6 f-stop from 14 bit RAW file. So calculation has to be done inside camera body.
Having Canon gear I can tell you I will be in big trouble if I underexpose 3+ stops and try to bring the exposure back in Lightroom. Recipe for disaster !! If you have trouble getting the correct exposure in a digital camera, sell your gear !! sheesh.
The most important thing that is not covered in this video is bit depth. If you're shooting on a 14-bit camera and underexposing your shot by 4 stops, then it is the same as you're shooting on a 10-bit one - you're never using the most significant 4 bits in this case. That is why it is important to amplify signal before ADC, not in post processing.
You are absolutely correct. I just tried a little test with my a6300 and there is a very noticeable difference, with more noise in the ISO 100 shot after increasing exposure in software compared to an ISO 3200 shot. Great vid.
Great video ! You got all the right points and explained they in a simple, but yet, technical accurate form. Some people should really understand more about electronics before saying things about photography
my understanding is that when you shot higher iso the camera does some nose reduction without you even asking for it. when you boost the shots in lightroom are you adding more noses reduction for the lower iso ones?
...IDK but, how about we all just forget about how any of this really matters, OR NOT, and continue taking photos with our cameras, processing them how we are and have in the past, posting and printing them and carry on like we are all normal? :) The reality of it is, none of this blaggar matters or is going to change anything in regards to the way we continue to take pictures (with the current technology) or buy cameras, is it? Not for me...
As someone who have actually studied electronics at the university, I am admiring how clearly and concisely you have explained where exactly Tony has made a mistake. Well done, Dave!
I agree. The conclusion here should probably be that in most cases, a modest boost in the ISO is prudent knowing that with RAW formats, you'll have extra bits to work with and much more powerful hardware to engage in noise reduction. A 10bit RAW file has roughly 4x the values possible to work with and a 12 bit has 16x as many, which is roughly 2 to 4 stops worth of data. Underdoing the ISO slightly by a stop or two is probably not unreasonable if it allows you to use the ISO that your camera produces the least amount of noise with. Part of the issue with both of these videos is that the lowest noise on the sensor isn't necessarily at the lowest ISO, it may be a couple settings up from the bottom, in which case, you get a bit of free lunch by just using that if you can and then using that as your base point for further work. Boosting that by a few stops in post production may well give you the best quality you're going to get.
RED doesn't offer the option to change ISO in post, RAW does. Any RAW does. But what you change is... not really something that exists... You're only adding gain... noise. I know this because I also own a camera that can shoot RAW video. I don't "change ISO" for 99% of the time.
Thank you for your clarification! You are right. Basically the understanding of a signal path is not to complex: At the frontend you have a nearly analog signal (not thinking about astro photography): the number of photons hitting the sensor within the color channel and the exposure time. The signal is changed to electricity by the sensor, thereby you can control the ampifying factor by the ISO setting. (the amount of photons you control by exposure time & aperture setting). Being analog you have effects like noise. Next step is the critical one: Analog-digital conversion: this means the analog signal is categorized in steps (i.e. 8 bit equals 256steps, 10 bits 1024 steps, 12bits 4096Steps) What ever is showing up in signal computing you live with this input steps. every noise from the sensor is also pressed in this steps. The only strategy for having the cleanest signal is hitting the white point already on sensor avoiding clipping (the analog signal is above range of A/D conversion) but giving as much steps as possible. The result is the very simple photographic rule: get the exposure right in camera as bright as possible without clipping the highlights. you can do everything in post but you can not get any information of your subject in your file which went not through A/D conversion.
It's more complicated than that if you don't send roughly the same size signal to the ADC (analog to digital converter) you will get less data in grey scale colour scale as well. So the picture will look grainy with lower light.
I would like to apologise to Shaun @ F-stoppers - I said in this video that they agreed with Tony's findings, however I now realise that isn't the case. I misunderstood their interpretation of what Tony was suggesting and believed they were fully agreeing with him, having watched the video again since I realise they are actually more in line with what I am suggesting here.
Sorry F-stoppers
I was going to raise that as the F-Stopper video did report visible differences between the underexposing 5 stops then boosting by 5 stops, even if they didn't know why.
We have a new video coming out next that explores this even more. The results aren’t exactly what we would expect either! -P
I look forward to seeing that.
I am also aiming to put together a more comprehensive video looking at how the different kind of ISO works, hopefully we can all settle this debate 😁
@@FStoppers I do wonder if there's a difference in how the image is processed. For example, I know the Sony Alpha series does noise reduction even on RAW files.
Not your fault. I had the same interpretation
The biggest problem with this topic when trying to simplify it into a RUclips video is how it disregards just how camera-dependent this topic is. Tony went hard in one direction and I feel you might have gone a little too hard in other. For some cameras ISO is almost entirely digital gain and might as well be a metatag, because you can change it in post all day. For other cameras like many Canon models, changing ISO is not reversible in post and has serious consequences. For other cameras still, it's a hybrid where it's analog to a certain ISO and then a larger digital component at higher ISOs, and then fully digital in extended ISOs. The list goes on and on. The a7 III is a fun example because it does demonstrate invariance but at different steps like you mentioned. There's a lower stage from 100-640ish, and then from 800 onward. My point is it's all too camera-dependent to be making any claims about ISO in general, unless the claim is look how funky ISO is. And so, I didn't mind Tony's video because I felt like that was his key point. I also enjoyed your video because your counterpoints further demonstrated that it's not as simple as most would think. Cheers! 👍😃
Thanks Gerald, we're all waiting for you to solve this issue once and for all 😁
@@DaveMcKeegan You did a great job! The animations were great too!
Thanks Gerald, that means a lot.
I know it's a complex, camera dependent topic, and yes I probably did go a bit far in the opposite direction to try and balance out what Tony was saying.
But the can of worms has been opened now 🤣 so think I might go for a more in depth video looking at the numerous different options and how they work
@@DaveMcKeegan Here's a resource that might help you in your journey: www.photonstophotos.net/Charts/PDR_Shadow.htm#Sony%20ILCE-7M3
This one's for the a7 III, but they've charted a ton of cameras. Compare that to a Canon and you'll see what's going on.
@@DaveMcKeegan Please do! I really dig this information and like Gerald mentioned it seems like ISO is generally all over the place - These videos help us all at least try to understand what we're messing with. Cheers
Coming from an audio background, I think iso is best understood as "gain".
The lower the signal, the more the noise will be visible/audible when amping it. The lower your iso, the higher your signal to noise ratio.
In audio, you try to get your signal input as high as possible without peaking. Basically with iso it's exactly the same. Only it's things like camera shake you're trying to avoid as well.
I was thinking very similar to you. It reminded me of my karaoke setup and how audio works
Most video cameras used gain. Now, with the ISO craze, they have the option to use gain or ISO values. This bothers me a bit because every camera has a different "native ISO", so knowing X camera has the least grain at ISO Z, doesn't mean Y camera will have the lowest grain at ISO Z. But, they'll both have the lowest grain at 0db gain...
@@rickyjanzen6684 in some ways the video guys definitely have the edge ✊
In olden days, you could set the ISO setting, shutter and aperture based on a light meter based on film sensors and get a reasonable result. So, I'm not really sure what's changed.
Although, it is worth noting that it never quite worked because digital sensors tend to blow out highlights in colors of red, yellow or magenta more so than film would. And you'd need to manually account for that if you've got any of those colors being directly lit, or you could wind up with unexpectedly blow areas in an image.
@@SmallSpoonBrigade the difference is: on film, the ISO is the actual light sensitivity on the film. Changing the ISO just told the camera what film is loaded.
All you could do was push or pull with a "wrong" setting.
Now, the ISO setting boosts the Signal from the base iso, kind of like the gain on an amplifier. You still really only have your base ISO unless you have a camera with multiple base values (forgotten the term).
But that's the reason why you have noise: the lower the light the higher the signal to noise ratio on your sensor, the more noise you get by bumping it up.
It's amazing how many exceptional photographers execute their craft beautifully while remaining blissfully unaware of these debates.
Keywords exceptional photographers not supposed youtube influencers
@@odemata87 I don't think that most of the youtube photographers actually consider themselves to be great photographers. Jared Polin doesn't, the F-Stoppers don't, Matt Granger doesn't and the Northrup family consider themselves to be generalists without a particular expertise. So it's all good.
No one really cares about these topic except the tech geeks and influencers who need to have a "topic" to talk about and generate view count.
In short - each camera is different and each camera has different technology and different tweaking made to them.
Many exceptional photos are taken on fully automatic settings too. If you just remember to use the lowest gain/ISO settings and expose using shutter speed and aperture as much as possible, you will be fine.
@@clifftotten7609: Your post is confusing. "... expose using shutter speed and aperture as much as possible..." does not match with "Many exceptional photos are taken on fully automatic settings too."
Tony Northrop: keeping youtubers employed since 2014. This shit is gold.
And keeping fanboys brainwashed since 2014.
Im pretty sure its not just the exposure the camera changes. Look at mobile cameras today. Software has already come a long way. How hard would it be to make one for a bigger camera?
@Steph Format the only stupid ones are those that defend liars.
Dave owes Tony a six pack!
@@melvinch Except attacking, trolling, being nasty and disrespectful - what is your purpose in this? I never ever read any explanation, argument, proper grown-up discussion fro your side. You are just nasty, petty wanna be "photographer" (even though I doubt it), who is seeking his 5 seconds of glory on forums. Your work and contribution are "fabulous" here and I will definitely try to find a way to get you of this social media.
This topic is something astrophotographers are well aware of since we are almost exclusively dealing with very low levels of signal and have to be very aware of what effect ISO has on noise. I think you explained the concepts very well for technically-minded beginners. What you are discussing starting at 04:10 is what astrophotographers call "read noise". Changing the ISO has typically been a balancing game between not increasing noise by setting the ISO too low, but maintaining enough dynamic range that one doesn't cripple color in brighter areas (for astrophotographers the stars). ISO-invariant cameras have been amazing because it means the ability to shoot at ISO 400 (more dynamic range) with still very low added read noise. I don't know of any serious astrophotographers shooting at ISO100 though for the reason you mentioned: no DSLR is truly "ISO-less" as that would mean zero read noise. My 2 cents.
"Out and out wrong" is too strong. After I saw Tony's video, I took some pictures and tested his results. As far as I was concerned, there was no noticeable difference and it made a huge difference to my thinking, which was treating ISO as thought I had a film camera. Even though you are right, you should be praising Tony for highlighting this and many more things like the true meaning of f1.8 stuck on the front of a camera. Yes I fell into that trap. f1.8 wasn't f1.8 and the camera manufactures should be shot for misleading people. Lying.
The problem is that not every iso step applies the same formula for analog gain/digital gain and this is different for every camera. Sometimes going from 200 to 400 applies digital gain and going from 400 to 800 dials back the digital gain and applies analog gain, thus the noise pattern in the shadows changes. Ideally you should test every iso in your camera to find the ones without digital gain and stick to those. Not only will it preserve dynamic range, but it'll force you to light your shots and/or choose an aperture that allows your camera to perform at it's best..
This is what I am curious about, thank you. so do you have a testing method for this? Or is there one provided some where and or do the camera manufacturers state their ideal iso setting? (should i set up a colour chart and photograph it with a metered setting to start and then go through my iso setting then apply post and see what setting needed the least amount of adjustment?
I'm an electrical engineer and I do signal processing. Your video is very correct. Amplifying a signal before it's processed (known in the business as "sensor adaptation") is indeed going to prevent having to amplify the noise added by the whole processing chain. What Tony stated is equivalent to saying a digital zoom is just as good as an optical zoom : it's not, because the optical zoom scales up the picture before any processing, whereas the digital zoom must create pixels for which the original signal doesn't contain any data.
Tony is a good source of a advice, but this is RUclips, world HQ of self-proclaimed experts. No matter how good someone seems to be, you had better double-check anything they say. Feel free to include me in the lot.
Hi Jean. In any precision electronic readout, you would like to be limited by the sensor's noise and get the most dynamic range out of it with the least amount of noise.
So let's get paint this picture, the sensor generates some signal and some noise, and by default, we get a 8-bit signal out in the final picture. Which option is better:
-Building a front end 2x amplifier that an input referred noise of half of the sensor and an 8 ENOB AD converter.
-Build a 9 ENOB AD converter with an input referred noise of half of the sensor and decimate one of the bits for a 2x amplification in the digital domain.
In both cases, we would end up with the same image in terms of SNR, and if we do smart choices for the AD design you can save significant area and power.
If you check, manufacturers use 12-14bit SNR a/d converters, but each pixel generates a 8 bit information... that is just digital amplification, and if the noise level is sensor limited you are able to digitally "amplify" that information without losing any information. And that is pretty in accord to the image that the guy show, from ISO 100 to 12800 is around 16x but the SNR is not nearly 16x worse... one can quantize it but it is barely 2x worse, wich means you are 1 bit off in you adc...
The digital zoom analogy is a bit off... if you happened to have a digital capture that is 2x better than what your eyes can even imagine to ever get, a 2x zoom would not result in any loss of information for you. And that is one point that the manufacturers play around, but with sensor noise.
In the end manufacturers do play around a lot with pre amplification-dynamic range-digital quantization to get what they look for with conversion time/power/area/flexibility/SNR.
@@jethro606 you are splitting hairs my friend. I understand what you wrote, but who else does ? My comment wasn't aim at fellow engineers and in the strictest sense, you have to agree I'm correct. Taking into account parameters like the limitations of the human eye, for example, isn't really what we're discussing here 😉
I'm an electronics designer, specializing in mixed signal design and data acquisition. When I watched Tonys' Video I thought exactly the same... naaah there is no way that's true... I said to myself either there is some biasing on the sensor and the bias level is tuned for the ISO setting or there is a PGA in front of the ADC to sort of "auto adjust" the signal or set based on the ISO settings... Then i looked it up ... at least from the info i could find nope... nothing.. nada... of the cameras i could find some sort of block diagram or schematic there is no such stage... it is possible that Tony is correct... at least for some cameras
@@TheNefastor Hey, I know what I commented was not for everyone to understand, but I guess you would and I am fine with that. Also I would like to attract more of other engineers attention so they could correct me in the case.
But if you got my point you would see that a pre amplifier is not always necessary. And like the other fellow said, you don't always find a front end PGA in such sensors because the ADC already bring noise levels so low that doing post digital amplification or analog amplification has nearly 0 difference. And that is what both guys showed in their videos, amplifying a signal 16x and not seeing 16x extra noise sort of means that.
Hello, I'm a male gigelo, I sleep with women for money.
In other words: For best results, know your camera. Thanks for the great explanation.
The most important part of the video is one some people seem to be missing. Yes, ISO is applied gain, after the light hits the sensor (as Ken Wheeler might say) but it's applied gain BEFORE it goes to the converters and then the processor,, thereby amplifying it at it's purest state, without amplifying any of the noise associated with said converters and processor. Is that correct?
That is true. But when the noise associated with converter and processor gets very very minimal, amplifying the signal before or after becomes almost irrelevant. That's the whole idea of ISO invariance.
I praise you, Tony and Gerald for even bringing this topic to discussion and debate! This was excellent education, and entertaining!
I think Tony’s point still stands.
FStoppers did actually see the colour shifts that you mention, but came to the conclusion (which I think is right) that for the difference seen the results are basically the same after post processing.
I think Tony’s point was to educate people here and show them something that they may never have realised. It certainly opened my eyes on this.
I don’t think Tony suddenly expects everyone to suddenly go out and shoot at ISO100 for a wedding reception without flash. His point is theoretical, and interesting.
Except ISO and post processing are fundamentally different because one works on an analogue signal while one works on a digital
ISO is like the preamps in audio equipment which will always yield better results than turning up the volume afterwards
The ISO values should be adjusted to reflect the sensor’s light transmission factor. This is what the ISO measurement is supposed to measure. ISO is supposed to be a standard for how much light the sensor captures, AFTER TRANSMISSION FACTOR. Also, the point of the RAW format is to capture exactly the readout from the sensor, so the AD converter output should be identical between different ISO values, if the camera circuitry isn’t applying electrical amplification. If they are, it would seem of critical importance that the amps are very high quality. If they are not high quality, bumping up the exposure in software will be similar in noise level, but waveform-clipping artifacts would still be a problem.
This is where I'm a bit skeptical. If the ISO we get in the camera is just in relation to the baseline for that particular sensor, then it won't be possible to use external meters without manually adjusting the readings, but I don't think that's the case. When I used to use an incidence light meter, I don't recall needing to adjust the output to compensate for the fact that I was shooting on a digital sensor.
I’m a beginner and what I got out of his video is DONT TRUST THE ISO SPECS ON A CAMERA SO MUCH !! THE STANDARD IS NO LONGER A STANDARD!! And I believe he’s talking to people like me the video wasn’t intended to school professionals !!! Of course it’s more technical in depth! Don’t get so defensive. He didn’t insist the earth is flat !
You missed the main point (while making a few good points). Tony is *primarily* pointing out that camera manufacturers exploit the inaccuracy (and non-standard lack of uniformity) of ISO as a unit of sensitivity measurement; thereby allowing for more flattering ISO specification declarations. It is largely misleading, and that's what he's pointing out.
Except it's always been like this, even back with film you would get variations in results from one brand to the next at the same ISO
Of course, but how many people know this, esp. beginners?
The discussion here generated by Tony's video enlightened me about some cameras using a different tpye of amplifications at different ISO settings; before I thought it was a straight linear amplification.
The manufacturers should be more open and tell us more about how they produce their raw files and JEPG's.
@@DaveMcKeegan Yes! Also stated by Tony (re: "always been like this"). But I still think reviewers such as yourself keep peeps honest by monitoring the fringes of speech - so keep up the good work! :-)
Kind off you to speak for Tony. I am of the opinion that if Tony had wanted to say that he would have. But he didn't. He deserves full credit for obfuscation, inaccuracy and ignorance.
Agreed. Tony is 100% right.
Also cameras of same make and model can vary. They are not 100% equally due to sensors are made of organic materials, if I don’t remember wrong. Can also differ when they where made.
ISO don’t care! Do we really think that camera manufacturers put a settings on the camera just so we could mess with one more knob? Think about it. This whole conversation is really out there. This is what I do know; Set the ISO as low as you can and shoot for correctly exposed image; Know your camera and know it’s limitations; Settings between cameras and settings even between models will vary; If it’s dark increase your shutter speed and open your f-stop and when you run out increase the ISO; You get what the light gives you and what the manufacturer provides. Nikon, Canon, Fuji and Sony all want you to be able to take the best picture possible. If the best solution was post software then they would provide that as a proprietary means of pushing your image. And they would drop the ISO setting. I don’t mind a good technology discussion, but this one is just absurd... to me!
Below I have two articles. I have summarized information from the Stanford article so nearly all credit should be given for the below summary to them. I’ve added my own explanations to some parts however.
The number of photons arriving on a pixel for any given exposure is governed by the Poisson Distribution. From this we get a mean pixel value and a standard deviation pixel value.
The signal to noise ratio or
SNR = (mean pixel value)/(standard deviation of pixel value)
SNR(dB) = 20*log10*(mean /sigma )
Therefore, doubling the width and height of a pixel increases its area by 4×, hence # of photons by 4×, hence SNR by 2× or +6 dB
Not all photons will produce an electron. Quantum Efficiency is a number we can calculate as
Q.E = #electrons/#of photons
For a typical digital camera this will be less than 50%. For back side illuminated sensors this will be greater since the wiring doesn’t obstruct photons. Larger pixels allow for longer measurements of small currents to be made and therefore a more accurate measurement.
It is reasonable to say that a full frame sensor with let’s say 24 megapixels will generate a larger signal to noise ratio than that of a crop sensor with 24 megapixels and thus less noise. This is assuming that the tech in the signal chain after the sensor i.e Amplifier, ADC etc is the same and therefore the noise contributions from the rest of the signal chain are the same. This is of course assuming the photo sites on the 24 megapixel camera are larger which would naturally be the case since they would be expanded to fit the larger surface area of a full frame sensor vs the crop sensor.
Pre gain stage before the ADC is typically an analog gain amplifier. If I assumed an inverting amplifier the gain equation is:
Vout=-Rf/Rin*Vin
Vin would be the signal of interest plus the inherent noise at the input of the amplifier pins (1/f noise, broadband noise etc). This would be where the base iso would come from and the noise would vary depending on the amplifier of choice.
The ratio of resistors chosen would determine the gain of the signal Vin and maybe be a “standard gain” whereas the quality of the amplifier would determine the noise level. One other option is for them to use a PGA (programmable gain amplifier) with values 10,100, 1000 (pre ADC).
Post ADC the signal can now be infinitely gained up digitally therefore there’s no limit as you said.
Citations and References:
pdfs.semanticscholar.org/e0af/181fd692ef06b7a114948164c399b8869be0.pdf
graphics.stanford.edu/courses/cs448a-10/sensors-noise-14jan10-opt.pdf
NERD!
Tony's main point seems to be that different cameras implement different standards for ISO. This doesn't make ISO fake. It really doesn't matter. Once you become familiar with your camera's ISO standard, you can use it to produce properly exposed pictures. Most modern cameras have an auto ISO that will nail the exposure.
Yes you can change the ISO in post, but not the shutter speed and aperture.
The real lesson here is never substitute gain for light, whether through ISO or in post. Choose slower shutter over higher ISO if the situation allows it.
No?
Depends, if you're shooting sports and need a sharp image, then ISO noise can be reduced in post a little, a blury subject can't.
What few realize is that in the film days, ISO was changed by using smaller or larger photosensitive particles. Higher ISO would be "more sensitive" because larger grains would have a higher chance at being hit by photons than smaller grains. This is also what gave higher ISO film that "grainy" look. Back then, the equivalent to what digital ISO does would be to develop your negatives and then overexpose or underexpose when producing your positives.
If we want to recreate film ISO in the digital world, the closest would be what is called "binning", where the camera groups adjacent photosites to be treated as a single pixel by adding the charge of all those before going to the ADC. As the ADC gets the added charge of 2 (1x2 binning) or 4 (2x2 binning), and lumps that charge into a single pixel, the pixels (not the photosites) is really more sensitive to light, because ot has the light from all those photosites. This will cost you resolution (and aspect ratio, but that is another topic), which would parallel the grainy look of higher ISO film.
I was saying we need to standardize the lens (use the same lens adapted for each test camera) and test for shutter speed consistency.
Good call on the sensor glass.
Someone should do a scientific test.
I agree with you. Also Fstoppers used for their test an old APC sensor camera in the form of a Nikon 1DH and were comparing it to the same ISO setting performance in 2 modern day Full frame sensor cameras. This in part would account for a difference. I do not think people always realise or appreciate that the size of the sensor is another crop factor that one has to consider. Other than Aperture and focal length multiplication. So an APC ISO equivalence of an ISO 400 setting on a full frame sensor camera is more akin to an ISO of approximately 156 (Canon) 177 (Nikon) APC sensor camera. Tony Northrup gives the formulae: ISO setting of FF divided by the crop factor (squared).
You guys are splitting hairs. ISO is a marketing scheme. I am a machinist at several ISO companies and some of those companies manufacture poor quality products. The rating standard means nothing. As far as photography is concerned ISO really doesn’t matter. Just shoot away and whatever looks good to the photographer and or customer will suffice. Sometimes good enough is good enough. Perfection is only found in nature.
Cornelius Fischer states:
"Haven't read the article, but it is right that high ISO does not mean, a camera sensor becomes more sensitiv for light.
This could only be done by applying a higher voltage to the CMOS sensor. Without active cooling (as for example done on CMOS/CCD sensors on fluorescence microscopes) the sensor would just burn on a higher voltage.
That's the reason why regular camera CMOS sensors have a native ISO. This is the baseline sensitivity based on the applied voltage.
So how is a higher "sensitivity" achieved. Simply by applying a signal gain factor. This is applied on the CMOS output signal voltage with an analog circuit and prior the ADC unit.
Sensor and ADC have their specifiy/fixed voltages, resulting in the native dynamic range. When I now using the gain (analog amplification of my sensor signal --> higher ISO), I still have a maximum limit for digitation on my ADC. Therefore, increasing ISO/gain results in a drop of dynamic range and at the same time an increase of signal noise because of a reduction of signal-noise-ratio.
The better all components work together and the better the signal-to-noise ratio of the baseline signal detection is, the less noise you will see on higher ISO values.
So, higher ISO will not mean more voltage/sensitivity on a sensor, but you will get a analog signal amplification. This always results in a different data, then by just pushing RAW file.
PS: I have worked as lab technician for about 15 years in the field of flow cytometry/microscopy and met some cool guys from Olympus/Sony/BD/Leica during that time. These guys are awesome with their knowledge about the technical things behind sensor systems. 😳"
"Jamie Moffat - technically not really the same. Overclocking really changes the clock frequency of the CPU.
An analog signal amplification increases the voltage of an analog signal. E.G. using a gain factor of two, input voltage is 1V, output will be 2V. The signal still has the same frequency.
I have worked on flow cytometer microscope which was using two CCD cameras which had a variable readout frequency. The system could measure the speed of a particle in a liquid flow and could adjust the line based CCD readout. At the end a particle passed the CCD chip while the signal get passed line by line downwards with the same speed as the particle.
That results in a true signal amplification right on the chip.
This is actually the same technique as we photographers use, then we do a longer exposure on a fast moving object while moving the camera with the object. The machine is just adjusting the line based readout to the speed ob the moving object.😎"
i maintain my silence there because of chelsea...
The Northrup channel would be much better if Tony left :)
I get the point. Haha 😄
😅🤣😂
Yes, yes and yes. Further for certain sensors, careful overexposure by higher ISO allows for a better dynamic range in the darker areas.
that's great are you making compelling content???? That is the main and will always be the main question otherwise we have a perfectly exposed perfectly grainless character list the dull picture.....
@@nelsonmoreno3680 I admit that my comment is sloppy, in that it does not explain detail. Further I forgot to mention that it is as well a factor for the highlights and to gain the full benefits might require a controlled overexposure in this area. The effect is, that you have more color information available in the range you want to expose, than if you had when you exposed in a lower iso range (by practical means having a higher dynamic range). Further down you find a video explaining what I mean. You can see there how clipping is happening, resulting in an image that looks completely dull, lacking color information.
Tony Northrup on the other hand is talking about noise, what creates a somewhat grainy picture in result. It can be fixed at the price of a more blurry picture. Dave is arguing that by technical means, before converting the image into digital, you can reduce this noise and he is correct with that.
And yes, you are right in what you mean usually the noise in the darker areas is stronger, since: there is less light information, it is more noticable to amplifying the blacks and with that the noise it has. But that is not what I mean. Check out the video to understand the details:
ruclips.net/video/2sshGdMgJxQ/видео.html
Tony’s right. 1) On most digital cameras, ISO is just digital gain. 2) And one reason the pictures lifted in post are noisier is because the camera is applying some processing even to RAW files. 3) He’s also right about ISOs being rather arbitrary.
He has more wrongs than right, making him the ideal false prophet of photography.
From what I've read most Canon cameras are not ISO invariant. I did my own test on my 6Dii and there was way more noise on a lower ISO with the same shutter and aperture when conditions were dark.
Maybe his statement is true for some, many, or maybe even most cameras. However, it seems like it isn't true for most Canon models and that isn't exactly a no name manufacturer.
Except arguably ISO isn't digital gain because it's the analogue signal that is being increased.
ISO in cameras is like audio recorders, it's the preamps that boost the audio before it's saved, not just turning up the digital volume later
Even if the signal boost is a digital one - it performs differently than the lightroom/photshop/whatever algorithm performs like, thus giving different results. The in camera boost is "tuned" to the specific sensor and processor.
@@DaveMcKeegan So, ISO values are equal to the amount of recording level on recorders? Or is it equivalent to the "High Gain" setting most (hand) wave recorders have. I have the Roland R-05 and I use always the Hi gain setting instead of turning it off and boost the sound levels in post during editting for example. Is it also better to record in the lowest recording level setting (in normal) and boost in in post anyway in order to get the cleanest sound since High gain settings is only a boosted analog signal (real world sounds hitting the microphone and gets amplified before getting it transformed to a digital signal)?
To add a little, ISO speed is not rated for raw at all. Here is from current standard, ISO 12232:2019: "ISO speed and ISO speed latitude values shall not be reported for raw images, however, because with raw images processing that affects the values has not been performed".
And pros shoot RAW not jpeg like the 99.9999% of the “mediocre is good enough” FB, IG, and YT crowd.
Tony's video itself was the reaction to somebody else's video published one or two days before. I watched that also, but don't remember who it was. It was about higher DR with higher ISO at highlights and that ISO actually only is a parameter metadata at raw photos.
When I pick up a camera I am mostly concerned about getting the shot. I need it for myself (to prove that I know what I'm doing). My client needs it(I'm under pressure). That gain(iso) is just another tool in the tool bag. I just need it to do what it is universally understood to do). I am so thankful to live in a world where I have many choices when it comes to photography. (Camera brands, third party lens, third party products, RUclips channels, and competition that makes this industry sizzle). I appreciate everyone's feedback. It is an incredible journey to learn and discover new things. Anyone who decides to make a RUclips channel based on photo/video is very much welcomed.
To clarify what Dave said, on a CMOS sensor, which is now use on pretty much every camera, before the AD converter ( which is an independent componant of the sensor ), there's actually a charge converter included on each individual photosite of the sensor, which produce a voltage depending of the charge of the photosite, which can be then send to the AD converter.
It's a little confusing, but without the charge converter for each photosite, it's not CMOS, it's a CCD sensor.
( IT CCD utilize lines of buffer photosite between normal one, and just serve as emptying active photosite, and transfering those charges to the charge converter, and then the AD converter.
there is olso the FT version which use a second "sensor" that just serve as buffer, so that the active sensor have a better fill factor. )
exactly, here is some old site about that from 2002 (the days CCD was the king of the picture and CMOS an unwanted baby), and light and color. very nice read.
micro.magnet.fsu.edu/primer/digitalimaging/index.html
The comment toward the end about shutter speed is the first thing that came to my mind. If you're putting your camera on something like aperture priority and you're taking a low-light photo without a tripod, increasing the ISO is very useful since that allows for a faster shutter. So, not only are you getting a better photo than using a low ISO+post production, you're also getting a better photo since you're not getting motion blur from unsteady hands.
digital is really just a single, reusable film frame (i.e. the sensor)
I did not understand how interference affects digital signal after ADs? What happens with the 1s and 0s?
As a photographer who shoots in low light I would never consider shooting in low ISO and boosting in post as the results would be notably undesirable.
Capturing images work a little bit more complicated. But I agree with Dave in most of the things he said. People tend to forget (or maybe do not understand the basics of physics) how devices they use work.
To conclude: there are many factors which influence the final results (a photo-cell size, it's capability of producing the current from some amount of photons, etc.). However, the ISO is adopted to our needs because we can somehow understand where we are when capturing photos.
I think you got caught up in the details... Though Tony did use those examples I believe he also expressed the difference was minimal. Not in those exact words... The gist was ISO was a standard in film and a ISO100 from mfg to mfg would result in very similar exposure settings without regard to camera mfg. However, digital world there is no standard. (Which completely defaults the use of ISO... Which means "equal") So though you are correct... I also believe Tony is as well. ISO in the digital camera is a completely false statement as each mfg, and even different cameras from the same mfg will require different camera settings in the same lighting condition.
ISO actually is an acronym for International Organization for Standardization, iso means equal
@@chrisholmesphotography2456 I disagree. I will concede ISO logo is used for the International Organazation of Standardization. Which initials would be IOS... ISO = Equal.
@@dtd2844 ISO is in fact the acronym, you can look it up very easily. They provide standards for many things.
@@chrisholmesphotography2456 still no... yes you look it up it will lead you to exactly what you say... except the acronym for International Organizations of Standardization would be IOS... therefore.. ISO still = equal.
Quote "According to ISO, ISO is not an abbreviation. It is a word, derived from the Greek isos, meaning "equal," which is the root for the prefix iso- that occurs in a host of terms, such as isometric (of equal measure or dimensions) and isonomy (equality of laws, or of people before the law). The name ISO is used around the world to denote the organization, thus avoiding the assortment of abbreviations that would result from the translation of "International Organization for Standardization" into the different national languages of members. Whatever the country, the short form of the organization's name is always ISO."
Link -> searchdatacenter.techtarget.com/definition/ISO
Sometimes its not always about absolute right or wrong, but I appreciate a good counterpoint every now and then. That's the type of thing that keeps you guys on your toes, and honest. BTW, I tired fixing it post with my MK III and it didn't work. Not even close.
Tony's point about ISO 100 is that it isn't set to a standard level. ISO 100 doesn't give consistent results across manufacturers or even cameras. You've tackled him on a different point.
Yes, however film didn't give truly consistent results either, it would vary from brand to brand as well
@@DaveMcKeegan I wonder how big the difference in ISO was with film compared with the differences in ISO within digital camera's today. If the difference in film is a lot smaller then digital could have been a standard within tolerances. Wondering when is something consistent. What makes the difference between consistent en not being consistent. Where to draw the line. In my opinion film was consistent enough, within manufacturing tolerance, to be able to say that there was a standard. ISO was the parameter that told you how long to develop the film. It must have been close enough to give good consistent results. Digital might be off a lot more than film ever was.
@@primefotoNL Yes, back in the day, film had to meet a certain standard within allowable tolerances so diferent brands of film were very close. Digital camera have a wider variance in that standard so ISO numbers aren't as closely accurate between brands as film was. I was shooting film from the 70's till the 2000's.
agree. ISO100 in different brands have different approach to the ISO100. It is not standardized anymore.
tsk tsk tsk! i could not even finish this guy's vid.
@@DaveMcKeegan Not just from brand to brand, but also from developer to developer - the same film can have significantly different actual ISO when developed in, say, D76 and Rodinal
After ADC signal is already digital. Passing around digital signal does not introduce any additional noise. Only possible explanation why ISO can't be applied in post, is either that gain is applied with analog amplifier before ADC. Or if gain is applied after ADC digitally, but ADC resolution is greater than amount of bits used per pixel in RAW format. Most probably it's first one.
I think the point you make, Dave, that raising the ISO means that the camera boosts the analog signal before conversion to digital, is central to your argument. It totally makes sense, and would account for the increased noise (and also notice color shift as well) if one merely shoots at native ISO, then raising values during editing. Digitizing an analog signal always results in data loss: a curve is converted to a series of discrete steps, and that loss results in noise. Increasing exposure of a native ISO image during editing magnifies that noise; by boosting the signal (shooting higher ISO) before digitizing, you are retaining as much information as possible before the image is digitized: you are magnifying the analog curve, not the discrete steps. So the bits of noise are smaller and fewer. Fstoppers even remarked that they noticed more noise in the greatly underexposed native ISO image, but then blithely went on to ignore this observation, saying "Tony is right." I'm not a Tony fanboy, but like T&C, and Fstoppers too, but in this case, I think they're wrong, and I believe you explained why. Thanks, and good job.
Boosting the analogue signal before quantisation is indeed the central point. Dave has mentioned, though not stressed that quantisation is not a lossless process, it actually removes data from the analogue signal to be able to represent it in a digital form, so all the amplification is better to be done before that happens. Dave said "the interference" which is, in my eyes too soft of a term, because parts of data actually do get thrown away, not just interfered with.
Also, and it has never been mentioned before, amplification of the analogue signal may, and probably does involve multiple stages like filtering, compression etc. Actually, the analogue part is exactly the place where the miracle of developing new sensors with better ISO range happens every year. Only that we, as a broad public are not aware of that fact.
Well, just to summarize of what I've just said - Tony had represented a pixel as an empty bucket for collecting photons, connected to the input of a ADC, while in the reality it is, of course, much more complex, hence the mistake.
Apart from quantization noise (which isn't that big of a deal when you do the math) the biggest source of noise is the inaccuracy of the ADC itself, the so called "read noise".
The Z6 perform great at high ISO because it has an extremely low read noise and good quantum efficiency being a BSI sensor, however at low ISO if you try to boost in post you will notice the different read noise between the normal pixels and those used for the AF, which results in the banding you can see when trying to open shadows 5-6 stops.
At high ISO the difference in read noise is unnoticeable so the banding is completely swamped by other sources of noise and becomes completely irrelevant.
Yeah, it's like with digital audio. Before digitisation you want to get as much analog gain as possible without any clipping. If you recorded everything with 36 dB of headroom and then boosted it in post, you'd be wasting a whole lot of bit depth.
The point is, that modern ISO invariant sensors do NOT boost the analog signal anymore. The analog/digtial converter is right behind the sensel wich brings other advantages too. This is the major flaw in this video. Analog amplification was performed on "older" sensors. This is also one reason why not every camera is ISO invariant.
Based on the description of the circuitry, an Analog signal (voltages) from the sensor passes through an Analog amplifier before being converted by an ADC to a Digital facsimile (numbers) of the original signal. The amplifier boosts the Analog signal and the thermal noise of the sensor plus introduces it's own thermal noise.
A voltage has an infinite range of possible values but an ADC must select a discrete number as a best fit to represent it, imposing quantization distortion, another source of noise. An ADC also has a limited number of numerical representations available to it (dynamic range) so converting a low range (under exposed) signal will vastly under utilizes the limited range available to the ADC.
Dave rightly states that amplifying the Analog signal as soon as possible minimizes the potential to pick up additional line noise and as the amplifier is in circuit at all times, using it or not using it introduces the same amount of additional noise.
Post processing simply scales (multiplies) the numbers of the Digital facsimile and potentially adds it's own noise (rounding errors). Processing a low range digital signal (all small numbers relative to the available range) into a full range one will involve a lot of rounding error.
To my mind, the less post processing you have to do to a wide dynamic range source, the less noise you will introduce in the system as a whole. Early amplification to get the best use of the available dynamic range of the ADC followed by minimal post processing to avoid injecting and enhancing noise just sounds sensible.
Think of ISO as conditioning the Analog signal to get the best performance from the ADC and minimize the potential of adding to or enhancing the unavoidable noise inherent at every stage. PS: I'm a telecoms guy.
Finally, thanks for clearing this all up! I've read up on it all after Tony posted it the way he did, as it sounded really fishy. Your explanation now was definitely the easeiest to follow.
Easy to follow but his main point is wrong. ISO invariant cameras do not boost an analog signal. It gets converted to digital pretty much right after light hits the sensel and the rest of the processing (including ISO) is on the digital part, so it can be altered later.
Actually it will and does effect you if you are you using an off camera light meter to manually set up your shots.
Great job describing how ISO works, as an engineer "sensor sensitivity" felt wrong but that's how everybody describes ISO and if it wasn't for Tony's boldness & ignorance many people would still be clueless. Should we flame Tony for not knowing what a DAC is and how signals work in electronics? Hell no, I pretty much like his content and we are all evolving here. It's actually very nice of youtubers to step up and put each others concepts in the test, that's how science works; you publish an idea and expect others to (dis)prove it.
I assume you mean ADC instead of a DAC.
@@henri.witteveen the point is none cares except the engineers and frankly I wouldn't build my own camera with so many well engineered options on the market. Just enjoy photography!
One of the issues I have with Tony is that he's not always consistent about what he does and doesn't test. Some of the stuff, is right despite making no sense. My Canon 70-200mm f2.8L IS was never sharp on my 10D or 7D, but on my 5D mk IV, it's significantly sharper than it was on either of those bodies, even if I crop the resulting image. Without Tony, I wouldn't have consider that to be possible.
But, in this case, the take home should be to apply moderate increases to the gain ahead of capturing it, knowing that depending upon the RAW format, you can probably get at least a stop or two of relatively clean signal from that point. Not to mention apply noise reduction with a much more powerful device and with smarter masking.
In your experiment, did you have ISO noise reduction On or Off? Most digital cameras employ digital noise reduction to the higher ISO images, and some do it whether the photographer wants it or not. You can't compare the noise and grain of images without stating what camera sensor and noise reduction setting was used. It could be that the images are noisier simpy because in-camera noise reduction was used in the higher ISO images but not in the lower to native ISO images. All of which can be mostly duplicated in post.
There was no noise reduction used (high ISO or long exposure)
All results were straight as the camera produces
@@DaveMcKeegan Tony is correct when he says ISO is "fake" as there is no standard to compare the value to, and not even the same manufacturers use the same definition of ISO between different cameras. Roger Clark gives base ISO as the "relative level of light compared to the maximum signal a pixel is capable of recording" which is dependent of the size/area of the pixel used. Tony is also correct when he states that a higher ISO is the digital amplification of the sensor readout. Again, R. Clark, "ISO is simply a post sensor gain and digitization range." and that at a certain level you can simply take a picture and make adjustments in post (as long as you have a good algorithm for managing "gain"). Tony is simplifying the math to make his point. You are (correctly) getting a bit more into the details.
All digital cameras have a range where they are ISO invariant. This is the range where images become sensor read noise limited. There is no IQ/dynamic range improvment from using a higher ISO value. For the Canon 5D Mark II it is around ISO 1600. The lower ISO values are limited by electronics downstream from the sensor. This is the area being debated.
I believe what you are noticing is the differences in the algorithms used between the camera manufacturer and Adobe to increase gain as digital gain is digital gain, ie.e, a simple multiplication
As R. Clark states in his article on "What is ISO on a digital camera? When is a camera ISOless?
ISO Myths and Digital Cameras", ...Future cameras will have 16-bit A/D converters thus reducing the need for ISO gain. Actually 18-bit A/D converters are needed for the top cameras, but we are not likely to see that anytime soon. With 16-bit A/Ds, we'll only need broad categories of ISO, like low, medium and high, and choose precise gain (ISO) with raw conversion. "As camera electronics become better, we will no longer need an ISO setting on the camera, as we can effectively choose it later."
So, who is wrong on ISO? Neither. If Adobe had the same algorithms that Sony or Canon uses, would you see the same image using the same true exposure? Most likely.
see: www.clarkvision.com/articles/iso/
Well done! 👍
you both have a point (tony and you). but i dont think there could be a much difference between boosting the singnal before and after a.d. i think the cameras put in noise reduction even if you disable this option
The difference is, as you pass a signal round you introduce more noise, if you amplify the signal later then you amplify the noise as well.
As a basic example if you get 4db of light signal (not the correct units) and you're going to boost it by 2 stops up to 16db, if the circuit introduces 1db of noise in the process then boosting the signal straight away will give you 16:1 signal to noise, however if you boost it afterwards you end up also boosting the noise by 2 stops so you'll end up with only a 4:1 ratio
So much depends on the bit depth of the AD converter, the quality and profile of the signal amplifiers, the correction of rounding errors, the noise cancelling algorithms, etc. Technically there should be no noise introduced after the AD conversion as the values are all fixed into memory. If noise could be introduced after the AD conversion then computers wouldn't work as no values would stay the same. Most of the difference comes in the bit depth of the raw file. For a 15 bit raw file, a correctly exposed image will have values from 0 to 32767 in steps of 1. A dark image taken at low iso may only have values from 0 to 1023 in steps of 1. When the dark image is pushed 5 stops in post the values will go to 0 to 32736 but in steps of 32 making any gradations in tone (and noise) very noticeable. Try this with JPEGS and the result is even more noticeable. So Yes - Always get your ISO correct in camera. - I think!
wooow man, best explanation ever!
I was drawing on a piece of paper trying to explain to myself like you described but I couldn't get it together good enough. Very well said.
Base gain, (transformed to base iso) is the gain for witch you match the pixel saturation (the full well) and the amplification saturation. For instance if your full well is 60ke- the corresponding digital number will be about 2^14 = 16384 for a 14bits camera.
Interesting...Your vid made me think of something: DPReview did an article recently about the Nikon Z6 banding issue, and to demonstrate they pushed a 100 ISO file up +6 EV in post. The banding is definitely present at ISO 100 pushed EV +6. However, if you look at their scene tool, and push the camera up +6 EV to 6400 ISO (which is an IN CAMERA setting), you'll notice that not only does the banding vanish, but additionally the noise levels are a bit better too. This seems to show that no, just using software to push your EV up is not the same as adjusting it in camera. The DP Review article is entitled "Nikon Z6 image quality and dynamic range impress, but not without caveats" for those that want to take a look.
If you push the image EV+6 using ColorLab instead of RGB you can go around the banding.
What you’re saying makes more sense to me based on photo processing experience. I tried taking pictures at ISO 100 in low light and boosting exposure in post processing. Results were not good.
Signal to noise ratio is the reason a recording artist uses a preamp to boost a quiet audio signal before analogue to digital conversion.
I agree on your your take on ISO invariance. But I don‘t on your point on ISO100. ISO 100 should be same for every camera and I am sure the manufacturers try. The base ISO in my world is what the sensor captures when only A/D conversion without boosting or reducing. Some cameras have base of ISO100, some ISO200 or even ISO64. Of course these values are not exact, but not wrong by a full stop. An additional variant of. ourse is the light transmission of the lens. This is what t-stops are for as apposed to f-stops. But practically 99% of images are shot within camera light metering through the lens, so nobody cares about the actual numbers of t-stop or ISO (in the sense of a comparable, standardized number). Also a partial stop is irrelevant when using post processing.
signal boosting level never changes by changing iso except for the cameras which have dual native iso, about all of photographic cameras have only one signal boosting level which we call it native iso ,other iso levels are applying digitally
A always like viewing Tony Northrup's video's. I don't have to agree with everything he says but I do learn when viewing because his videos make you think. He also brings out "thinking" among rival video bloggers too, judging by the recent rebuttals his videos have taken. Obviously this ISO discussion has led to multiple rebuttals yet there is nothing wrong with that. Discussion is a learning tool for many.
Another topic video Tony put out recently (?) was a discussion of JPG versus Raw. That video drew the ire from at least one video blogger in froknowsphoto who took exception to the video. Again, no harm comes from opposing points of view.
I don't see the point of getting so emotional in opposing points of view. YMMV
I liked your passion about this topic which is steeped in technicalities. My criticism of various RUclipss is advertising. Are you sponsored? That is my first question.
The real issue with ISO (vs ASA) is S/N ratio. That is the heart of the issue. Increasing signal over noise. That is the electronic way of viewing this.
I came to the conclusion that it is not noise that we are trying to suppress but higher ISO has less range. Lower ISO increases range or how f-stops we can capture.
But I do like the push back here. It is good for the community.
You should be aware you're making a statement logically wrong
You're comparing your Sony a7III results with F-stoppers \ Tonys Nikon D850 results
None of you is able to make statements about the majority of digital cameras of different manufactors
Note:. - Most sensors nowerdays are produced by Sony (Fuji X-Trans, even latest Nikons)
- The digital images we potographers actually recieve (let it be RAW or JPEG) are heavily processed Sensor signals (selective denoising (referring to "edge noise"), vignetting correction depending on the lens mounted, and most obviously: converting the Bayer/X-Trans Filter Colourpatterns to sRGB \ Adobe RGB Colourroom and sampling in missing information\holes created by phase-detect pixels.
All of this is done well before the RAW image is processed in the camera, is NOT affected by any settings in the camera menu (Sidenote: genious tec: Sony a7II colourchannel correction for manually compensating for CAs of attached vintage lenses (don't know where in the processing chain this correction is made as it's a well kept secret and removed in later cameras)), AND note: ALL processing (-software) between photocell charge (from sensor readout to the RAW file) is a manufactors biggest secret and a result of insane development.
Talking about ISO: When using 35mm film, which is a physical part beeing interchangeable between film-cameras standardisation of the films sensitivity was a must have to propperly expose an image.
ISO could be determined by actual testing and was a physical property of film (reaction time of the chemical ingredients)
Nowerdays ISO is one of many imput parameters used to influence the image-processing and is therefore depending on software/programming of an individual body and even the lens used (f. ex.: Fujifilm camera and lens firmware versions)
As a matter of fact: by shooting the same image (with the same settings) trough the SAME LENS, wich is UNABLE to communicate with the camera body you could actually make out differences in ISO between camera manufactors, and even bodys of the SAME manufactor.
Mainly ISO is a matter of what a camera manufactor is LETTING you use and the manufactor is deciding which compromises can be made reguarding image-quality, low-light (high-ISO) -capability, clean-image (low-ISO) -capability
Examples: - Nikon D850 extended Low ISO of 32 (producing incredibly clean landscape Images)
- Sony a7III extended High ISO of up to 102400 (great low light images reguarding noise)
Low- and High-ISO capabilitys are quiet different topics:
High ISO means plenty of work to do in denoising (typical grain, also colour-noise, without creating processing artefacts)
while maintaining good colour-resolution, hitting correct Whitebalance with auto WB, while accurate and quick focusing
with mirrorless cameras is still possible, etc.
To summ it up: mostly a Software issue (AF with contrast AF included; AF with the use of phase detect on Sensor is also depending on the engineering of the sensor), all affected by processing power
Low ISO really comes down to physics and engineering: how much charge a photocell of the sensor can handle while maintaining valid information, inter-photocell isolating, readout process, etc.
CONCLUSION:
on modern Digital cameras ISO is mainly a parameter you dial in to influence processing the Digital Image,
Each camera manufactor is deciding which values for ISO he's letting you use (which is also heavily influenced by marketing),
ISO is mostly a software issue, and sceems to (with digital) no longer describe a "standard", but a setting not consistent between camera brands or even different bodys of the same brand
(compare Fujifilm X-T3 ISO 200 to that of an Sony a7III with same shutterspeed and lens transmission "t-stop")
I hope that did / does help you understand the fascinating world of photography a tiny bit better.
But reguardless of your camera / lens/ or even ISO setting:
Go outside and take photographs of whatever you like, AND: have fun !
That noise level is going to be different depending on what you used to process your RAW image (your camera or lightroom for example).
Pixel count, signal to noise ratio and dynamic range are only real measurable values of the sensor.
Iso is strictly software and image processor related feature. And yes you can get same or better results if you use native ISO and then crank it up in your computer but that would not be so much practical to do.
In the film industry they like to find that native ISO that gives them best image that sensor can do and never even touch that thing again. And if you have to push or pull brightnes you do it in software that comes with camera by changing that ISO value.
Even on real film stock things works more or less the same. You push or pull your stock in dark room to get brighter or darker image that also introduce more visible noice.
But, if iso really is arbitrary I should be over to overexpose in camera and bring it back. In my 80D after 2 stops adjustment the picture is useless. Taking an image that needs brightened works well. I agree about the noise and what device processes, I've been saying the same thing. If it didn't matter I should be able to recover the picture in both directions.
@@fixitrod4969 Thats because Canons sensors depends very much on image processor and their RAW files are not exactly the best ones. Go to dpreview and download image from Nikon Sony or Panasonic and push it few stops and then do same for Canon. Canon lucks dynamic range on their sensors and and highly rely on their processors and in camera software to fix things up even in raw. Thats why there is not so much use of Canon raws in terms of pushing or puling exposure.
So ISO is 95% Fake... Thanks for clearing this up...
no its not ... he made it very clear just you havent got it. Than you can say ISO is fake whatsoever even in film. because also film can be darker or brighter even the same ASA ISO or DIN (which ever measurement you use) because its dependable on developing as well as digital cameras on boosting signal. Its the same premise.
...actually, ASA was written on the Kodak box but it was exactly equivalent to ISO. ISO was on Ilford boxes. In the dark days of film we spoke in ASA or ISO interchangeably according to our preference.
@Puppet Master -
If people set their camera to ISO 100 or ASA 100, (That is, if an ASA setting actually existed on digital cameras, using older analogue terms here), rather than employing older analogue terms, what difference does it really make? I for one consider ASA and ISO equivalent meanings for photographers, even if they're not so equivalent for more scientifically minded people.
More like it's 5% fake.
@Puppet Master They weren't exactly the same to start with, but eventually the two standards organizations merged their two standards in various revisions.
Reading into this too much.. Tony was spot on.. When he said u could just adjust exposure in lightroom he didnt mean it as an exact science but its bloody close enough
Except he didn't say "they look close enough" , he said they looked "identical" which is incorrect, and that was on a more modern camera, on older cameras that difference would be even greater and could lead people into believing there is no difference when there actually is
Another amateur mislead by Tony. Great work.
Awesome explanation. So much clearer than most others. Great work!
As far as I know the noise amount with may camera stays almost the same but I get annoying color shifts if I underexpose a photo 1.5~2 stops or more where the blacks have too much red. I've also noticed that my camera (Sony A68) in general makes good photos except at ISO 800. That's like the sweet spot of the worst look I'd say.
I believe he was generally correct, the iso setting is fake because its based off film standard and this is digital. Like you have said the camera companies aren't held to a strict standard just an estimation so 100 might be 85 might be 105 or whatever, and that was the point. The 4 stop difference to just shoot at 100 will get you close, but you are correct it would be better to shoot at the correct exposure for the best result. I think he just mad that iso 100 should be iso 100 no matter what camera you are using and pointed this out by showing the results of different cameras. This is not a new problem, I have worked in the dark room before and NO film was exactly correct, they were close generally with the better brand films but they were Never 100% correct.
Best comment here!
If you think iso wasn't faked by the manufacturers back in the film days, boy, I have news for you...
ISO is not fake. It only applies to the jpg. There is no ISO standard for RAW files
Precision is not the issue. Tony said you can fix it later in post and that's simply not true. Granted there is some amount of latitude but its never as good as native iso.
In the film days (some of us still shoot film) we would buy our film in batches and test the first roll for iso. The same can be done for raw files which are not subject to any iso standards. No need to test jpgs though as they are already controlled by the iso standard
Well done. In the olden days ISO was either ASA or Din and was supposed to represent a standard measure of the sensitivity of film and offer a guide line for exposure. Film is analagous to the sensor in today's digital cameras. In film days one changed the film sensor depending on expected light conditions. That's why I loaded some cartridges with 12 or fewer frames. In today's digital I take ISO as a point of departure not arrival. Despite what the camera is able to capture with all that intervening circuitry, I still have the challenge of what software and monitor will let me render and see; every software has it's own language. I think recently Northrup back peddled his opinions about ISO.
I bet Tony is laughing at how many people he has baited into making vidoes.
Absolutely! At the time of my reply Tony has over 232K views of his video while this guy only has nearly 73K views.... Views = $ …. This guy is getting Tony paid! Tony is laughing, alright- all the way to the bank!
Tony would have had that many views with or without my video
I doubt many people will be watching my video who hadn't already seen Tony's
Tony is NOT baiting anybody! There is a TON of real misunderstanding about what GAIN actually is. Actually, the ISO term was fine in film but it doesnt belong in analog/digital sensors! In fact, the ISO concept is screwing photographers up today. Gain/ISO has nothing to do with exposure of light to a photosite. Only shutter and apertute controll "exposure".....NOT gain/ISO. The faster people understand this, the better we all are.
@@DaveMcKeegan Tony's video should be reported as fake news and removed, spreading untrue information, comparable to 'earth is flat'.
Laughing to the bank
I thought it sounded a bit off (doesn't help that he can come over as a bit smug). What I don't get is is there really that much interference in camera? Good to see your tests though to prove your point.
Thank god. I was hoping someone would respond correctly.
If they all have different ISO100, right, how can that work with the doubling and halving ISO being equivalent to opening or closing the aperture 1 stop, or doubling or halving the exposure time ?
SD cards are a hoax! Just shoot without them, and if you want your images on your computer later, use your webcam.
Off course you can shoot without SD card.
You know in your head what picture you took. No need to store it somewhere else.
There was a man years ago who could put images from his brain onto film. He did not need the camera. But he did need the film.
Johannes Labusch hahaha
Are these photos you tested RAW or JPEG ?
It's correct to shoot RAW when comparing different ISO's since many camera process or interfere with the jpeg photo and it might be the reason why you're getting different noise or colors. How do we know for sure that a certain camera doesn't process and reduce the image noise over a certain ISO level?
These were all shot in uncompressed RAW
This. I was so confused watching Tony’s video when he was talking about raising ISO in camera and in post being the same.
He has the talent of stirring shit and benefiting from it.
At 2.30 you state that Tony is totally wrong. Others have tried this and confirmed that you can get virtually the same result post camera as in camera. In camera ISO = equals gain, nothing more. Doing that at different stages, I agree, introduce minor changes in noise. Tony was not advocating ISO 100, he only advocated leaving ISO on auto and worry about other aspect of controlling the photographic image that have a significant impact on image quality.
- If an A/D has infinite precision, there is _no quality loss_ using the base ISO and correcting in post.
- If an A/D is very good, there is _no practical loss in quality._
- If an A/D is poor, there is a _slight loss in quality._
The _only_ reason adjusting in post causes quality loss is due to rounding errors.
I think Tony should have been clear that there is a difference. But the point is you shouldn't be too concerned with getting the ISO perfect when using modern cameras.
It's not easy to get everything perfectly correct in a video, and easy to criticize, as you probably noticed with your mistake regarding F-stoppers.
Karl A
Every post processing software is different. On top, being completely off in ISO during shooting will deviate you from correct exposure... without signal (too short exposure) you will always end up with a crappy image, no matter if you‘re using high ISO in camera or pulling up the exposure in post.
I just have one question.... What did he say at the end? that is his conclusion? I have just made an exercise ISO 100, 400, 6400, and really there is not much difference that post cannot solve. I am more inclined to the shutter speed and Transmission of the lenses, than in the ISO. If shooting RAW there is not that much of a difference, If shooting JPEG because you are in a wedding and have to deliver immediately, then yes, use a correct balance between ISO, Sutter, and aperture.
I don’t know your channel and technical background on this topic, RUclips just recommended this video to me, but I would like to point out, that you indeed got something major wrong here.
It is at 4:10.
You claim, that the gain/amplification of the sensel voltages is performed before it gets through the ADC and that is NOT the case with current ISO invariant sensors.
It used to be a dedicated ADC, seperate from the sensor for that. On these modern sensors, the sensel voltages go through an ADC directly on the sensor more or less directly after each sensel and that is the big thing to consider. This "short way" reduces the base noise/increases the SNR and shifts the ISO adjustment to the digital section.
Most of these cameras still do have a dual native ISO or “dual-gain architecture”, so images with ISO 100 actually do have less noise and more dynamic range than at 1600. The two base ISO ranges of an A7 III are 100-500 and then 640+.
You can’t change these values but the “rest” of the ISO is just a multiplication factor baked in the RAW file and this can be changed through RAW converters like Lightroom.
But I’m totally with you in saying that shooting ISO 12800 and 1600 + exposure boost of the RAW file are not 100% the same. You actually do lose a little bit if dynamic range and get very slightly more noise since no Sensor is 100% invariant but the results of a modern sensor under these circumstances actually are VERY close.
For my background on this: I work for one of the largest camera manufacturers (industrial, not consumer) and have a direct insight on current and upcoming sensor technology.
Thank you for the insights
I appreciate not all sensors work in that way, however Dual ISO architecture is costly to make and not all sensors are ISO invariant, so there are still cameras on the market that do work in their manner.
The point I was trying to highlight is that unless you have a true ISO invariant camera, then you will have difference results from increasing exposure in camera to in post.
I am going to do another video that delves more in depth into the difference ways ISO is done
@@DaveMcKeegan Yes, im totally with you there. Most cameras do not use this swapped ADC/ISO adjustment. It is mainly Sony who implements this in their more modern cameras but since Sony also manufactures sensors for other companys, they are not the only ones on the market.
I think the main point Tony was trying to make is, that in the "real world" it doesnt really matter if you correct your image brightness +3 stops in post (if you have the right camera that is). With Canon cameras for example you won't have such a good time trying this extensively.
I was about to write what you just said, but you already did! Nicely done! I’m also an engineer.
Good, civil discussion, guys.
Higher ISO as well means you need to down sample your resolution. Perform the "pixel binning". That is what a film did, faster film has larger silver halides for better light capture but so on lower detailing capability.
With electronic sensor (it is not digital until converted) people do not want to give up with their "valuable pixels" that they pay for.
In film era everyone learned that faster film meant you get less definition.
In digital era most so not ever learn that they should lower the resolution and apply different demosaic method.
In film era everyone knew that slower film have larger enlargement capability or possibility crop tighter.
In digital era most believe they can print as large as ever regardless the ISO setting.
Now try this.
Take 24mm lens and use ISO 200 to photograph a scene.
Then take 50mm lens and set ISO 1600 and capture same scene with four frames and stich them together. Down sample it to same resolution as sensor has with single frame. Again, you exchange speed to definition.
Same thing can be done with single frame, you capture full frame with ISO 200 and then ISO 3200.
Now you get to files, full resolution and 1/4-1/6th of the original size (in area). So 50-33% scale from original.
Meaning, just like in film era you didn't use 135 film if you wanted 100x70 cm enlargement with ISO 64, but you used 120 film with ISO 64. Same speed, same sensitivity, same light gathering power, but just larger frame, that you exchanged to a definition in larger enlargements.
When you didn't need more than A3 size prints, you didn't use medium format.
When you used faster film, you knowledged that you do drop the possible enlargement size each stop faster shutter speed you want.
With digital, that ain't automatic, as there ain't physical silver halides but a physical static pixels in grid, regardless what sensitivity (speed) you use. It is photographer tasks to knowledge that how the print size changes based ISO speed.
And only image resources tests this!
The give you their visual testing results from the print sizes based the ISO changes.
They literally tell that example with ISO 200 you can make great 35" wide print, but at ISO 1600 a 16x11" and at ISO 12800 only anymore a 8x10".
Because they understand that sensitivity and it's trade-off.
That is as well why it is wrong to talk about "crop sensors" when people do not crop but they frame everything same.
No, of you want to talk about crop sensor, then you perform every comparisons with same focal length, same F-stop, same shutter speed and same ISO, and then you keep same perspective to subject and take the photos and then crop the larger format area to same as smaller sensor and that way compare the crop qualities.
What larger sensors offer is larger field of view with everything else same (except smaller sensors usually have better sensor technology and results a better quantum efficiency and so on better image quality at pixel level, that you elongate by enlarging files X times more with smaller sensors than larger sensors. Because people think wrongly that quality difference stays same regardless the output size (don't get it that larger ISO = smaller print with same quality).
Because they do not understand the context why photography is made!
1) How many here does constantly prints larger than 100 x 75 cm?
2) how many does max 75 x 50 cm?
3) How about family albums like A6 and A5 size?
If you belong to first group, then get a FF If you need speed (ISO 8000-51200) or get MF if you need quality from 20-40 cm viewing distance. Otherwise 4/3" does as good results in any other target audience.
Prints smaller than 150 x 75 cm at ISO 200-800. Prints smaller than 80 x 55 cm at ISO 1600-3200.
Prints smaller than A3 after ISO 6400.
Digital form real world use (4-10" screens with less than 4 Mpix) you don't get difference to go larger formats.
I think T&C wanted to give an elementary answer for the benefit of those beginners out there that need it broken down in an elementary way. I can appreciate that even though I'm not a beginner. Sometimes we forget how difficult is was to grasp certain understandings back when we first started out. Dave Mckeegan, great job and great info too.
That's is exactly the danger of his channel. He is misleading beginners ! You will learn more photographic truth from the Angry Photographer, backing up facts with optical principles. Phony Northrup on the other hand, mixed truths with lies that sounds believable for beginners.
@@melvinch Oh please, the angry photographer is a hack. He talks and talks yet never backs up his word with any photo proof. At least T&C show their work when testing a camera or lens. All the angry photographer does is say trust me I know then does his weird laugh.
@@williamneuman7783 AP has photos on Flickr. They may not be artistic but are technically competent. Phony Northup's photos on the other hand, are amateurish at best. And I'm not the only one who think that way, you can google that.
And most importantly, Phony Northrup has a fucking face that says "I'm a blatant Sony shill" written all over.
Well IF that's true I don't think it's intentional. Everyone is right in there own heart so I don't think it would be intentional. I personally haven't seen any issue where he was apsolutelly completelly wrong on an issue eather. With that said i know there's not a 100% of anything and so theres exceptions in every topic. Some people think that rair exceptions are the proof that someone is 180° wrong and that's just not the case even tho there is marriet in the accusation.
@@3439jc you're exactly the kind of supporter Phony Northrup depends on.
you made good points ...thanx for posting
only thing clear to me about ISO is there is most camera shooters don't fully understand it.
but the bottom line for me would be if they DID fully understand it, would their shots improve ?
i have spent a LOT of time studying ISO and my images haven't improved by a significant amount.
I'm not a great specialist, but I'm sure that in the 80-90's we could have had the same debat about different film makers, (Kodak, Ilford,Fuji). The different negatve films rated 100 ISO surely gave different performances in different cameras with different lens. An interesting debat though kind a sort of tempete in a glass a water... I'm glad that I don't think all about that when I take pictures.
But I think that the emphasis that Tony Northrup made which is the most important for me, as that the different manufactors of digital cameras today have all adapted their own ISO standards(and that they promote as being the most senstive). 100 ISO from one camera from another and the values of ISO we can reach is really based on the manufactor's own definition of 100 ISO.
In another 10 years I'm sure that we will have 100 megapixel sensors shooting at 25600 ISO on the market without 1 grain of noise in the frame, what will 100 ISO mean then? Maybe that already exists...
Anthony Salema they did and changed emulsion to emulsion that’s why we bought a case and tested the film. Found the correct iso and used a Wratten filter to color correct.
The "boosted" image in post production looks more noise because of the way Lightroom handles the data, you would get different results with different softwares but basically is what the camera does just as Tony said. The thing is that every manufacturer applies their own noise reduction to the picture to hide the noise
Check out F-stoppers latest video, they tested the same thing but using several different pieces of software including manufacturers own processing computer software, and none of them could equal the result of the correct ISO shot.
The thing with in-camera noise reduction is it works in one of two ways, either a filter like you see in lightroom that softens out the noise (but that softens the images), and true noise reduction which takes a dark frame with the same settings to measure the interference and then remove it from the original file (but that takes time for the camera to process and needs to switched on
Shout out to the Tony and Chelsea fanboys who are going to dislike this video because you called out their favorite RUclipsr for being misleading. I had a feeling that video (Tony’s) was going to be wild by the title. But i couldn’t explain why it like you did. Salute sir
This video itself is misleading. His test scene has a lot more light on the higher ISO side.
Danwarb - if you're referring to the sample shots around 5:30 then there isn't more light
The scene remained consistent, I just increased the ISO (the same as Tony's experiment)
I was just showing strips of each file of which the higher ISO shot happened to include the light but the light was present in the 100 shot as well
It's why @6:00 I compared the same part of the image for iso100 Vs iso1600 and I went with the sky rather than foreground so ambient light was guaranteed to be consistent
depends on the camera. I never shoot over iso 800 on my a7ii, as it looks utterly horrible. The internal processing is really bad. I got way better results using photoshop to push the image later.
ISO = signal gain in digital cameras. No "ISO" control in video cameras/camcorders, only signal boost/gain. No ISO in digital still cameras either. ISO really is just a holdover label from film, when ISO determined the size of the silver crystal halide particles in the film emulsion which determined proper exposure times. So technically, there is no ISO in any camera that doesn't shoot film. People know what it means. Of course the meaning is hitting that point where the average person learning photography will never shoot film. Technical, technical; does not matter.
"ISO" adjustment is applied to the analog signal captured by the image sensor prior to digital conversion. Doing this in software would be the equivalent of applying changes post A/D conversion. Not the same, even if they can produce similar results.
Case in point: Black Magic cameras (and other dual native ISO cameras) make use of this to full effect with two separate native ISO base lines. If the "magic" wasn't taking place at analog signal level prior to A/D conversion, there wouldn't be much point to dual native ISO cameras, because it wouldn't produce the results actually seen.
Tony is technically correct about ISO, but incorrect with regards to what was stated with regards to signal gain applied to an analog signal being the equivalent of applying exposure correction in software to a converted digital image. No ifs and or buts about it.
Yes, ISO now is more like audio preamps, however Tony's idea is that it's like turning up the volume in post and getting the same results
@@DaveMcKeegan Yeah that's exactly what went through my mind - using a microphone preamp to adjust gain in your recording or just pushing the fader in your DAW. sounds different and you lose resolution.
@@DaveMcKeegan Well, with an ISO invariant sensor you can to a certain degree. See Dpreviews RAW ISO- Invariance test tool www.dpreview.com/reviews/nikon-d810/13
It's really a shame and misleading to label the sensitivity of the sensor "ISO". I appreciate why "ISO" was used for digital cameras--it was a term very familiar to film users and does relate to aperture and shutter speed in the exposure triangle the way film emulsion sensitivity to light does--but considering that most people taking pictures today have never shot a frame with a film camera in their lives it would be much clearer to call "ISO" in digital cameras something else. If it were instead called "gain" (I'm just throwing out a word) then camera manufacturers could assign values to it without reference to some historic standard value and without the temptation to compare one camera's base value to another's (which apparently is futile anyway). And, if I understand your video correctly, it would also be clearer--"0 gain" would then be the base value without any arithmetic boost between analog signal and digital value. That seems much more understandable to me.
David Jenkins there is no exposure triangle. Exposure is only determined by the aperture and the time. Nothing else. However I do agree with your overall point about ISO being used in digital cameras instead of gain. It is definitely more familiar to those people who came from film.
Just a comment with some math. Normal file (JPG) 8 bit covers 256 shades of grey each color. Each 1 bit increment doubles informationso - double f stop. So 14 bit RAW file covers 6 f-stop depth of sensitivity.
Question is if this increment goes both sides (into shadows and into light)? If it goes both sides we should get information 3 f-stop below measured and 3 f-stop above.
There is simply no numbers (information) to retrieve more than 6 f-stop from 14 bit RAW file.
So calculation has to be done inside camera body.
Having Canon gear I can tell you I will be in big trouble if I underexpose 3+ stops and try to bring the exposure back in Lightroom. Recipe for disaster !! If you have trouble getting the correct exposure in a digital camera, sell your gear !! sheesh.
The most important thing that is not covered in this video is bit depth. If you're shooting on a 14-bit camera and underexposing your shot by 4 stops, then it is the same as you're shooting on a 10-bit one - you're never using the most significant 4 bits in this case. That is why it is important to amplify signal before ADC, not in post processing.
Hello,
I'm in sensor manufacture company and you cannot be more right !
You are absolutely correct. I just tried a little test with my a6300 and there is a very noticeable difference, with more noise in the ISO 100 shot after increasing exposure in software compared to an ISO 3200 shot. Great vid.
Great video ! You got all the right points and explained they in a simple, but yet, technical accurate form. Some people should really understand more about electronics before saying things about photography
my understanding is that when you shot higher iso the camera does some nose reduction without you even asking for it. when you boost the shots in lightroom are you adding more noses reduction for the lower iso ones?
...IDK but, how about we all just forget about how any of this really matters, OR NOT, and continue taking photos with our cameras, processing them how we are and have in the past, posting and printing them and carry on like we are all normal? :)
The reality of it is, none of this blaggar matters or is going to change anything in regards to the way we continue to take pictures (with the current technology) or buy cameras, is it? Not for me...
Well I don't agree that it dosent matter. I prefer less noise so it is important too shot at the right f-stoo, exposuretime and ISO.
@@idontwantacallsign *YAWN*...yeah ok, good for you...continue on taking photos as if you were normal...
...it went above your head...
But then what would the RUclipsrs do? 😂
It does matter when there's a noticeable quality loss and time loss when taking ISO100 photos and increasing the exposure
If there's one thing I took away from this video, it's that "a turd in an elevator" is absolutely a new part of my vocabulary.
As someone who have actually studied electronics at the university, I am admiring how clearly and concisely you have explained where exactly Tony has made a mistake. Well done, Dave!
I agree. The conclusion here should probably be that in most cases, a modest boost in the ISO is prudent knowing that with RAW formats, you'll have extra bits to work with and much more powerful hardware to engage in noise reduction. A 10bit RAW file has roughly 4x the values possible to work with and a 12 bit has 16x as many, which is roughly 2 to 4 stops worth of data. Underdoing the ISO slightly by a stop or two is probably not unreasonable if it allows you to use the ISO that your camera produces the least amount of noise with.
Part of the issue with both of these videos is that the lowest noise on the sensor isn't necessarily at the lowest ISO, it may be a couple settings up from the bottom, in which case, you get a bit of free lunch by just using that if you can and then using that as your base point for further work. Boosting that by a few stops in post production may well give you the best quality you're going to get.
It all comes down to what Dave says from 4:00 on. Very well explained Dave.
Ha... they are pretty clever into getting views with viral content regardless if it' make logical sense.
RED offers the option to change ISO in post without any degredation.
John Johansson at that price tag, I’d hope so :-) ha
RED doesn't offer the option to change ISO in post, RAW does. Any RAW does. But what you change is... not really something that exists... You're only adding gain... noise. I know this because I also own a camera that can shoot RAW video. I don't "change ISO" for 99% of the time.
Thank you for your clarification!
You are right.
Basically the understanding of a signal path is not to complex:
At the frontend you have a nearly analog signal (not thinking about astro photography): the number of photons hitting the sensor within the color channel and the exposure time.
The signal is changed to electricity by the sensor, thereby you can control the ampifying factor by the ISO setting. (the amount of photons you control by exposure time & aperture setting).
Being analog you have effects like noise.
Next step is the critical one: Analog-digital conversion: this means the analog signal is categorized in steps (i.e. 8 bit equals 256steps, 10 bits 1024 steps, 12bits 4096Steps)
What ever is showing up in signal computing you live with this input steps. every noise from the sensor is also pressed in this steps.
The only strategy for having the cleanest signal is hitting the white point already on sensor avoiding clipping (the analog signal is above range of A/D conversion) but giving as much steps as possible.
The result is the very simple photographic rule: get the exposure right in camera as bright as possible without clipping the highlights. you can do everything in post but you can not get any information of your subject in your file which went not through A/D conversion.
Thus shoot ETTR.
It's more complicated than that if you don't send roughly the same size signal to the ADC (analog to digital converter) you will get less data in grey scale colour scale as well. So the picture will look grainy with lower light.