"THE NOISE IS CAUSED BY UNDEREXPOSURE" or "THE NOISE IS CAUSED BY RAISING THE BRIGHTNESS IN POST" This is easy to test yourself, but I did it for you here: ruclips.net/video/xVS_amf0DQk/видео.html (warning: crappy screen recording). As that demonstrates, underexposed images can be cleaner than properly exposed images. Or not. There is no true correlation between exposure (meaning, the brightness of the image) and noise when shooting raw. Many of us (myself included) have previously been taught that high ISO does cause high noise, and it's much harder to unlearn a concept than it is to learn it for the first time. Here's a nerdy summary for my more technical audience: First, make sure you understand the difference between causation and correlation. In a car, your speedometer *correlates* with your speed, it does not cause it. Your accelerator *causes* your speed. High ISO *correlates* with high noise. If you're using high ISO with a properly exposed picture, your image is going to be noisy... However, lowering your ISO does not lower the noise, just like forcing down the needle on your speedometer doesn't slow your car. Low light *causes* high noise. Regardless of the ISO, regardless of the exposure, pictures with less light will have more noise than pictures with more light. Therefore, if you want to reduce noise, add light. What I hope to accomplish by demonstrating the difference between correlation and causation is to change how we think about ISO... MANY photographers in low-light situations limit themselves to a low ISO, like ISO 800, because they incorrectly believe that choosing a high ISO will increase their noise. Instead, they get an underexposed picture that has the same amount of noise (but requires additional processing). Hopefully, this lesson teaches those photographers that they only way to reduce noise is to increase the light their sensor is getting. I would also like all photographers interested in technical image quality (noise, sharpness, dynamic range) to be conscious of using giving their sensor as much light as it can handle - and using auto ISO as a measurement is a good way to do this. If you're camera isn't choosing the base ISO for the exposure, then you're not getting your camera's maximum image quality. You can't always add more light, but it's a good to quickly note, "Oh, I'm at ISO 200 - can I use a longer shutter or a wider aperture to get myself down to a proper exposure at ISO 100? Can I add light in some other way?" One stop more light improves image quality about as much as upgrading from APS-C to full-frame.
Great addendum. I am definitely one who wondered if there was a difference in picture quality between under-exposing a picture on purpose to lower the ISO and then bringing the exposure up in post, and just using a higher ISO.
I think you missed some subtleties: Your test relied on using a dual ISO and iso invariant camera, if you had used an iso variant camera the low iso shots would have had much more noise when boosted in post. This obviously doesn't contradict your point.
@@splashstrike In fact in that case high ISO produces less noise than low ISOs. That's why people say "choose the HIGHEST ISO you can without clipping the Highlights" to get the lowest amount of noise
In the linked video you show an under exposed image to a "properly" exposed image. However the under exposed was at ISO 100 and the properly exposed image is at ISO 10,000. You then raise the under exposed image 3 f-stops. If you wanted to compare increasing the exposure to ISO the under exposed ISO image would be approximate ISO 800. Comparing ISO 800 to ISO 10,000 is really not a fair comparison. Yes, shot noise does have a major affect on noise in low light levels, however I don't see you discussing that here.
He can keep on with the semi - nonsense coming out. Like biggest breakthrough in digital is that you can upload to the computer fast. Tony had run out of things to take about
It’s not film…it’s an electronic camera with adjustable gain. This is effectively an amplification circuit which is amplifying the signal coming off the sensor. What you’re talking about is signal to noise ratio. Obviously, if you have plenty of signal (light) you have less noise… less signal, more noise. It’s simple… and nothing like a speedometer. Also proving your point with a dual base ISO camera just confuses the argument totally.
ISO, beyond the basic sensitivity level of your sensor (usually 100-200) is a gain amplification applied to the signal coming from the sensor. If you use it to compensate underexposure, instead of using the real exposure parameters (duration and aperture) it will definitely amplify noise.
Yeah this whole video is basically just a misunderstanding of gain and SNR The noise always exists regardless. When you don't have enough light in the photo (because of a fast shutter/slow aperture) then you have to apply gain to the image (either in camera with ISO, or in post using exposure/brightness adjustment) you amplify the noise Tony gets a lot of things right but he's completely misunderstood the science here - at least in how he explains it. The source of the gain is irrelevant, and the solution is always the same: more light, so that you have to apply less artificial amplification/gain
Tony just blew my mind! In the audio industry we often refer to this as signal to noise ratio. I now realize the sensor noise is static, and that the iso setting is just adding gain. I never realized how much photography has in common with audio engineering until now. Light is a wave just like sound. My camera is a transducer just like a microphone, so all the rules of reflection, absorption, and distance (inverse square) apply.
I've transitioned from audio to photography and can confirm that many concepts are very similar if not the same. I have a lot of issues with this video though. Most of them have been addressed in the comments.
yup and an amplifier won't necessarily distort a clean signal (if the latter is clean at source). If, instead, is distorted such as an electric guitar with an overdrive effect, then even at low volume the sound will be 'dirty', regardless.
Hello, An automation engineer here with some knowledge of signal processing. I don't agree with your explanation: there is always noise, but amplifying the signal to properly expose an image increases its visibility, nothing else. The true exposure is defined by the light transmission of a lens (approximated by the diaphragm) and the exposure time (shutter speed). The camera can also have several basic sensitivity levels: simply take the signal power spectrum of an image and apply a gain to it. You can amplify the signal by using a higher ISO in your camera or in post-production, but the result is very similar with the raw format. But it's usually best to do it as close to the source as possible to avoid adding transmission noise before amplification and to work with the actual analog information and not with the digital information that has already lost some of it due to the quantization process.
Hey Tony, please look at this comment. I was about to comment something similar - you should look into signal processing, S/N ratio, signal amplification/boost/gain. You consider yourself a nerd, maybe doing real research would be nice - observing through experimemtation is fine, but it's no good at all if you have no clue on what is really happening. I have to congratulate you, not everybody can get it wrong (far from the target, btw) twice on the same subject in front of millions of people - maybe on the third? :)
For what it's worth, I don't think quantization significantly affects the noise floor on most cameras as long as you're working with high-bit depth raw files, at least until you get into the 10+ stop gain range. I haven't even run the numbers on my camera though. :)
No no no no He knows this, he is just trying to give everyone a different point of view. I too am educated on this, so I understand and agree on your explanation. But also on the thinking outside the box view of Tony.
Not trying to take it out of context, but let me try to understand your key point. "You can amplify the signal by using a higher ISO in your camera or in post-production, but the result is very similar with the raw format". So you're saying it's very similar, and results appear to be very similar (Tony's point and experiment), but generally it's recommended to amplify the signal (and noise) as close to the source (camera)? From what I saw from the video, I don't think Tony is in opposition to your point. He probably didn't intend for people to take underexposed pictures with ISO 100. Those are experiment to prove his point, and they are done in extremes (rightfully). He's saying to us, that let's not be ISO-centric, because ISO is a imaginary construct. ISO basically is a means for the camera to increase the amplification. Let's be aperture and shutter speed centric, and use them to exposure a picture properly, because that's the actual "hardware" gains you'll get. I don't see contradiction from what he's saying and what you're saying, and you guys appear to agree from my point of view actually (but of course there are others who got very aggro-ed because Tony is saying something that opposes their "mantra")
A few points for people still confused about this: 1. When we say "noise" we generally mean perceptible noise. Grain we can see. All sensor readouts contain noise (the noise floor of the signal). It's only a problem when we can perceive it. At the levels we deal with in photography, it's caused by a few things, but mainly the random quantum behavior of photons, and electronic readout noise. 2. Dynamic range is the difference between the brighest part of the image and the noise floor. When a sensor is underexposed, this difference - the dynamic range - is smaller. The noise is still there, but it's too dark to see. However, a smaller dynamic range means a larger gain is required, and that gain applies to both the parts of the image we want (the signal) *and* the noise. 3. On mirrorless/DSLR cameras, ISO knobs usually apply gain before the camera's image processor gets the data. The exposure slider in post does the same thing, but after. On some cameras, ISO gain happens during the analog stage, which can reduce the effects of read noise, but most cameras *generally* have low read noise. In any case, applying gain with the ISO knob or in post makes very little difference when we're talking a few stops (ie. ISO200->ISO800 or +2eV). 4. The effect of adjusting ISO is different depending on your camera's mode. I think this is the key that has most folks confused. If you use your camera's automatic modes (program mode, aperture priority, shutter speed priority, etc) then raising ISO *will result in more perceptible noise* because your camera will underexpose the sensor. It is this underexposing which results in a lower dynamic range, lower signal:noise ratio, and more perceptible noise after gain is applied. 5. In manual mode, the only thing that affects perceptible noise is dynamic range, and this is determined by your exposure, not your ISO. The noise is always there. It's part of sensing our world. It's only a problem when we can perceive it, and that only happens when we amplify it. We only amplify it because the bright parts of the image aren't bright enough, and this only happens because we underexpose the sensor. The same is true in the audio world: in audio, we don't call it ISO, we just call it gain. When you turn the gain on a microphone up, you hear a hiss. That's noise, amplified. You can speak more quietly (akin to underexposing a camera sensor), but will acheive a higher signal:noise ratio with less hiss when you turn down the gain and speak more loudly (akin to a proper exposure). Lastly - this is not something to get worked up about. These are subtle differences for camera and math nerds. We find them interesting, but at the end of the day, it's the art that counts. If you still don't understand the point Tony's trying to make, then don't worry about it! Composition, storytelling and emotion are far more important than the details of how cameras calculate numbers from a bunch of voltages. :)
@@abdulhasibtamim6905 I think this is actually the point that causes so much contention between people! In auto exposure mode (ie. aperture priority, shutter priority, full auto, etc) the camera adjusts the exposure when you change the ISO. So if you double your ISO, the camera exposes the picture for half as long, or closes down the aperture. This "causes" noise from a user perspective. However, in manual mode, changing the ISO doesn't affect the exposure, because exposure is under full manual control. So, changing the ISO in manual mode doesn't change the level of noise (on ISO-invariant cameras, anyway), because the exposure hasn't changed. This is why there's been some frustration with the "exposure triangle" lately. ISO is not exposure anymore than the gain knob on a microphone determines how loud you're speaking or how far you are from the mic. They're connected, but not in the way a lot of people are taught from the beginning, and it gets extra confusing when we consider automatic exposure.
@@GordLamb Wherever you get the information I think it's not right. I just test with sony a7III, take a test shot with aperture priority and dial the same ISO, Aperture, SS in manual and got the exact same result. With the help of 'difference' layer blending mode in photoshop consolidates this too. So finally I am 100% sure that there is no difference between manual and auto exposure with regard to noise. I read many articles in Internet and watch lots of videos and come to know that there are three fundamental types of noise. (i) Analog noise: which comes from (a)heat noise, (b)shutter shock, (c)photon noise. (ii) Intermediate noise: comes from (a) read noise - basically created during analog to digital signal conversion and also depends on quality of sensor itself (b) analog amplification noise - based on ISO you selected (eg if base native ISO was selected then no amplification, if ISO is 1 stop higher than base ISO then signal will be amplified by 2 hence signal noise will also be amplified as such) (iii) Digital Noise: Comes from any amplification to the digital data (ie further exposure was added in post or in-camera JPEG conversion).
Exposure is determined solely by shutter speed, lens aperture, and scene luminance. Insufficient exposure will make more noise visible. To get less noise, you have to raise the exposure. ISO just maps an exposure to a desired lightness.
Your sensors sensitivity to light is fixed. It cannot be changed in any way. How much light reaches your sensor is determined (camera wise) by aperture and shutter speed. If little light reaches your sensor, the signal to noise ratio goes up as the sensor starts to mis-read the vague light. ISO is a factor the processor/software of your camera applies to the readings of the sensor. High ISO when used to compensate for lack of light will hence amplify the noise that is already there. So the little word "cause" is the debate here. In my mind, high ISO does not cause noise as such but will amplify it if present. So if you shoot with auto ISO on and the ISO value goes ballistic then you know light is missing. I think that is Tony's point. And a good one. So see the ISO as the canary in a coal mine. And don't worry about all the tech stuff.
For all non-technical people: it's the low signal to noise ratio (SNR) that ruins your picture, not the boosted gain that is applied to make the image brighter.
@@reldies5364 it's ultimately the signal to noise ratio that ruins your photo. The ratio never changes after you apply the amplification since the noise also gets amplified together with the signal. If your SNR ratio is 10dB at 64 ISO then it's still 10dB at 6400 ISO.
If you a confused, the point Tony is trying to make is if you are shooting in the dark and already have aperture and shutter speed maxed out, lowering the ISO won’t do you any good in noise reduction. You need to physically get more light in the scene. Thanks for the tip Tony, I remember trying to reduce ISO thinking the darker image will have less noise when I brighten it in post before.
@@-xox- Not only more attractive, but also more necessary because a lot of us (okay maybe just me) think that high ISO introduces noise. So the thinking needs addressing.
Technically high ISO causes the noise because it turns up the gain of the sensor. Of course when turning up the gain, the dynamic range is shifted, you need less light no to overdrive the sensor. Same principle applies for audio gear. There is gain only, no shutter neither aperture. If you turn up the gain it can pick up low level sounds but the noise goes up by also. If you increase the gain but the sound is loud, it will clip, like tony's photo. Of course audio can be more complicated, there can be several stages of amplifiers.
@@dzsemx Thing is it isn't creating noise.. it's only amplifying it. Gain doesn't affect the S:N ratio, only the amount of perceivable noise. The noise itself is created at the time the sensor is exposed and read. Same with audio; turning up the gain doesn't create the hiss, it just makes it loud enough to hear.
Gordon Lamb Under-exposing an image is like having too low an audio signal. You have a fixed level of sensor noise so if you underexpose then the signal to noise ratio is low. If you then amplify the final under-exposed image (ie boost in post) then you amplify the noise too!
0:43 uses dual gain sensor to show high ISO doesn't introduce noise in comparison to a low ISO image corrected in post with the high ISO image clearly superior in the low-mid tones. DOH! An apples to oranges comparison where the sensor is not operating with the same parameters, effectively being a completely different sensor with regards to dynamic range due to the gain applied at the sensor and not after the ADC conversion as is usually the case for these types of comparisons (digital gain). Given that there are many strategies with regards to gain, it cannot be claimed as a universal that ISO settings are effectively noise neutral to the output when your first example clearly demonstrates that this is not the case.
I have to say, that I was very skeptical of your premise, however, I just got done performing the same procedures you showed, and I was shocked to get the same results with my D800!! Thank you and your wife for all the hard work and effort you put into these videos!!
He's testing this on a Sony A7Riv which is ISO invariant above ISO 320 (ie. where the second amplifier kicks in). However, if you choose another camera which doesn't have ISO invariance (eg. Canon 5D Mark III) then the results should be quite different...
I was thinking the same thing at first, but the more I thought about it, it just makes obvious sense. Although, he doesn't explain it very well...as per usual. While all of the example photos were noisy due to the exposure increase in LR, what he really proved is that using a higher ISO in camera provided less noise than increasing the exposure in LR (at least with the camera he used). The confusing thing is that the point he was trying to make was entirely different. He was pointing out that a lack of light is what causes noise - which is rather obvious so we tend to not ever think about it that way. And his solution was to use the auto iso as a gauge to tell if more light is needed...which anyone should be using the viewfinder info, histogram, or just basic photography knowledge to deduce anyways. He was bored and needed a video topic...
Maybe it would be more correct to say that ISO is more like a volume control. You increase the volume control to hear the music better, but all the hum, static, and other noise is increased as well. The volume control is not causing the noise, just making it more apparent.
This is one of Tony's "duh!" videos. Raising ISO to avoid slow shutter speeds or to avoid large apertures essentially is underexposing a photo. So it is the same as taking an underexposed photo into Lightroom and increasing the Exposure slider. It turns out that more expensive wide-aperture lenses and/or Lens stabilization (to allow for a slower shutter speed) and/or tripods all allow for an exposure with the right amount of light. Anything else is using ISO as a cheat that will bite you later down the workflow. From other videos on RUclips, all sensors produce noise every time you take a picture. The more real light you are able to obtain during the exposure overrides that noise more and more (called signal to noise ratio). And yes, larger sensors do indeed have a lower native noise level than smaller sensors, all other things being equal.
This is quite interesting actually. I see a few point of views: 1) Yourself, that feel that this is very intuitive and should be clear to all. Tony is overexplaining. 2) Myself, who I feel that the video is sufficiently explained and I feel like that's a good point, why didn't I think of it. I always thought that there's some magic way ISO amplify signal without increasing noise too much. Seems like I was mostly wrong. 3) Some others, who feel that claiming ISO doesn't change anything?!?! That's the mantra of photography. How dare Tony! He must be mad! And that's cool! Though I feel some comments were more rude than they should be
@@samuelpua9771 I fully agree with you Samuel. I was not trying to be flippant. I love everything Tony reports on and he often injects clarity in something I think I already know but not well enough to take advantage of in certain situations. Thanks for the comment.
You need a combination of large aperture and higher ISO to campture dimly lit moving targets though. Cannot go by on tripods and other forms of stabilisation unless subject movement is within acceptable limits. This high ISO 'cheat' is mandatory at times. Also, the video demonstrates that on dual gain sensors, raising ISO yields better results than correcting exposure in post.
@@tfm2934 That sounds great! Sounds like you're prioritising on large aperture and fast shutter speed (to capture moving targets), thereby thinking of ISO last. Sounds exactly like what Tony recommends in the video. Obviously, the underexposed pictures are part of the experiment to see what's the effect of the 3 items, ISO (digital gains) being the least effect (usually on dual gain sensors, but otherwise minimal effect). You would still be taking properly exposed pictures, just with aperture/shutter-speed centric mindset, which is what you're doing exactly
I think a comparison to a tachometer (measuring RPMs) is more accurate than a speedomoeter. When the RPMs get too high, you change gears, which is effectively what you're doing by adjusting your shutter speed/aperture.
@@smartbiney801 I'd also like to know how to shoot a dress rehearsal of a ballet without adding noise. Sure there's lighting, but it's still fairly low light. You need a fairly fast shutter speed... the lowest I used was 1/160. My photos were good, but the noise was problematic.
So if I uderexpose an image and drag the exposure up in post it'll be noisier than if I exposed it properly in the 1st place? What I've learned was that we need to end this lockdown so Tony can go outside and play!
lmao, thank you, this is indeed the reason why this video is so confusing. He tells it as if it is something new. If you're underexposing you can either turn up your iso or turn up the exposure in post, but both will result in more noise. So don't under expose if you can.... hahaha, yes that make sense, thanks! Edit: Adding to that, images shot with high iso also always will have high noise. Just as Tony says in his video, if you shoot with a too high iso on a bright sunny day, you will still get noise, because of the high iso! Overexposing and than reducing the exposure in post will still result in more noise than if you didn't shoot with the high iso. In conclusion, you can see high iso as a result of underexposing, which means noise. The real lesson is that you shouldn't see iso as a real element of the way of lightning your photos. It as fix that can boos your exposure at the cost of noise. You can also do that in post, but doing it already in the camera may let you use some noise reduction tricks and hardware your camera has build in. So to summarise, use iso as a last resort. if you see that it is high, see if you can fix your exposure in another way. That is however something I've always done, but this video helps with making you think how it works.
Not noisier; the noise was always there... you just couldn't see it until you raised the luminance. Analogy - you're in a messy room with the lights off. Turn the lights up, and suddenly you see the mess. It was always there... it was just too dark to see.
@@GordLamb Yes exactly, the noise is there from the moment you're underexposing. If you than boost everything to correct the exposure, you're also boosting the noise and therefore you can see it (more clearly).
@@Zegmaar_Bas Right, though technically the noise itself is always there, even in a correctly exposed image. It's just not perceivable, until like you say, you amplify it along with the rest of the image (which obviously you wouldn't do if it wasn't underexposed).
@@Zegmaar_Bas In other words always shoot with the lowest ISO you can muster. Maybe use a tripod at night instead of hand held high ISO shots. Then again, people think shooting with big apertures is a good trick at night to allow more light to enter the camera, but you have depth of field considerations as well. So shoot with the biggest aperture you can get away with and the slowest shutter you can get away with and the lowest ISO you can get away with before cranking ISO.
If I understand correctly, the takeaways are: 1. Underexposing causes noise; a properly exposed image will always have less noise (regardless of ISO) than an underexposed image pushed in post, even for ISO-less sensors 2. Among properly exposed images, the one with the highest ISO setting will be the noisiest
1. Underexposing doesn't cause noise if you keep the aperture, light and shutter speed the same... the noise is pretty much the same whether you properly expose it or adjust it in post. 2. Yes, high ISO correlates with noise, but does not cause noise. That's why I suggest using auto ISO as a speedometer, so you'll have some sense for how much noise you'll get in the final image. If you want to reduce the noise, you need to look at light, shutter speed, or aperture, because those have a causal relationship with noise. ISO does not have a causal relationship with noise.
@@TonyAndChelsea You still have it backward. Shutter speed/aperture doesn't cause noise. Those are the physical paths of the light. The ISO itself is the ONLY part that actually introduces noise. The sensor doesn't care about the aperture/shutter speed, those are separate physical phenomena, the sensor itself, translating the energy of the photons into digital bits, is where noise is introduced. "noise" is just another way of saying "uncertainty of measurement". The aperture and shutter do not directly effect the measuring process. The only setting that actually changes that process, is the ISO setting. The concept of using auto-ISO as a speed limit is quite sound. And what many of us already do. That makes perfect sense and is a useful message to share. But please don't double down on the incorrect parts of your video in order to justify the useful/correct parts. That's just ego.
OK, that's why the first image that was underexposed via ISO was the cleanest of all boosted images. It wasn't exactly as good as the high ISO image due to the dual gain (does pushing 6 stops in post border on the limits of ISO invariance though or is it unrelated?). Taking it even further, for ISO-less sensors at least, one could say there is no longer an exposure triangle, as those cameras effectively have a single ISO and exposure is only determined by aperture, shutter speed and available light.
@@TonyAndChelsea Are you sure the noise is about the same when adjusting in post vs. using higher ISO? I was under the impression a lot of cameras actually apply analog gain depending on ISO. At leas with my old 50D it makes a huge difference, noise is brutal if I adjust it in post instead of using a higher ISO in-camera.
I think the takeaway here should have been a properly exposed image reduces noise. In a dark environment, If you have the ability to add more light (ie. a constant light source or flash) then you will reduce noise, if you have the ability to lower your shutter by using a tripod thereby exposing the sensor to more light you will reduce noise, if you have a faster prime lens and you shoot wide open thus exposing the sensor to more light then you will reduce noise. I think this has to do with the signal to noise ratio no? Simply bumping the ISO will not in and of itself increase the amount of light available - thus more noise introduction. It's a fascinating topic though. Thanks for taking this one on Tony. Always one to face the controversy head-on - lol ;)
ISO is a signal amplifier. The signal in photography is light. Less signal = more noise. If you boost the exposure of a scene, either when taking the photo or during post-production, there will be more visible noise because of the lack of light. The information lacking in the picture because of not enough light is filled by random pixels of different colors by the digital cameras.
Anybody practicing "exposing to the right" knows that sensor saturation is paramount. If a high ISO helps/is required, OK, then use it. In my experience, using a lower ISO with proper exposure to the right results in less noise.
@David You can always shoot with the histogram enabled and avoid clipping in real time before you shoot. I own a basic DLSR (SL3) and do that with live view. With mirrorless cameras you can do that in the viewfinder. But I agree that avoiding clipping must be No 1 priority.
If you lower the image information level into the noise floor and then amplify both the image information and the noise in an attempt to recover the image, the image noise will be more apparent. The ISO setting amplifies the signal arriving on the sensor and by default the sensor’s noise, therefore the noise is a higher proportion of the signal if there is less signal. Amplifying the signal amplifies the noise. High ISO is used because there is insufficient exposure of the sensor to light reflecting from the subject, raising ISO is used to amplify what is being received. Otherwise, we just accept that we run out of road and use the widest aperture and longest time necessary to decrease the proportion of our signal that is noise. The noise is inherent to the sensor and the amount of signal amplification applied by the ISO setting, not the aperture or the shutter speed, those are independent of the sensor and its noise floor. In any shot overexposing the sensor pulls the signal further above the noise floor. By recovering the image in post, because noise floor was already very low you reduced its effects by reducing image brightness to get the appearance of a correct exposure. Going the other way round and underexposing the sensor pushed the signal into the noise floor so changing the image brightness in post to compensate reveals the noise the image is imbedded in. The processing engine of a camera already does a lot of heavy lifting to fix the balance between noise and image. Adjusting the iso partially digitally processes noise effects away after amplifying the signal. Newer cameras tend to cope with under exposure better than older ones because noise processing may be better.
If you're going for a meaningful analogy, this is exactly the opposite: It's the low light hitting the sensor that makes people use high ISO and get noise. So the low light is the cause and the high ISO is the effect. Just like jumping off the roof causes you to hit the ground.
It's even worse because he says the shutter speed and aperture cause noise, but ISO does not. It's like saying jumping off a cliff or out of a plane will kill you, but when you jump from a building, it's the sudden impact that does it
Is there PROOF that jumping off a building doesn't kill? What if a person gets a heart attack BECAUSE he is fearful at the start of the fall? Just wondering. :-)
Tony is actually right that high ISO isn't the demon. I found this out by mistake one time - eons ago when I accidentally kept one of the very older (Canon 60D) cameras at 1600 ISO when moving from a darker room to a brighter room. I changed the shutter speed but not the ISO and the fact that the sunlit room was awash in light, I was amazed at how clean the capture looked even at 1600 ISO - bear in mind this was back when 1600 ISO was pretty much a max limit for digital captures. It was the first time I realized that a bit of overexposure at higher ISOs was a GOOD thing - and that the real issue was combining underexposure (lack of light) with high ISO was the culprit. It's better to overexpose a bit with high ISO images and absolutely damning to not have enough light and shoot high ISO.
When i shoot the milky way with 1600 iso, the image has more noise then when i shoot outdoor with sunlight using iso2000. The missing of light cause the noise
I don't think that's a useful comparison. You would need to compare the sun-lit image at ISO 100 with an equivalent scene shot at ISO 2000 and 20x the shutter speed.
Studying astronomy has made understanding photography so much easier. We work with CCDs and I'm responsible for checking the gain of the detector, which we use a monitoring script for. The gain is basically how sensitive the detector is to photons, and how much it can convert photons into electrons. The unit for gain (in astronomy) is electrons-per-DN where DN is digital number or the photon number. So if your gain is 2 e-/DN then for every photon that hits your sensor/detector/CCD you get 2 electrons of measurement. So with ISO if you crank it up from 400 to 800, you're effectively going from let's say a gain of 1 e-/DN to 2 e-/DN. Same for 800 to 1600 and so on. The amount of noise in a single pixel follows the poisson noise (shot noise) and is equal to square root of N, where N is the number of photons hit. If you don't have enough photons in an image (aka its severely underexposed), your shot noise relative to your photon count is going to be sizeable. If you have plenty of photons in an image (appropriately exposed) your shot noise is going to be minimized. So if you had 1 photon hit a pixel, your shot noise is 1. Whereas if you had 100 photons hit a pixel, your shot noise is 10. 10/100 is less than 1/1. So basically if you appropriately expose your image given the gain of your detector (ISO), you can minimize your shot noise. This is basically what Tony demonstrated. We have calculators set up to optimize the exposure for scientists who use our facilities, we call them exposure time calculators or ETCs. It's like the instant readout you get on your camera for your shot, but in this case scientists put in the object they want to look at in space plus what filter they want to observe in (infrared, ultraviolet, blue/green etc.) and it does a series of calculations to tell you what the exposure time should be, what the detector should be pre-flashed with light to, and what the final estimated brightness will be.
@SwitchRich For a given (correct) exposure, if all settings are locked and you increase the f-stop, you will lower your dynamic range / signal:noise ratio. This is not an opinion, but rather a fundamental property of math and physics. You can demonstrate it with a camera in a carefully controlled experiment, or you can run the math with a whiteboard. (apologies if you're being sarcastic, but on this topic ya never know :)
I think I get it, but let me try to restate it in simple language for us simple old people: If I am taking pix of my lovely spring flowers, and I notice that I need to crank up my ISO to get "proper" exposure, what my camera is REALLY telling me is that my fstop is not low enough, or that my shutter speed is too high. And if I reply to my camera, "But camera, I need my aperture closed down this far for proper depth of field, and I need a higher shutter speed to stop motion blur because the wind is blowing," then the camera will reply, "Then shine some light on the flower, or wait until the wind calms down, and for heaven's sake use your tripod, because there's not enough light now to keep the noise down!!" Am I close?
It all comes down to the signal to noise ratio. If you cut down the light to the sensor, the ratio between your image and the noise floor decreases. You recover your image either through amplifying the signal+noise either in the camera (higher ISO) or in post processing. Same difference. So noise is caused by lowering the level of the signal (light) to the sensor. EE's point of view.
Mr. Northrop you have saved me, I have a Canon 90D and Sigma Art 18-35 F/1.8, my pictures always show noise even at base ISO, I was ready to quit and now I have watched and tested what you have said I have got rid of the noise problem. A massive thank you.
The process that you make in the computer when lifting the exposure is very different compared to shooting in a higher ISO (2 different stages in the image making/processing- Pre Gain VS After Gain). This whole video is extremely misleading. It only demonstrates the fact that you should get your exposure right and don't count on lifting it back in post. The claim that there is no correlation between High ISO to digital noise is ridiculous. The only thing that i can agree with is the statement at the end about using higher ISO as a last resort .
The signal and noise are both fixed in the raw file at the time of capture. Post-processing does not add noise. "The claim that there is no correlation between High ISO to digital noise is ridiculous."
Modern Sony sensors are ISO invariant which means that shooting at for example ISO 3200 and shooting at ISO 800 but raising the exposure two stops will look exactly the same. But I agree his conclusion was a bit extreme, it doesn't work that way on older digital cameras.
@@TonyAndChelsea I understand the point you are saying that the lack of light is the factor that creates noise. What i don't agree with is saying that high ISO doesn't cause noise, because it does. This due to an electronic enhancement of the signal right after it is being captured by the sensor. There is no getting around it. High ISO generates more noise if you shoot the same exposure right out of the camera. The comparison you made while trying to bring up the exposure in post is incorrect.
This is somehow true when you are talking about Canon's cameras that do not have "ISOless" sensors, try that on a Sony camera and you will see that there is no such thing as Pre-Grain VS After Grain. ISO is just a number that indicates the amount of signal amplification that will be implemented in the final image. Light, exposure, and sensor performance will determine the amount of noise on your image, ISO itself does not have any effect on image noise as long as you are able to give your sensor a good amount of light exposure. I am sick and tired of Video Low light "tests" that compare noise performance at different ISO speeds while keeping the exposure and the light the same. These are the dumbest tests anyone can ever make, yet everyone just keeps doing them just to prove Einstein right when he was talking about human stupidity and the Universe.
You are wrong on so many fronts! ISO is a gain control, it takes what you capture and amplifies it. If you have noise mixed with the light, SNR, then you will amplify both, if there is plenty of Signal (light), but very little noise in the capture, let's say, 90% signal and 10% noise, then any gain you apply, will increase both, but 2X 90% is far better than 2X 10%. In low light situations where the SNR is more in favour of noise, 60/40 split, using the gain will result in 2X 60% noise as opposed to 2X 40% signal. You don't need to be a mathematician to realise the bigger percentage will always win out. Electronic components also create their own noise, which adds to the total noise when engaging high ISO settings. The worst mistake you made, was using post production to try to compensate for shutter and aperture settings, as post production can add quite a lot, if not excessive noise to the already bad image. The answer is quite simple, you get the best SNR, ( light), from the composition you can, before you apply the ISO gain. In other words, in low light situations, drop your shutter speed, or pick a bigger aperture before you resort to picking a higher ISO. If you don't, and your capture has a high percentage noise level, upping your ISO will only boost the noise level and make the image worse. The more light that hits your sensor, the less noise there will be!
I've never seen Tony respond to so many comments :) He's right. Test this with your own gear (without blowing out the highlights) before disputing it. I was surprised with my testing!
Just did. And this he isn't correct. He also knows it, which is doubly annyoing. The concept of "ISO as a speed limit" is pretty clever and correct. However, do it yourself: compare at the same aperture, ISO 100 @ 1/200 to ISO 400 @ 1/800. Same exposure. Tell me which has more noise. 🤔
@@shammyh You changed your shutter speed to test. You didn't prove anything. You have to match exposure in post to test this theory. Try keeping lighting, apeture, and shutter the same, take a pic at iso 100 and one at iso 400. Adjust the exposure in post 2/3 stop or so to make them the same exposure. Then look at the noise.
Sometimes Tony is the perfect image of that guy who butts into a conversation you're having with "ACKSHUALLY" and goes off, and you just roll your eyes.
probably you are saying it later in the vid (hope so) But it's the lack of light, that causes noise (i think). It's the signal to noise ratio, what's important here. and with higher iso, they crank up the voltage of the sensor, therefore your noise ratio is higher. So i guess i am disagreeing with you here. Higher ISO does cause more noise, but the main culprit in the story is the lack of light. And because higher ISO's are being used at lower light environments, it's easy to point your finger at the ISO, it's just a little bit more complicated then that.
I normally think Tony gets stuff right. But Flawed testing here. The image noise primarily exists in the shadows. So if you underexpose shots. You have more noise. Regardless of how you end up being underexposed. Shutter, aperture, light or ISO. If you have the same decent exposure, but THEN increase ISO. You will get more noise. That doesn’t happen for shutter or light or aperture. If you have the same decent exposure but vary shutter, light and aperture for each. And keep ISO the same, Then you will have consistent noise level.
He clearly demonstrated the opposite in the video, in a way that you can easily reproduce at home. Keeping ISO the same but varying SS, aperture and light change the noise level, as they change the amount of light.
Increasing ISO does induce noise if used to correct exposure...If you increase ISO (or raise brightness in post) to correct the exposure 'when' there is insufficient light on the sensor you WILL increase the noise. In other words, if you use ISO (gain) to correct for underexposed, you will introduce noise in order to gain the correct exposure.
Great explanation Tony. I have always been an available light shooter and tended to towards wide apertures on expensive fast primes. I recently started playing around with some slower but modern Leica/Voigtlander/Zeiss M lenses adapted to my Fujifilm APS-C camera. I realized what I really enjoyed was clean images with sharpness and tone more than bokeh; shooting a scene at f5.6 or tighter gave me what I wanted more than at f1.4. The consequence was I just started setting my ISO to the most tolerable setting for that time of day and shooting with a mini tripod. Instead of wasting money on glass I started to use the creative restrictions of shooting with a $30 mini-tripod as my guide to setting up shots. I was always afraid to "waste" the expensive glass on more moderate f-stops but I knew I didn't want blurry photos; by admitting what I wanted I sold off some expensive glass and just focused on what was sharp as well as a few cool stabilization techniques. The noise (or signal to noise ratio) is always about available light and separating it from ISO is important in thinking about exposure; ISO is gain to boost signal (and noise) and the more likely you need to add gain the more likely you need more light in your shot for your preference in noise level. Like you, I always viewed the % of my maximum allowable ISO for the setting as a measure of my success in curating the shot. Excuse the long comment, hopefully someone finds that valuable.
What you're saying is that high ISO boosts noise (and signal) which Tony says doesn't occur. He specifically says in his opening statement, 'High ISO does not cause high noise' - so which is it to be believed, you or Tony? This theory of Tony's is fine for the studio, but has no basis in real life outdoor photography where the light conditions can only be altered by either a) moving the subject into a better lit spot, but if that isn't possible then b) by providing extra light in the way of flash or reflectors. Yeh, try that on a race track! Tony should have been more specific in what he was outlining because in circumstances that require fast shutter speeds and the widest aperture your lens is capable of in lighting that necessitates a higher ISO, then you will get more noise - fact! Lowering the shutter speed will of course lower the ISO for the same lighting conditions and subsequently lower the noise, but will increase blur (and I know which I would prefer) Alternatively, if the aperture is reduced in order to preserve the shutter speed and the ISO left as is (lower than required for the A/L, ) underexposure will occur and on none ISO variant cameras (which I imagine the majority of us have) then noise will be increased in post to the same extent as it would have been had the ISO been increased with the same aperture at the taking stage. So where's the advantage in underexposing and increasing the brightness in post? None from what I can see.
@@robinlikes2learn what is 100×0? Zero. What is 100×5? 500. ISO doesn’t multiply noise that isn’t there; it doesn’t cause noise, it amplifies it. If you increase light and thus increase signal to noise sufficiently there is negligible noise to be amplified.
thanks, this is really thought provoking. Just want to add something here, noisy image is caused by not high enough Signal-to-noise ratio, meaning the sensor cannot collect enough "good" light signal to combat the noise naturally occurring in the circuit of the sensor. So high ISO is really not to be blamed, it's by increasing the ISO, you are allowed to use a faster shutter, which in turns reduces the amount of "good" light that hit the sensor.
At 1:13, increasing exposure gain on pic 1 to match with pic 2 on post. As far as I know, in/decreasing exposure gain on post depends on ISO base of the picture. In this case, for example, if you gain the exposure of pic 1 +3 stop (idk how much exactly exp. gain on pic 1 to match pic 2 exp., so I assume +3 stop) from base ISO 6400, it would be similar to ISO 51200 on exposure. Noise should be worst than before, right? Lets say, you take pic 3 with 1/640s f2.8 ISO51200 , it would be nearly same as pic 1 after exp. gained. Pic 2 will be much more clearer with long shutter speed as compensation.
For non ISO-invariant camera it's actually the other way around. Higher ISO means lower noise since the signal are amplified before the read-out noise been added in, so higher ISO means better signal-to-noise ratio and lower noise.
good point, we can understand the video like this: increasing ISO = increasing exposure in post editing, they're same technical light boosting which causes more noise, but the ISO in the camera might work a little better in some way, that's why (ISO100 + increasing exposure in post editing) looks worse than High ISO
That's his point. :) Neither gain applied on the camera using the ISO knob, nor gain applied in post create noise; they merely amplify the noise that existed at the time of capture. Dynamic range is determined by exposure levels, not gain, and captured dynamic range dictates the required final amplification. If you want to reduce noise, you need to capture a greater dynamic range, and you do that by correctly exposing the sensor, not applying gain with the ISO knob.
Gordon Lamb oh yeah, push the gain on your camera and correct exposure by using ND filters, aperture and shutter speed and tell me how the image looks.
@@ralphgeronilla so the whole point here is that gain/ISO has nothing to do with exposure. It's not part of exposure. ND filters, scene luminance, shutter speed, and aperture are really the only variables. So long as those variables are configured to maximize dynamic range of the sensor, you'll end up with the best possible result.
I have an older Canon 50D. At ISO 400 is is noiceless with a well exposed image. At ISO 1600 it's very noisy with the same well exposed image. Same lens, same camera, same aperture (f2.8) and the shutterspeed is two stops higher. I've always thought of ISO in digital cameras along the lines as radio reception. At the base ISO that the camera has - and it has only one - the noise is excellent. A radio picks up sound from a local station and it sounds good. If we want to listen to a radio station whose signal is weak because it's a distance away, we turn up the sound and hear the broadcast more clearly but we also hear popping and buzzing as the background signal noise. When we raise the ISO in the camera we are telling the camera that we will accept the weaker signal (less light) and through camera noise reduction and RAW noise reduction algorithms we reduce the noise to make it acceptable.
I just took 2 (identical) shots of a house ... both in aperture priority at f/13.... both with correct exposure... One was shot at iso 200 and one shot at iso 3200.... not surprisingly- one yielded 1/200th sec shutter speed- the other 1/3200th... No prizes for guessing which image yielded the most noise... I'll leave you to decide if it was the high iso or the 1/3200th sec shutter speed that was the cause...
The noise is always there. The same noise is in every photo here. When more light hits the sensor it overrides the noise. So Tony's right, more light (signal-to-noise ratio) less noise seen in image. Look through your viewfinder with the lens cap on. There's your noise.
I found this out myself by trying to limit my camera to low iso which meant I was underexposing and got noisy images so now I let the camera choose iso and keep checking the histogram for correct exposure and yes using the auto iso as a “speedometer” is a handy tip .
I found that out too. I posted it above. If I let the camera find the correct ISO, then expose to the right based on my histogram, not only did the noise subside, but the color was richer. The noise was still prevelent, but way more manageable then exposing correctly.
So I came a little late to this, but it's as simple as to think that raising ISO is just like raising your exposure bar in lightroom. So it creates noise by retrieving the details in the shadows and mid tones. So technically it is not the ISO what creates noise, actually noise isnt created, its always there, it's just a matter of how visible it will be. And it will be more visible when there is less light, when there is more light, the light overlaps the noise (SNR).
The noise on your camera's sensor is actually pretty static. No matter what your settings are, the noise stays about the same. ISO increases the brightness in the same way that turning up the gain on a mic increases the volume. Any hiss on the mic is also amplified when you do that... just like the noise on your sensor will be amplified when you do that. Ditto with increasing the brightness in post. The goal here is to increase the signal to noise ratio. In this case, the "signal" is the light you're gathering. Thus, if you want to "reduce noise", you simply gather more light by decreasing shutter speed or f-stop. The extra light helps drown out the noise.
@@Falcrist sure but on the real world you use the iso to compensate for the lack of light... so what the hell are we talking about? you should always have a correct exposure... if you recover the dark part of the image on post it's obvious that you are going to have tons of noise in the picture...
@@aizel330stream Honestly, I'm not sure what the point of the video is. High ISO does indirectly "cause more noise" in the final image by making you increase f-stop and shutter speed to compensate. That reduces the amount of light gathered and thus the SNR of the image is lower. But if he wants to be pedantic, he's doing it wrong. Even the high shutter speed and f-stop doesn't cause the noise. The noise was already there, you're just not gathering enough light to drown it out.
@@aizel330stream He's not trolling lol There's *some* value to understanding what I explained... Not much, but some. I just think this video is a mediocre explanation.
in short: noise is a function of total light gathered which is affected by aperture (flow) and shutter (time). ISO is a gain applied to the signal post-capture.
Great information. Thank you for sharing this as has helped changed my perspective. I am attempting to apply this insight to getting better drone footage at night. I have been having trouble getting the right settings to reduce noise while still getting enough brightness in the key areas of the image. The aperture is fixed so its been coming down to adjustments in the shutter speed and ISO. I have also found that that for the Anafi PLOG produces the better footage. And that HDR adds a ton of noise. Do you have any recommended settings for capturing night video via Drone?
Since high ISO do not cause noise Tony , please show us a well-exposed image at very high ISO that do not contain noise. (It's sad to write and say things that are NOT true! Very sad...)
I proved this to myself last night. We frequently have whitetail deer in our back yard. In the last month I upgraded my Canon Rebel T6 to a 90D, and upgraded my glass from the 75-300mm kit lens to a Sigma 150-600 Contemporary. I live on the west side of a hill, so light fades in my back yard well before sunset. I first tried photographing the deer at 600mm F6.3 (wide-open for this lens), 1/1000, and auto ISO (25,600). I got nice sharp images, but there was a lot of noise. Taking the advice of this video, and another of Tony's videos (on finding out how low you can get your shutter speed hand-held), I moved the shutter speed down to 1/250. The image was underexposed, but when I looked closely there was little to no noise in the image. Thanks, Tony!
Why would you want to use high ISO outside with plenty of light? You'd have to crank your f-stop number up and increase your shutter speed, thus creating more noise.
@@Joe-hm1zk Increasing your ISO actually affects how many stops of dynamic range you have above middle gray, so there are some situations where it can be beneficial. For example, if you're taking a photo outside where you care more about the clouds than the ground, increasing your ISO will capture more dynamic range in the clouds at the expense of the ground.
ISO doesn't cause noise. Fast shutter speed and high f-stop doesn't cause noise. The noise on a given sensor is always about the same. The goal is to drown the noise out by collecting more light. That means opening up the aperture and/or slowing the shutter speed in order to gather more light. The light you're gathering is the "signal", so you're increasing your signal to noise ratio. Increasing the ISO does the same thing that turning up the gain on a mic will do: increase the hiss you get from that mic. It would be better to reduce the gain and move the mic closer to the subject (if possible).
@@osirismarbles5177 I think it comes down to the fact you have 256 levels of brightness. Looking at the ground and sky scene, to have shadows be above brightness 0, and white clouds below 255, all your shadow area's might only be between 0-10, and all the clouds 245-255. But if you don't mind all your sky being over exposed and 255, the shadow area's can now be anywhere from 0-150 brightness, giving you a lot more subtle details. And vice versa with the sky.
While under exposing and correcting in post certainly reveals more noise, it’s also true that the sensor’s signal to noise ratio is more favorable at lower ISO. I think he has a point about less light causing more noise, but I don’t think it’s true to say that noise is not at all related to ISO.
Yeah this seems a little misconstrued. Using iso to boost your exposure in low light does increase noise esp in shadows... I don’t understand the point he’s trying to make. Like yeah it’s technically the lack of light that makes it noisy, but cleaner gain sensors don’t look as noisy in the shadows even when both are at native.
The logic of your argument is not fully correct here. Noise is intrinsic to the light itself, but the more light (exposure) you get, the less noise. So, the question is what gives more exposure. Low f-number and long exposure time (low shutter speed)! But often the problem is that the shutter speed is limited because of motion blur, and the f-number is limited because of the focus depth that we want. Now, to have less noise always the lowest ISO is the best. Because the higher the ISO, the faster we get to saturation level, thus we should end our exposure. This is unless for any of those limiting factors we can not fully expose our full dynamic range (exposing to the right of the histogram). ONLY THEN, we can increase the ISO to get actually less noise, and that is because it will reduce the part of the readout noise that lies after the ISO amplification circuitry. ISO amplification is just a like a multiplier, it does not increase the real exposure.
I actually just figured this out yesterday doing macro photography. It was a very cloudy and overcast day. I went outside and I was using a flash with a paper plate setup on my Canon 100mm f2.8 lens.. I was shooting at 1/200s of a second f16 with ISO 6400. Because I was using the flash to introduce extra light. When I checked my photos in Lightroom there was no extra noise. Originally I was upset I had to crank up my ISO to 6400 because I knew that the amount of noise being introduced with the aggravating. Or at least I thought so. But the end result came out phenomenal. After doing a Google search of this I could not find a definitive answer until your video Tony. Thank you again for the wonderful upload and I hope you and Chelsea have a blessed and wonderful week.
This isn't a very productive video, for beginners watching this it will confuse and mislead them. I understand that this video is intended for pros and makes for a very clickbaitey title but I think overall it will confuse people who dont know any better. Its also very misleading since Tony uses 6400 ISO (A higher ISO) on all of the pictures and specifically chooses the under exposed images to bring back in post which will always introduce noise into the image. Bringing back exposure in post introduces noise and Tony knows this, you might know this, I know this but someone who is just beginning might now know this.
So it's similar to gain, and noise caused by electronic (digital) amplification of sensor when there's not enough light (Sorry but speedometer comparison was a bit confusing to me)
Electronics inside the camera like the sensor and ADC this is what makes noise. The noise is just a constant and by making shorter exposures it will look like we have more noise (because the noise is the same all the time, though it will be a bit bigger in higher temperatures).
The “noise” Tony is referring to is his own voice. The solution is to turn off this RUclips video and watch someone who knows what they’re talking about. Hope that helps!!
Dead wrong as usual, Silver Fox. You’re raising the exposure of the ISO100 shot so much to get it to see details that are in your ISO6400 shot that it’s causing noise. You can’t push that far in post without getting noise in the shadows. That’s why you’re getting more noise in the ISO100 shot. And btw, the 6400 shot also has a lot of noise - because it’s taken at a high ISO. It looks better to you because the ISO100 shot that you DESTROYED in post by pushing it so far in the shadows has degraded. ISO is, has always been, and will always be applied gain in a DSLR or mirrorless camera. It’s applied AFTER the shot is captured to artificially apply more light to the shot. Anything done artificially in the camera (and even in your destruction of the ISO100 shot in post) to compensate for the light that should have been in the shot to begin with will degrade the image and add noise. Not opinion, it’s a fact. Maybe you are ignorant - nothing wrong with that. But you should (1) learn how a camera really works; and (2) apologize on a future video to all those you are trying to learn the same thing you should already know by now. We’re not perfect, but ou need to admit when you’ve made a mistake.
He does make some sense, I don't think his statements can be entirely dismissed. Think of bird or sports photography in daylight. We shoot with high ISO so we can get the shutter speed as low as possible. An ISO of 6400 in sunlight will definitely not show as much noise as 6400 in low light. So it is certainly acting as a speedometer in this case, directly correlating to faster shutter speed.
Spot on. Low iso's don't prevent noise in my milky way shots if I fail to gather enough light. I often need 3 or 4 minute long exposures of my foreground to prevent high noise when i bring the shadows up. Gathering light is like catching rain in a bucket; if you don't have enough you're screwed regardless of how good your sensor is.
Tony, As pro's, we are taught that "shadows create detail". That's fine in theory, but not always applicable in real world situations. For example, in my real estate photography business, I shoot a majority hand-held at a higher ISO's than I would like - this is dictated by pricing, and number of homes I shoot daily. I have found with my EOS R and 16-35L f4 IS, a comfortable ISO to be around 800. This means I am shooting a standard room 1/60 @ f8 with flash at 1/4 power (angled 45 degrees back over my head) . I am able to add enough fill to not create bounce shadows, while retaining detail shadows and not blowing out highlights (windows).
Interesting manipulation of information...not real, but nice. ISO works like a booster for low light environment...it boosts everything....also the noise is provided by parts of sensor that are not excited by photons, so increase of ISO is the increase of sensitivity of this parts... A good sensor can have reduced noise as the software can analyze and eliminate part of noise (wrong signals from photodiodes) by comparing pairs of them and reducing fake electric signals. This is why so called "night shots with reduced noise" have smaller size, as the matrix of photodiodes will be used partially for canceling the noise induced.
when you adjust the exposure in post(shadows highlight ....), you virtually increase your iso, noise is only dependant of the iso value (in camera + post) when you have a short stutter speed, you pull the post processing iso up, so more iso and more noise technically in low light it's better to stay at the lowest iso where your camera is iso invariant, 400 or 800 iso most of the time, so you won't clip the highlight (in a concert for exemple) (also be carefull with the new mirrorless and banding effect)
You say "noise is only dependant of the ISO value (in camera + post)" and "dependant" is an interesting word. Noise correlates with the in-camera ISO value, but it's not a causal relationship. Adjusting exposure in post does not cause noise at all; all the noise is in the raw file at the time of capture. "technically in low light it's better to stay at the lowest iso where your camera is iso invariant, 400 or 800 iso most of the time, so you won't clip the highlight (in a concert for exemple)"
@@TonyAndChelsea "If you move into a brighter scene (maybe they increase the room lighting) your camera will not be able to drop to a lower ISO" then you need to adjust. Most of the time I'm in auto iso, or base iso. Some rare case, I lock at 800 iso cause I don't want my camera to use 400 iso or 640 iso. PS when I say noise I only care about final image for the rest, It's just a matter of phrasing and yes the more light your sensor get, the less noise you image will get.
I'm really happy you posted this vid, I was starting to watch some of your material. I'm glad you stopped that quickly by revealing your level of competence.
Yeah, digital ISO is really "gain" which is what the exposure slider does. Typically, camera processors are more efficient than what your computer can do in post. The real takeaway from this is to shoot your image perfectly exposed or as bright as you can without clipping highlights (ETTR) then adjust it down in post. It is always better to shoot with a higher ISO to get the exposure where it needs to be than it is to underexpose and lift in post.
Christopher Goetting They aren‘t more efficient actually, it‘s only that the ISO control boosts the actual voltages before the Analog/Digital Conversion. As the picture on the Computer is digital, it has to boost the digital signal, which was already processed, even in RAW.
@@TonyAndChelsea I feel like this way of framing the problem is overly complicated. When one brightens an image in post, that's really just an amplification of signal and noise, increasing their respective visibility to the human eye. Technically the signal to noise ratio doesn't change, so it's correct to say that the noise was always there. But brightening in post processing (or raising ISO in camera) is what makes the noise visible. Therefore, I'd argue, a more intuitive way to frame this is to say that increasing the ISO or brightening in post increases the *visible* noise of an image. [To clarify: This is about this particular comment thread. The overall point of the video seems valuable nonetheless.]
Easier said than done when shooting wildlife with telephoto lenses and the need to use fast shutter speeds to prevent motion blur! The light outdoors at any given place on any given day is fixed. The beauty of larger sensors is that you can dial up the ISO more and still see less noise. ISO 6400 on my Panasonic G85 is nearly unusable. ISO 6400 on a Sony a7s II is VERY usable. And that can be crucial in a situation where you don't have much control over the light or as much flexibility to reduce shutter speeds. I have found sports and wildlife photography can really push a camera to its limits like no other type of photography.
Less light doesn't cause noise, it causes a underexposed image or a low key image... however noise is produced when you force your sensor by increasing iso to be more receptive of the little light is there. If you had a 4 stops underexposed image and gain light by each of the settings the result would be very different. Shutter speed would cause movement, aperture would cause shallow depth of field and ISO WOULD CAUSE NOISE.
@@eduardobianchi4564 No, watch another video from Tony. About ISO. If you correct underexposed RAW by 4 stops in post it's the same as raising ISO in camera before taking picture. Do You think this channel has millions of subscribers and don't know how exposure work?
Simple setup: take images with the lens cap on. Change ISO. Change Shutter speed. Change Aperture. Do the same for each setting of ISO. Look in Lightroom. Do not adjust exposure. Look at the difference. Then increase the exposure for EVERY image to the SAME increase of exposure (+5). Look at the difference. On my Canon 5D MK IV, at ISO 100, the images are black and stay black, regardless of shutter speed or aperture. At ISO 32000, the noise from the sensor can be seen in the image before exposure adjustment and is accentuated drastically when exposure is increased. This means there is, at least for Canon, a direct correlation to increased noise at higher ISO settings. (ISO100,1/60,F1.8;ISO16000,1/60,F1.8;ISO32000,1/60,F:1.8;ISO100,30",F16;etc.etc.etc.) The point about low light is valid, but to say that ISO doesn't produce noise, at least for Canon, is pretty easy to show is not correct by doing the lens cap test.
Great video. But to nitpick, f/ stop and shutter speed and lack of light don’t *cause* noise. The noise is always in the sensor. What f/ stop and shutter speed and light affect is the signal to noise ratio (SNR). When you float the ISO and it reaches high levels, it’s an indicator that SNR is dropping.
The ISO in this cases is caused by the shadow lifting in your editing. I think the correct way to compare in this case would be having the on image correctly exposed, and the other one over exposed, but correcting it with an ND so both would be at correct exposure. That would be a fair comparison
Why? The result would be identical. First decreasing the light by using an ND filter, then restoring it by lowering your shutter speed. The same amount of light hits the sensor. Are you implying that the duration of the exposure alone is a determining factor in noise? Interesting though, will test it tomorrow :)
TFM no. I’m saying that because he adjusted the dark image to be brighter, and he should have done the opposite. Adjusting the brighter image to be darker. Because it’s obvious that by pushing the shadows it will come noise. So I just mentioned ND’s so you are to get lower speeds without clipping and able to recover the highlights just that
@@GuillePozzi correcting with an ND reduces incoming light which is not the same as darkening the image in post. You'd reduce incoming light therefore increase noise.
@@tfm2934 I do understand, but what I'm saying is that he is matching exposure in post. Which will obviously cause noise when increasing the brightness/exposure. So what he needs to do to correctly test this issue, is by matching exposure IN CAMERA, and the only way to do that (without altering other parameters) is by using an ND. And I do understand what you said, but I've done this test before (properly, and no, ND does not increase noise, and yes ISO causes noise).
@@GuillePozzi I think we agree with eachother :) But I'd expect the ND filter to offset the addition of a light source so the reuslts would be identical. Same light, same ISO, same noise. Agreed?
Fair argument, except the iso isn't bound to shutter speed or f stop, you can set it manually. And regarding that, iso introduces noise! because in video you can't change any other settings.. This is called SPS in engineering or paramater dependance in equations. velocity and acceleration are bound to each other as with the throttle/spedometer example. Tonys argument only makes sence with camera in shutter, auto iso or aperture priority
YES.... YES.... and YES.... Thats EXACTLY what happens. ISO is sensitivity of the image sensor to light, i.e sensitivity is increased / amplified electronically. If there is less or no light, then the output of increased sensitivity of the image processor would be noise because it has nothing to amplify. Sadly. this video would only make sense to people who are new to photography.
Shoot something exposed correctly at base ISO and then shoot it again exposed correctly with the ISO cranked higher and there will be more noise...this is nonsense, its like saying improper shutter speed doesnt cause motion blur a fast moving object does
That's true! This is why I suggest watching the ISO like a speedometer. But the ISO is not causing the noise, the low light is. It's the difference between causation and correlation, and that's important to understand because correcting the problem cannot be done simply by changing the ISO.
i trust you (and have through other controversies), and have learned just about everything from yall (ive watched literally every video), this one seems wrong to me though (according to what i've learned from you). interested in seeing your response,
I'm speechless, can't digest the logic of this video, how the hell he compare black under exposed photo at 1/640 sec F2.8 ISO 6400 to slightly over exposure at 1/6 sec F2.8 ISO 6400 and then made conclusion that short shutter cause noise.
I watched your previous video about ISO long time ago, and I have to say you're right. The conclusion is, every photographer needs to get the correct amount of light before taking any photos.
"THE NOISE IS CAUSED BY UNDEREXPOSURE" or "THE NOISE IS CAUSED BY RAISING THE BRIGHTNESS IN POST"
This is easy to test yourself, but I did it for you here: ruclips.net/video/xVS_amf0DQk/видео.html (warning: crappy screen recording). As that demonstrates, underexposed images can be cleaner than properly exposed images. Or not. There is no true correlation between exposure (meaning, the brightness of the image) and noise when shooting raw.
Many of us (myself included) have previously been taught that high ISO does cause high noise, and it's much harder to unlearn a concept than it is to learn it for the first time. Here's a nerdy summary for my more technical audience:
First, make sure you understand the difference between causation and correlation. In a car, your speedometer *correlates* with your speed, it does not cause it. Your accelerator *causes* your speed.
High ISO *correlates* with high noise. If you're using high ISO with a properly exposed picture, your image is going to be noisy... However, lowering your ISO does not lower the noise, just like forcing down the needle on your speedometer doesn't slow your car.
Low light *causes* high noise. Regardless of the ISO, regardless of the exposure, pictures with less light will have more noise than pictures with more light. Therefore, if you want to reduce noise, add light.
What I hope to accomplish by demonstrating the difference between correlation and causation is to change how we think about ISO... MANY photographers in low-light situations limit themselves to a low ISO, like ISO 800, because they incorrectly believe that choosing a high ISO will increase their noise. Instead, they get an underexposed picture that has the same amount of noise (but requires additional processing). Hopefully, this lesson teaches those photographers that they only way to reduce noise is to increase the light their sensor is getting.
I would also like all photographers interested in technical image quality (noise, sharpness, dynamic range) to be conscious of using giving their sensor as much light as it can handle - and using auto ISO as a measurement is a good way to do this. If you're camera isn't choosing the base ISO for the exposure, then you're not getting your camera's maximum image quality. You can't always add more light, but it's a good to quickly note, "Oh, I'm at ISO 200 - can I use a longer shutter or a wider aperture to get myself down to a proper exposure at ISO 100? Can I add light in some other way?" One stop more light improves image quality about as much as upgrading from APS-C to full-frame.
Exactly! 🙂
Great addendum. I am definitely one who wondered if there was a difference in picture quality between under-exposing a picture on purpose to lower the ISO and then bringing the exposure up in post, and just using a higher ISO.
I think you missed some subtleties: Your test relied on using a dual ISO and iso invariant camera, if you had used an iso variant camera the low iso shots would have had much more noise when boosted in post. This obviously doesn't contradict your point.
@@splashstrike In fact in that case high ISO produces less noise than low ISOs.
That's why people say "choose the HIGHEST ISO you can without clipping the Highlights" to get the lowest amount of noise
In the linked video you show an under exposed image to a "properly" exposed image. However the under exposed was at ISO 100 and the properly exposed image is at ISO 10,000. You then raise the under exposed image 3 f-stops. If you wanted to compare increasing the exposure to ISO the under exposed ISO image would be approximate ISO 800. Comparing ISO 800 to ISO 10,000 is really not a fair comparison. Yes, shot noise does have a major affect on noise in low light levels, however I don't see you discussing that here.
“The fourth corner of the exposure triangle.”
Maybe it's a pyramid ? Prism ?
@@Arovna dun dun DUNNNNNN
@@dubisirie2 just as long as it's not a mobius strip or klein bottle
Tetrahedron
He is too smart for us!
TONY: "People haven't hated me enough during this lockdown, what sould I do?"
😂
He can keep on with the semi - nonsense coming out. Like biggest breakthrough in digital is that you can upload to the computer fast. Tony had run out of things to take about
@@paulscottfilms That post was edited?
1+1=2, 1x2=2; therefore 0+1+1=2 and is better than the other two equations.
Tell them you want the lockdown to continue till there’s a vaccine.
Jared Polin Is going to dislike this 👍
why?
@@tylermai1436 they like to beat each other up 🤣
😂
With hair like that he can only have ever worked in low light.
Fro Doesn't Know hahaha!
It’s not film…it’s an electronic camera with adjustable gain. This is effectively an amplification circuit which is amplifying the signal coming off the sensor. What you’re talking about is signal to noise ratio. Obviously, if you have plenty of signal (light) you have less noise… less signal, more noise. It’s simple… and nothing like a speedometer. Also proving your point with a dual base ISO camera just confuses the argument totally.
ISO, beyond the basic sensitivity level of your sensor (usually 100-200) is a gain amplification applied to the signal coming from the sensor. If you use it to compensate underexposure, instead of using the real exposure parameters (duration and aperture) it will definitely amplify noise.
Yeah this whole video is basically just a misunderstanding of gain and SNR
The noise always exists regardless. When you don't have enough light in the photo (because of a fast shutter/slow aperture) then you have to apply gain to the image (either in camera with ISO, or in post using exposure/brightness adjustment) you amplify the noise
Tony gets a lot of things right but he's completely misunderstood the science here - at least in how he explains it. The source of the gain is irrelevant, and the solution is always the same: more light, so that you have to apply less artificial amplification/gain
Tony just blew my mind! In the audio industry we often refer to this as signal to noise ratio. I now realize the sensor noise is static, and that the iso setting is just adding gain. I never realized how much photography has in common with audio engineering until now. Light is a wave just like sound. My camera is a transducer just like a microphone, so all the rules of reflection, absorption, and distance (inverse square) apply.
I've transitioned from audio to photography and can confirm that many concepts are very similar if not the same. I have a lot of issues with this video though. Most of them have been addressed in the comments.
Really good comparison!
yup and an amplifier won't necessarily distort a clean signal (if the latter is clean at source). If, instead, is distorted such as an electric guitar with an overdrive effect, then even at low volume the sound will be 'dirty', regardless.
Interesting
3 years after the original comment. Can confirm, as well.
Hello,
An automation engineer here with some knowledge of signal processing. I don't agree with your explanation: there is always noise, but amplifying the signal to properly expose an image increases its visibility, nothing else. The true exposure is defined by the light transmission of a lens (approximated by the diaphragm) and the exposure time (shutter speed). The camera can also have several basic sensitivity levels: simply take the signal power spectrum of an image and apply a gain to it. You can amplify the signal by using a higher ISO in your camera or in post-production, but the result is very similar with the raw format. But it's usually best to do it as close to the source as possible to avoid adding transmission noise before amplification and to work with the actual analog information and not with the digital information that has already lost some of it due to the quantization process.
Hey Tony, please look at this comment. I was about to comment something similar - you should look into signal processing, S/N ratio, signal amplification/boost/gain. You consider yourself a nerd, maybe doing real research would be nice - observing through experimemtation is fine, but it's no good at all if you have no clue on what is really happening. I have to congratulate you, not everybody can get it wrong (far from the target, btw) twice on the same subject in front of millions of people - maybe on the third? :)
You missed the point of the video big brain 😂
For what it's worth, I don't think quantization significantly affects the noise floor on most cameras as long as you're working with high-bit depth raw files, at least until you get into the 10+ stop gain range. I haven't even run the numbers on my camera though. :)
No no no no
He knows this, he is just trying to give everyone a different point of view. I too am educated on this, so I understand and agree on your explanation. But also on the thinking outside the box view of Tony.
Not trying to take it out of context, but let me try to understand your key point. "You can amplify the signal by using a higher ISO in your camera or in post-production, but the result is very similar with the raw format". So you're saying it's very similar, and results appear to be very similar (Tony's point and experiment), but generally it's recommended to amplify the signal (and noise) as close to the source (camera)?
From what I saw from the video, I don't think Tony is in opposition to your point. He probably didn't intend for people to take underexposed pictures with ISO 100. Those are experiment to prove his point, and they are done in extremes (rightfully).
He's saying to us, that let's not be ISO-centric, because ISO is a imaginary construct. ISO basically is a means for the camera to increase the amplification. Let's be aperture and shutter speed centric, and use them to exposure a picture properly, because that's the actual "hardware" gains you'll get.
I don't see contradiction from what he's saying and what you're saying, and you guys appear to agree from my point of view actually (but of course there are others who got very aggro-ed because Tony is saying something that opposes their "mantra")
A few points for people still confused about this:
1. When we say "noise" we generally mean perceptible noise. Grain we can see. All sensor readouts contain noise (the noise floor of the signal). It's only a problem when we can perceive it. At the levels we deal with in photography, it's caused by a few things, but mainly the random quantum behavior of photons, and electronic readout noise.
2. Dynamic range is the difference between the brighest part of the image and the noise floor. When a sensor is underexposed, this difference - the dynamic range - is smaller. The noise is still there, but it's too dark to see. However, a smaller dynamic range means a larger gain is required, and that gain applies to both the parts of the image we want (the signal) *and* the noise.
3. On mirrorless/DSLR cameras, ISO knobs usually apply gain before the camera's image processor gets the data. The exposure slider in post does the same thing, but after. On some cameras, ISO gain happens during the analog stage, which can reduce the effects of read noise, but most cameras *generally* have low read noise. In any case, applying gain with the ISO knob or in post makes very little difference when we're talking a few stops (ie. ISO200->ISO800 or +2eV).
4. The effect of adjusting ISO is different depending on your camera's mode. I think this is the key that has most folks confused. If you use your camera's automatic modes (program mode, aperture priority, shutter speed priority, etc) then raising ISO *will result in more perceptible noise* because your camera will underexpose the sensor. It is this underexposing which results in a lower dynamic range, lower signal:noise ratio, and more perceptible noise after gain is applied.
5. In manual mode, the only thing that affects perceptible noise is dynamic range, and this is determined by your exposure, not your ISO.
The noise is always there. It's part of sensing our world. It's only a problem when we can perceive it, and that only happens when we amplify it. We only amplify it because the bright parts of the image aren't bright enough, and this only happens because we underexpose the sensor. The same is true in the audio world: in audio, we don't call it ISO, we just call it gain. When you turn the gain on a microphone up, you hear a hiss. That's noise, amplified. You can speak more quietly (akin to underexposing a camera sensor), but will acheive a higher signal:noise ratio with less hiss when you turn down the gain and speak more loudly (akin to a proper exposure).
Lastly - this is not something to get worked up about. These are subtle differences for camera and math nerds. We find them interesting, but at the end of the day, it's the art that counts. If you still don't understand the point Tony's trying to make, then don't worry about it! Composition, storytelling and emotion are far more important than the details of how cameras calculate numbers from a bunch of voltages. :)
Well said, Someone that is actually educated here and knows photography. Thank You 🙏✌️
Wow. Great insight.
But I don't quite get no 4 and 5 point. Do you say ISO in manual mode is different than any auto modes?
@@abdulhasibtamim6905 I think this is actually the point that causes so much contention between people! In auto exposure mode (ie. aperture priority, shutter priority, full auto, etc) the camera adjusts the exposure when you change the ISO. So if you double your ISO, the camera exposes the picture for half as long, or closes down the aperture. This "causes" noise from a user perspective.
However, in manual mode, changing the ISO doesn't affect the exposure, because exposure is under full manual control. So, changing the ISO in manual mode doesn't change the level of noise (on ISO-invariant cameras, anyway), because the exposure hasn't changed.
This is why there's been some frustration with the "exposure triangle" lately. ISO is not exposure anymore than the gain knob on a microphone determines how loud you're speaking or how far you are from the mic. They're connected, but not in the way a lot of people are taught from the beginning, and it gets extra confusing when we consider automatic exposure.
@@GordLamb Ok. What's the source of this information?
@@GordLamb Wherever you get the information I think it's not right.
I just test with sony a7III, take a test shot with aperture priority and dial the same ISO, Aperture, SS in manual and got the exact same result. With the help of 'difference' layer blending mode in photoshop consolidates this too. So finally I am 100% sure that there is no difference between manual and auto exposure with regard to noise.
I read many articles in Internet and watch lots of videos and come to know that there are three fundamental types of noise.
(i) Analog noise: which comes from (a)heat noise, (b)shutter shock, (c)photon noise.
(ii) Intermediate noise: comes from (a) read noise - basically created during analog to digital signal conversion and also depends on quality of sensor itself (b) analog amplification noise - based on ISO you selected (eg if base native ISO was selected then no amplification, if ISO is 1 stop higher than base ISO then signal will be amplified by 2 hence signal noise will also be amplified as such)
(iii) Digital Noise: Comes from any amplification to the digital data (ie further exposure was added in post or in-camera JPEG conversion).
Exposure is determined solely by shutter speed, lens aperture, and scene luminance. Insufficient exposure will make more noise visible. To get less noise, you have to raise the exposure. ISO just maps an exposure to a desired lightness.
Your sensors sensitivity to light is fixed. It cannot be changed in any way. How much light reaches your sensor is determined (camera wise) by aperture and shutter speed. If little light reaches your sensor, the signal to noise ratio goes up as the sensor starts to mis-read the vague light. ISO is a factor the processor/software of your camera applies to the readings of the sensor. High ISO when used to compensate for lack of light will hence amplify the noise that is already there. So the little word "cause" is the debate here. In my mind, high ISO does not cause noise as such but will amplify it if present. So if you shoot with auto ISO on and the ISO value goes ballistic then you know light is missing. I think that is Tony's point. And a good one. So see the ISO as the canary in a coal mine. And don't worry about all the tech stuff.
For all non-technical people: it's the low signal to noise ratio (SNR) that ruins your picture, not the boosted gain that is applied to make the image brighter.
Well said, Someone that is actually educated here and knows photography. Thank You 🙏✌️
🙏 agree, what's wrong with teaching people what iso actually is!
High IS or increasing the brightness in post amplifies the noice so yes ISO ruins your picture.
@@reldies5364 it's ultimately the signal to noise ratio that ruins your photo. The ratio never changes after you apply the amplification since the noise also gets amplified together with the signal. If your SNR ratio is 10dB at 64 ISO then it's still 10dB at 6400 ISO.
@@sctm81 True until amplification causes clipping of the signal (AKA blown highlights)
If you a confused, the point Tony is trying to make is if you are shooting in the dark and already have aperture and shutter speed maxed out, lowering the ISO won’t do you any good in noise reduction. You need to physically get more light in the scene. Thanks for the tip Tony, I remember trying to reduce ISO thinking the darker image will have less noise when I brighten it in post before.
Wrong title. Should lower exposure cause more noise
But this title is more attractive. You know there will be less people watching this video if they use your 'correct' title, right?
@@-xox- Not only more attractive, but also more necessary because a lot of us (okay maybe just me) think that high ISO introduces noise. So the thinking needs addressing.
Technically high ISO causes the noise because it turns up the gain of the sensor. Of course when turning up the gain, the dynamic range is shifted, you need less light no to overdrive the sensor. Same principle applies for audio gear. There is gain only, no shutter neither aperture. If you turn up the gain it can pick up low level sounds but the noise goes up by also. If you increase the gain but the sound is loud, it will clip, like tony's photo. Of course audio can be more complicated, there can be several stages of amplifiers.
@@dzsemx Thing is it isn't creating noise.. it's only amplifying it. Gain doesn't affect the S:N ratio, only the amount of perceivable noise. The noise itself is created at the time the sensor is exposed and read. Same with audio; turning up the gain doesn't create the hiss, it just makes it loud enough to hear.
Gordon Lamb Under-exposing an image is like having too low an audio signal. You have a fixed level of sensor noise so if you underexpose then the signal to noise ratio is low. If you then amplify the final under-exposed image (ie boost in post) then you amplify the noise too!
0:43 uses dual gain sensor to show high ISO doesn't introduce noise in comparison to a low ISO image corrected in post with the high ISO image clearly superior in the low-mid tones.
DOH!
An apples to oranges comparison where the sensor is not operating with the same parameters, effectively being a completely different sensor with regards to dynamic range due to the gain applied at the sensor and not after the ADC conversion as is usually the case for these types of comparisons (digital gain).
Given that there are many strategies with regards to gain, it cannot be claimed as a universal that ISO settings are effectively noise neutral to the output when your first example clearly demonstrates that this is not the case.
...and is fixed by adding more light...i am sure he didnt keep the ISO the same after adding more light LOL
I have to say, that I was very skeptical of your premise, however, I just got done performing the same procedures you showed, and I was shocked to get the same results with my D800!! Thank you and your wife for all the hard work and effort you put into these videos!!
He's testing this on a Sony A7Riv which is ISO invariant above ISO 320 (ie. where the second amplifier kicks in). However, if you choose another camera which doesn't have ISO invariance (eg. Canon 5D Mark III) then the results should be quite different...
The title should be: „+5 exposure in Lightroom causes more noise than with no adjustment”. Duh...
Kamil Pekala or Camera ISO is better than post.
I was thinking the same thing at first, but the more I thought about it, it just makes obvious sense. Although, he doesn't explain it very well...as per usual.
While all of the example photos were noisy due to the exposure increase in LR, what he really proved is that using a higher ISO in camera provided less noise than increasing the exposure in LR (at least with the camera he used).
The confusing thing is that the point he was trying to make was entirely different. He was pointing out that a lack of light is what causes noise - which is rather obvious so we tend to not ever think about it that way.
And his solution was to use the auto iso as a gauge to tell if more light is needed...which anyone should be using the viewfinder info, histogram, or just basic photography knowledge to deduce anyways.
He was bored and needed a video topic...
Maybe it would be more correct to say that ISO is more like a volume control. You increase the volume control to hear the music better, but all the hum, static, and other noise is increased as well. The volume control is not causing the noise, just making it more apparent.
This is one of Tony's "duh!" videos. Raising ISO to avoid slow shutter speeds or to avoid large apertures essentially is underexposing a photo. So it is the same as taking an underexposed photo into Lightroom and increasing the Exposure slider. It turns out that more expensive wide-aperture lenses and/or Lens stabilization (to allow for a slower shutter speed) and/or tripods all allow for an exposure with the right amount of light. Anything else is using ISO as a cheat that will bite you later down the workflow. From other videos on RUclips, all sensors produce noise every time you take a picture. The more real light you are able to obtain during the exposure overrides that noise more and more (called signal to noise ratio). And yes, larger sensors do indeed have a lower native noise level than smaller sensors, all other things being equal.
This is quite interesting actually. I see a few point of views:
1) Yourself, that feel that this is very intuitive and should be clear to all. Tony is overexplaining.
2) Myself, who I feel that the video is sufficiently explained and I feel like that's a good point, why didn't I think of it. I always thought that there's some magic way ISO amplify signal without increasing noise too much. Seems like I was mostly wrong.
3) Some others, who feel that claiming ISO doesn't change anything?!?! That's the mantra of photography. How dare Tony! He must be mad!
And that's cool! Though I feel some comments were more rude than they should be
@@samuelpua9771 I fully agree with you Samuel. I was not trying to be flippant. I love everything Tony reports on and he often injects clarity in something I think I already know but not well enough to take advantage of in certain situations. Thanks for the comment.
You need a combination of large aperture and higher ISO to campture dimly lit moving targets though. Cannot go by on tripods and other forms of stabilisation unless subject movement is within acceptable limits. This high ISO 'cheat' is mandatory at times. Also, the video demonstrates that on dual gain sensors, raising ISO yields better results than correcting exposure in post.
@@tfm2934 That sounds great! Sounds like you're prioritising on large aperture and fast shutter speed (to capture moving targets), thereby thinking of ISO last. Sounds exactly like what Tony recommends in the video.
Obviously, the underexposed pictures are part of the experiment to see what's the effect of the 3 items, ISO (digital gains) being the least effect (usually on dual gain sensors, but otherwise minimal effect). You would still be taking properly exposed pictures, just with aperture/shutter-speed centric mindset, which is what you're doing exactly
@@samuelpua9771 Exactly. I think the video could have been simpler on that point. But I agree with you that many comments are overly rude!
I think a comparison to a tachometer (measuring RPMs) is more accurate than a speedomoeter. When the RPMs get too high, you change gears, which is effectively what you're doing by adjusting your shutter speed/aperture.
I very very much appreciate the effort taken by
mentor Tony to see & understand
this profoundly important aspect of picture - taking !!!
The lack of a greeting made me hear the opening as “Hi, ISO.” 😂
😂
Tony & Chelsea Northrup Good video though. Definitely good to get some short informative video like this.
@@TonyAndChelsea can you make a video explaining how you would effectively use this technique in a concert please. Thanks a lot for this information
@@smartbiney801 also a test using a super high ISO, such as 40,000-102400
@@smartbiney801 I'd also like to know how to shoot a dress rehearsal of a ballet without adding noise. Sure there's lighting, but it's still fairly low light. You need a fairly fast shutter speed... the lowest I used was 1/160. My photos were good, but the noise was problematic.
So if I uderexpose an image and drag the exposure up in post it'll be noisier than if I exposed it properly in the 1st place? What I've learned was that we need to end this lockdown so Tony can go outside and play!
lmao, thank you, this is indeed the reason why this video is so confusing. He tells it as if it is something new. If you're underexposing you can either turn up your iso or turn up the exposure in post, but both will result in more noise. So don't under expose if you can.... hahaha, yes that make sense, thanks!
Edit:
Adding to that, images shot with high iso also always will have high noise. Just as Tony says in his video, if you shoot with a too high iso on a bright sunny day, you will still get noise, because of the high iso! Overexposing and than reducing the exposure in post will still result in more noise than if you didn't shoot with the high iso.
In conclusion, you can see high iso as a result of underexposing, which means noise. The real lesson is that you shouldn't see iso as a real element of the way of lightning your photos. It as fix that can boos your exposure at the cost of noise. You can also do that in post, but doing it already in the camera may let you use some noise reduction tricks and hardware your camera has build in. So to summarise, use iso as a last resort. if you see that it is high, see if you can fix your exposure in another way.
That is however something I've always done, but this video helps with making you think how it works.
Not noisier; the noise was always there... you just couldn't see it until you raised the luminance.
Analogy - you're in a messy room with the lights off. Turn the lights up, and suddenly you see the mess. It was always there... it was just too dark to see.
@@GordLamb Yes exactly, the noise is there from the moment you're underexposing. If you than boost everything to correct the exposure, you're also boosting the noise and therefore you can see it (more clearly).
@@Zegmaar_Bas Right, though technically the noise itself is always there, even in a correctly exposed image. It's just not perceivable, until like you say, you amplify it along with the rest of the image (which obviously you wouldn't do if it wasn't underexposed).
@@Zegmaar_Bas In other words always shoot with the lowest ISO you can muster. Maybe use a tripod at night instead of hand held high ISO shots. Then again, people think shooting with big apertures is a good trick at night to allow more light to enter the camera, but you have depth of field considerations as well. So shoot with the biggest aperture you can get away with and the slowest shutter you can get away with and the lowest ISO you can get away with before cranking ISO.
If I understand correctly, the takeaways are:
1. Underexposing causes noise; a properly exposed image will always have less noise (regardless of ISO) than an underexposed image pushed in post, even for ISO-less sensors
2. Among properly exposed images, the one with the highest ISO setting will be the noisiest
1. Underexposing doesn't cause noise if you keep the aperture, light and shutter speed the same... the noise is pretty much the same whether you properly expose it or adjust it in post.
2. Yes, high ISO correlates with noise, but does not cause noise. That's why I suggest using auto ISO as a speedometer, so you'll have some sense for how much noise you'll get in the final image. If you want to reduce the noise, you need to look at light, shutter speed, or aperture, because those have a causal relationship with noise. ISO does not have a causal relationship with noise.
@@TonyAndChelsea Correlation not causation.
@@TonyAndChelsea You still have it backward. Shutter speed/aperture doesn't cause noise. Those are the physical paths of the light. The ISO itself is the ONLY part that actually introduces noise. The sensor doesn't care about the aperture/shutter speed, those are separate physical phenomena, the sensor itself, translating the energy of the photons into digital bits, is where noise is introduced. "noise" is just another way of saying "uncertainty of measurement". The aperture and shutter do not directly effect the measuring process. The only setting that actually changes that process, is the ISO setting.
The concept of using auto-ISO as a speed limit is quite sound. And what many of us already do. That makes perfect sense and is a useful message to share.
But please don't double down on the incorrect parts of your video in order to justify the useful/correct parts. That's just ego.
OK, that's why the first image that was underexposed via ISO was the cleanest of all boosted images.
It wasn't exactly as good as the high ISO image due to the dual gain (does pushing 6 stops in post border on the limits of ISO invariance though or is it unrelated?).
Taking it even further, for ISO-less sensors at least, one could say there is no longer an exposure triangle, as those cameras effectively have a single ISO and exposure is only determined by aperture, shutter speed and available light.
@@TonyAndChelsea Are you sure the noise is about the same when adjusting in post vs. using higher ISO? I was under the impression a lot of cameras actually apply analog gain depending on ISO. At leas with my old 50D it makes a huge difference, noise is brutal if I adjust it in post instead of using a higher ISO in-camera.
I think the takeaway here should have been a properly exposed image reduces noise. In a dark environment, If you have the ability to add more light (ie. a constant light source or flash) then you will reduce noise, if you have the ability to lower your shutter by using a tripod thereby exposing the sensor to more light you will reduce noise, if you have a faster prime lens and you shoot wide open thus exposing the sensor to more light then you will reduce noise. I think this has to do with the signal to noise ratio no? Simply bumping the ISO will not in and of itself increase the amount of light available - thus more noise introduction. It's a fascinating topic though. Thanks for taking this one on Tony. Always one to face the controversy head-on - lol ;)
I honestly think that this lockdown is taking its toll on Tony’s sanity.
LMAO
hahahhahhahahh
haahahha this reply deserves more attention
ISO is a signal amplifier. The signal in photography is light. Less signal = more noise. If you boost the exposure of a scene, either when taking the photo or during post-production, there will be more visible noise because of the lack of light. The information lacking in the picture because of not enough light is filled by random pixels of different colors by the digital cameras.
Anybody practicing "exposing to the right" knows that sensor saturation is paramount. If a high ISO helps/is required, OK, then use it. In my experience, using a lower ISO with proper exposure to the right results in less noise.
@David You can always shoot with the histogram enabled and avoid clipping in real time before you shoot. I own a basic DLSR (SL3) and do that with live view. With mirrorless cameras you can do that in the viewfinder.
But I agree that avoiding clipping must be No 1 priority.
If you lower the image information level into the noise floor and then amplify both the image information and the noise in an attempt to recover the image, the image noise will be more apparent. The ISO setting amplifies the signal arriving on the sensor and by default the sensor’s noise, therefore the noise is a higher proportion of the signal if there is less signal. Amplifying the signal amplifies the noise.
High ISO is used because there is insufficient exposure of the sensor to light reflecting from the subject, raising ISO is used to amplify what is being received. Otherwise, we just accept that we run out of road and use the widest aperture and longest time necessary to decrease the proportion of our signal that is noise.
The noise is inherent to the sensor and the amount of signal amplification applied by the ISO setting, not the aperture or the shutter speed, those are independent of the sensor and its noise floor.
In any shot overexposing the sensor pulls the signal further above the noise floor. By recovering the image in post, because noise floor was already very low you reduced its effects by reducing image brightness to get the appearance of a correct exposure. Going the other way round and underexposing the sensor pushed the signal into the noise floor so changing the image brightness in post to compensate reveals the noise the image is imbedded in. The processing engine of a camera already does a lot of heavy lifting to fix the balance between noise and image. Adjusting the iso partially digitally processes noise effects away after amplifying the signal. Newer cameras tend to cope with under exposure better than older ones because noise processing may be better.
It’s kind of like saying, jumping off a building doesn’t kill you, it’s the sudden impact at the end that does.
If you're going for a meaningful analogy, this is exactly the opposite: It's the low light hitting the sensor that makes people use high ISO and get noise. So the low light is the cause and the high ISO is the effect. Just like jumping off the roof causes you to hit the ground.
It's even worse because he says the shutter speed and aperture cause noise, but ISO does not. It's like saying jumping off a cliff or out of a plane will kill you, but when you jump from a building, it's the sudden impact that does it
Is there PROOF that jumping off a building doesn't kill? What if a person gets a heart attack BECAUSE he is fearful at the start of the fall? Just wondering. :-)
Tony is actually right that high ISO isn't the demon. I found this out by mistake one time - eons ago when I accidentally kept one of the very older (Canon 60D) cameras at 1600 ISO when moving from a darker room to a brighter room. I changed the shutter speed but not the ISO and the fact that the sunlit room was awash in light, I was amazed at how clean the capture looked even at 1600 ISO - bear in mind this was back when 1600 ISO was pretty much a max limit for digital captures. It was the first time I realized that a bit of overexposure at higher ISOs was a GOOD thing - and that the real issue was combining underexposure (lack of light) with high ISO was the culprit. It's better to overexpose a bit with high ISO images and absolutely damning to not have enough light and shoot high ISO.
What's the point of the a7s series? For some reason I thought it was excellent at low light & high iso. Hm.
Yes, but underexposure is always going to be a culprit no matter what camera.@@jordanbabcock9349
When i shoot the milky way with 1600 iso, the image has more noise then when i shoot outdoor with sunlight using iso2000. The missing of light cause the noise
Absence of light
Good point!
You explained it better than Tony
I don't think that's a useful comparison. You would need to compare the sun-lit image at ISO 100 with an equivalent scene shot at ISO 2000 and 20x the shutter speed.
Brent Schumer my comparison was not accurate by any means, only something I noticed while shooting
In a Milky Way photo, you want the sky to look dark, thus you are underexposing it and you'll have noise.
Studying astronomy has made understanding photography so much easier. We work with CCDs and I'm responsible for checking the gain of the detector, which we use a monitoring script for. The gain is basically how sensitive the detector is to photons, and how much it can convert photons into electrons. The unit for gain (in astronomy) is electrons-per-DN where DN is digital number or the photon number. So if your gain is 2 e-/DN then for every photon that hits your sensor/detector/CCD you get 2 electrons of measurement.
So with ISO if you crank it up from 400 to 800, you're effectively going from let's say a gain of 1 e-/DN to 2 e-/DN. Same for 800 to 1600 and so on.
The amount of noise in a single pixel follows the poisson noise (shot noise) and is equal to square root of N, where N is the number of photons hit. If you don't have enough photons in an image (aka its severely underexposed), your shot noise relative to your photon count is going to be sizeable. If you have plenty of photons in an image (appropriately exposed) your shot noise is going to be minimized.
So if you had 1 photon hit a pixel, your shot noise is 1. Whereas if you had 100 photons hit a pixel, your shot noise is 10. 10/100 is less than 1/1. So basically if you appropriately expose your image given the gain of your detector (ISO), you can minimize your shot noise. This is basically what Tony demonstrated.
We have calculators set up to optimize the exposure for scientists who use our facilities, we call them exposure time calculators or ETCs. It's like the instant readout you get on your camera for your shot, but in this case scientists put in the object they want to look at in space plus what filter they want to observe in (infrared, ultraviolet, blue/green etc.) and it does a series of calculations to tell you what the exposure time should be, what the detector should be pre-flashed with light to, and what the final estimated brightness will be.
The first thing I said to myself was here we go again Tony hehe
Honestly... This is exactly what I said to my self immediately I saw the title
Lol.... ikr
LOL true
@SwitchRich It means less light = more noise, more light = less noise
@SwitchRich For a given (correct) exposure, if all settings are locked and you increase the f-stop, you will lower your dynamic range / signal:noise ratio. This is not an opinion, but rather a fundamental property of math and physics. You can demonstrate it with a camera in a carefully controlled experiment, or you can run the math with a whiteboard.
(apologies if you're being sarcastic, but on this topic ya never know :)
I think I get it, but let me try to restate it in simple language for us simple old people: If I am taking pix of my lovely spring flowers, and I notice that I need to crank up my ISO to get "proper" exposure, what my camera is REALLY telling me is that my fstop is not low enough, or that my shutter speed is too high. And if I reply to my camera, "But camera, I need my aperture closed down this far for proper depth of field, and I need a higher shutter speed to stop motion blur because the wind is blowing," then the camera will reply, "Then shine some light on the flower, or wait until the wind calms down, and for heaven's sake use your tripod, because there's not enough light now to keep the noise down!!" Am I close?
Yes, James, this is an awesome explanation!
Thanks! Now if only I knew how a camera having a "dual gain sensor" tweaked this situation...
It all comes down to the signal to noise ratio. If you cut down the light to the sensor, the ratio between your image and the noise floor decreases. You recover your image either through amplifying the signal+noise either in the camera (higher ISO) or in post processing. Same difference. So noise is caused by lowering the level of the signal (light) to the sensor. EE's point of view.
Mr. Northrop you have saved me, I have a Canon 90D and Sigma Art 18-35 F/1.8, my pictures always show noise even at base ISO, I was ready to quit and now I have watched and tested what you have said I have got rid of the noise problem. A massive thank you.
The process that you make in the computer when lifting the exposure is very different compared to shooting in a higher ISO (2 different stages in the image making/processing- Pre Gain VS After Gain). This whole video is extremely misleading. It only demonstrates the fact that you should get your exposure right and don't count on lifting it back in post. The claim that there is no correlation between High ISO to digital noise is ridiculous. The only thing that i can agree with is the statement at the end about using higher ISO as a last resort .
The signal and noise are both fixed in the raw file at the time of capture. Post-processing does not add noise.
"The claim that there is no correlation between High ISO to digital noise is ridiculous."
Agree
Modern Sony sensors are ISO invariant which means that shooting at for example ISO 3200 and shooting at ISO 800 but raising the exposure two stops will look exactly the same. But I agree his conclusion was a bit extreme, it doesn't work that way on older digital cameras.
@@TonyAndChelsea I understand the point you are saying that the lack of light is the factor that creates noise. What i don't agree with is saying that high ISO doesn't cause noise, because it does. This due to an electronic enhancement of the signal right after it is being captured by the sensor. There is no getting around it. High ISO generates more noise if you shoot the same exposure right out of the camera. The comparison you made while trying to bring up the exposure in post is incorrect.
This is somehow true when you are talking about Canon's cameras that do not have "ISOless" sensors, try that on a Sony camera and you will see that there is no such thing as Pre-Grain VS After Grain. ISO is just a number that indicates the amount of signal amplification that will be implemented in the final image. Light, exposure, and sensor performance will determine the amount of noise on your image, ISO itself does not have any effect on image noise as long as you are able to give your sensor a good amount of light exposure.
I am sick and tired of Video Low light "tests" that compare noise performance at different ISO speeds while keeping the exposure and the light the same. These are the dumbest tests anyone can ever make, yet everyone just keeps doing them just to prove Einstein right when he was talking about human stupidity and the Universe.
You are wrong on so many fronts! ISO is a gain control, it takes what you capture and amplifies it. If you have noise mixed with the light, SNR, then you will amplify both, if there is plenty of Signal (light), but very little noise in the capture, let's say, 90% signal and 10% noise, then any gain you apply, will increase both, but 2X 90% is far better than 2X 10%. In low light situations where the SNR is more in favour of noise, 60/40 split, using the gain will result in 2X 60% noise as opposed to 2X 40% signal. You don't need to be a mathematician to realise the bigger percentage will always win out. Electronic components also create their own noise, which adds to the total noise when engaging high ISO settings. The worst mistake you made, was using post production to try to compensate for shutter and aperture settings, as post production can add quite a lot, if not excessive noise to the already bad image. The answer is quite simple, you get the best SNR, ( light), from the composition you can, before you apply the ISO gain. In other words, in low light situations, drop your shutter speed, or pick a bigger aperture before you resort to picking a higher ISO. If you don't, and your capture has a high percentage noise level, upping your ISO will only boost the noise level and make the image worse. The more light that hits your sensor, the less noise there will be!
I've never seen Tony respond to so many comments :) He's right. Test this with your own gear (without blowing out the highlights) before disputing it. I was surprised with my testing!
Just did. And this he isn't correct. He also knows it, which is doubly annyoing.
The concept of "ISO as a speed limit" is pretty clever and correct.
However, do it yourself: compare at the same aperture, ISO 100 @ 1/200 to ISO 400 @ 1/800. Same exposure. Tell me which has more noise. 🤔
@@shammyh You changed your shutter speed to test. You didn't prove anything. You have to match exposure in post to test this theory. Try keeping lighting, apeture, and shutter the same, take a pic at iso 100 and one at iso 400. Adjust the exposure in post 2/3 stop or so to make them the same exposure. Then look at the noise.
A lot of people are missing the point that high iso isnt inherently noisy. Which i always thought it was. Great vid for a beginner like me.
Sometimes Tony is the perfect image of that guy who butts into a conversation you're having with "ACKSHUALLY" and goes off, and you just roll your eyes.
probably you are saying it later in the vid (hope so)
But it's the lack of light, that causes noise (i think). It's the signal to noise ratio, what's important here.
and with higher iso, they crank up the voltage of the sensor, therefore your noise ratio is higher.
So i guess i am disagreeing with you here. Higher ISO does cause more noise, but the main culprit in the story is the lack of light. And because higher ISO's are being used at lower light environments, it's easy to point your finger at the ISO, it's just a little bit more complicated then that.
I normally think Tony gets stuff right.
But Flawed testing here.
The image noise primarily exists in the shadows.
So if you underexpose shots. You have more noise. Regardless of how you end up being underexposed. Shutter, aperture, light or ISO.
If you have the same decent exposure, but THEN increase ISO. You will get more noise.
That doesn’t happen for shutter or light or aperture.
If you have the same decent exposure but vary shutter, light and aperture for each. And keep ISO the same, Then you will have consistent noise level.
He clearly demonstrated the opposite in the video, in a way that you can easily reproduce at home. Keeping ISO the same but varying SS, aperture and light change the noise level, as they change the amount of light.
Increasing ISO does induce noise if used to correct exposure...If you increase ISO (or raise brightness in post) to correct the exposure 'when' there is insufficient light on the sensor you WILL increase the noise. In other words, if you use ISO (gain) to correct for underexposed, you will introduce noise in order to gain the correct exposure.
Great explanation Tony. I have always been an available light shooter and tended to towards wide apertures on expensive fast primes. I recently started playing around with some slower but modern Leica/Voigtlander/Zeiss M lenses adapted to my Fujifilm APS-C camera. I realized what I really enjoyed was clean images with sharpness and tone more than bokeh; shooting a scene at f5.6 or tighter gave me what I wanted more than at f1.4. The consequence was I just started setting my ISO to the most tolerable setting for that time of day and shooting with a mini tripod. Instead of wasting money on glass I started to use the creative restrictions of shooting with a $30 mini-tripod as my guide to setting up shots. I was always afraid to "waste" the expensive glass on more moderate f-stops but I knew I didn't want blurry photos; by admitting what I wanted I sold off some expensive glass and just focused on what was sharp as well as a few cool stabilization techniques. The noise (or signal to noise ratio) is always about available light and separating it from ISO is important in thinking about exposure; ISO is gain to boost signal (and noise) and the more likely you need to add gain the more likely you need more light in your shot for your preference in noise level. Like you, I always viewed the % of my maximum allowable ISO for the setting as a measure of my success in curating the shot. Excuse the long comment, hopefully someone finds that valuable.
What you're saying is that high ISO boosts noise (and signal) which Tony says doesn't occur. He specifically says in his opening statement, 'High ISO does not cause high noise' - so which is it to be believed, you or Tony? This theory of Tony's is fine for the studio, but has no basis in real life outdoor photography where the light conditions can only be altered by either a) moving the subject into a better lit spot, but if that isn't possible then b) by providing extra light in the way of flash or reflectors. Yeh, try that on a race track! Tony should have been more specific in what he was outlining because in circumstances that require fast shutter speeds and the widest aperture your lens is capable of in lighting that necessitates a higher ISO, then you will get more noise - fact! Lowering the shutter speed will of course lower the ISO for the same lighting conditions and subsequently lower the noise, but will increase blur (and I know which I would prefer) Alternatively, if the aperture is reduced in order to preserve the shutter speed and the ISO left as is (lower than required for the A/L, ) underexposure will occur and on none ISO variant cameras (which I imagine the majority of us have) then noise will be increased in post to the same extent as it would have been had the ISO been increased with the same aperture at the taking stage. So where's the advantage in underexposing and increasing the brightness in post? None from what I can see.
@@robinlikes2learn what is 100×0? Zero. What is 100×5? 500. ISO doesn’t multiply noise that isn’t there; it doesn’t cause noise, it amplifies it. If you increase light and thus increase signal to noise sufficiently there is negligible noise to be amplified.
thanks, this is really thought provoking. Just want to add something here, noisy image is caused by not high enough Signal-to-noise ratio, meaning the sensor cannot collect enough "good" light signal to combat the noise naturally occurring in the circuit of the sensor.
So high ISO is really not to be blamed, it's by increasing the ISO, you are allowed to use a faster shutter, which in turns reduces the amount of "good" light that hit the sensor.
This is flat earth conspiracy theory in photography.
With the difference that its popular.
not flat.
At 1:13, increasing exposure gain on pic 1 to match with pic 2 on post. As far as I know, in/decreasing exposure gain on post depends on ISO base of the picture. In this case, for example, if you gain the exposure of pic 1 +3 stop (idk how much exactly exp. gain on pic 1 to match pic 2 exp., so I assume +3 stop) from base ISO 6400, it would be similar to ISO 51200 on exposure. Noise should be worst than before, right?
Lets say, you take pic 3 with 1/640s f2.8 ISO51200 , it would be nearly same as pic 1 after exp. gained. Pic 2 will be much more clearer with long shutter speed as compensation.
Wouldn't this only be true for cameras that are ISO-invariant?
Yes it would, but someone just HAS to start brand wars every once in a while
For non ISO-invariant camera it's actually the other way around. Higher ISO means lower noise since the signal are amplified before the read-out noise been added in, so higher ISO means better signal-to-noise ratio and lower noise.
It's not a fair comparison as you made the pics brighter in post. A fair comparison would be iso 6400 f2.8 , 1/8th of a second vs iso 100 f2.8 8 sec.
good point, we can understand the video like this: increasing ISO = increasing exposure in post editing, they're same technical light boosting which causes more noise, but the ISO in the camera might work a little better in some way, that's why (ISO100 + increasing exposure in post editing) looks worse than High ISO
Raising exposure in post is simulating high sensor gain. This is a misleading video
I concur
That's his point. :) Neither gain applied on the camera using the ISO knob, nor gain applied in post create noise; they merely amplify the noise that existed at the time of capture. Dynamic range is determined by exposure levels, not gain, and captured dynamic range dictates the required final amplification. If you want to reduce noise, you need to capture a greater dynamic range, and you do that by correctly exposing the sensor, not applying gain with the ISO knob.
Gordon Lamb oh yeah, push the gain on your camera and correct exposure by using ND filters, aperture and shutter speed and tell me how the image looks.
@@ralphgeronilla so the whole point here is that gain/ISO has nothing to do with exposure. It's not part of exposure. ND filters, scene luminance, shutter speed, and aperture are really the only variables. So long as those variables are configured to maximize dynamic range of the sensor, you'll end up with the best possible result.
Because it is the same. Apply gain on post or do it in camera, gain is still gain
I have an older Canon 50D. At ISO 400 is is noiceless with a well exposed image. At ISO 1600 it's very noisy with the same well exposed image. Same lens, same camera, same aperture (f2.8) and the shutterspeed is two stops higher.
I've always thought of ISO in digital cameras along the lines as radio reception. At the base ISO that the camera has - and it has only one - the noise is excellent. A radio picks up sound from a local station and it sounds good. If we want to listen to a radio station whose signal is weak because it's a distance away, we turn up the sound and hear the broadcast more clearly but we also hear popping and buzzing as the background signal noise. When we raise the ISO in the camera we are telling the camera that we will accept the weaker signal (less light) and through camera noise reduction and RAW noise reduction algorithms we reduce the noise to make it acceptable.
RUclips - "Tony, do you want to say something crazy or something that will piss off the internet?" Tony - "Yes"
Hehehe
I just took 2 (identical) shots of a house ... both in aperture priority at f/13.... both with correct exposure...
One was shot at iso 200 and one shot at iso 3200.... not surprisingly- one yielded 1/200th sec shutter speed- the other 1/3200th...
No prizes for guessing which image yielded the most noise... I'll leave you to decide if it was the high iso or the 1/3200th sec shutter speed that was the cause...
Coming up next: Focal length has nothing to do with the FOV.
Well its funny because to some extent it doesnt ! xD
The noise is always there. The same noise is in every photo here. When more light hits the sensor it overrides the noise. So Tony's right, more light (signal-to-noise ratio) less noise seen in image. Look through your viewfinder with the lens cap on.
There's your noise.
I found this out myself by trying to limit my camera to low iso which meant I was underexposing and got noisy images so now I let the camera choose iso and keep checking the histogram for correct exposure and yes using the auto iso as a “speedometer” is a handy tip .
Noel Chignell but then lose detail , even more so when crops
I found that out too. I posted it above. If I let the camera find the correct ISO, then expose to the right based on my histogram, not only did the noise subside, but the color was richer. The noise was still prevelent, but way more manageable then exposing correctly.
So I came a little late to this, but it's as simple as to think that raising ISO is just like raising your exposure bar in lightroom. So it creates noise by retrieving the details in the shadows and mid tones. So technically it is not the ISO what creates noise, actually noise isnt created, its always there, it's just a matter of how visible it will be. And it will be more visible when there is less light, when there is more light, the light overlaps the noise (SNR).
is this a troll? I don't get it.
The noise on your camera's sensor is actually pretty static. No matter what your settings are, the noise stays about the same.
ISO increases the brightness in the same way that turning up the gain on a mic increases the volume. Any hiss on the mic is also amplified when you do that... just like the noise on your sensor will be amplified when you do that. Ditto with increasing the brightness in post.
The goal here is to increase the signal to noise ratio. In this case, the "signal" is the light you're gathering. Thus, if you want to "reduce noise", you simply gather more light by decreasing shutter speed or f-stop. The extra light helps drown out the noise.
@@Falcrist sure but on the real world you use the iso to compensate for the lack of light... so what the hell are we talking about? you should always have a correct exposure... if you recover the dark part of the image on post it's obvious that you are going to have tons of noise in the picture...
@@aizel330stream Honestly, I'm not sure what the point of the video is.
High ISO does indirectly "cause more noise" in the final image by making you increase f-stop and shutter speed to compensate. That reduces the amount of light gathered and thus the SNR of the image is lower.
But if he wants to be pedantic, he's doing it wrong. Even the high shutter speed and f-stop doesn't cause the noise. The noise was already there, you're just not gathering enough light to drown it out.
@@Falcrist exactly, that's why I don't get if he's actually trolling or what... ahahahah
@@aizel330stream He's not trolling lol
There's *some* value to understanding what I explained... Not much, but some.
I just think this video is a mediocre explanation.
in short: noise is a function of total light gathered which is affected by aperture (flow) and shutter (time). ISO is a gain applied to the signal post-capture.
2:28 "But officer, I wasn't speeding, the accelerator was speeding"
Great information. Thank you for sharing this as has helped changed my perspective. I am attempting to apply this insight to getting better drone footage at night. I have been having trouble getting the right settings to reduce noise while still getting enough brightness in the key areas of the image. The aperture is fixed so its been coming down to adjustments in the shutter speed and ISO. I have also found that that for the Anafi PLOG produces the better footage. And that HDR adds a ton of noise. Do you have any recommended settings for capturing night video via Drone?
Since high ISO do not cause noise Tony , please show us a well-exposed image at very high ISO that do not contain noise.
(It's sad to write and say things that are NOT true! Very sad...)
I proved this to myself last night. We frequently have whitetail deer in our back yard. In the last month I upgraded my Canon Rebel T6 to a 90D, and upgraded my glass from the 75-300mm kit lens to a Sigma 150-600 Contemporary. I live on the west side of a hill, so light fades in my back yard well before sunset. I first tried photographing the deer at 600mm F6.3 (wide-open for this lens), 1/1000, and auto ISO (25,600). I got nice sharp images, but there was a lot of noise. Taking the advice of this video, and another of Tony's videos (on finding out how low you can get your shutter speed hand-held), I moved the shutter speed down to 1/250. The image was underexposed, but when I looked closely there was little to no noise in the image. Thanks, Tony!
If you moved the shutter speed to 1/250 sec you would have increased the exposure/light so why would the image be underexposed ?
@@liverpix It was a long time ago. I may have messed with the ISO as well. Sorry, I just don’t remember the situation.
Tony: High iso doesn't cause noise.
Also Tony: Don't use high iso if you can, because your pictures will look bad.
Why would you want to use high ISO outside with plenty of light? You'd have to crank your f-stop number up and increase your shutter speed, thus creating more noise.
@@Joe-hm1zk Increasing your ISO actually affects how many stops of dynamic range you have above middle gray, so there are some situations where it can be beneficial. For example, if you're taking a photo outside where you care more about the clouds than the ground, increasing your ISO will capture more dynamic range in the clouds at the expense of the ground.
@@GordLamb That's interesting. I'll have to experiment. So lower ISOs increase dynamic range in darker areas?
ISO doesn't cause noise. Fast shutter speed and high f-stop doesn't cause noise.
The noise on a given sensor is always about the same. The goal is to drown the noise out by collecting more light. That means opening up the aperture and/or slowing the shutter speed in order to gather more light. The light you're gathering is the "signal", so you're increasing your signal to noise ratio.
Increasing the ISO does the same thing that turning up the gain on a mic will do: increase the hiss you get from that mic. It would be better to reduce the gain and move the mic closer to the subject (if possible).
@@osirismarbles5177 I think it comes down to the fact you have 256 levels of brightness. Looking at the ground and sky scene, to have shadows be above brightness 0, and white clouds below 255, all your shadow area's might only be between 0-10, and all the clouds 245-255. But if you don't mind all your sky being over exposed and 255, the shadow area's can now be anywhere from 0-150 brightness, giving you a lot more subtle details. And vice versa with the sky.
While under exposing and correcting in post certainly reveals more noise, it’s also true that the sensor’s signal to noise ratio is more favorable at lower ISO. I think he has a point about less light causing more noise, but I don’t think it’s true to say that noise is not at all related to ISO.
people who haven't watched the previous videos Tony made on ISO invariance and how ISO is fake are going to be so confused lol
Yup, my first one. Is this some inside joke?
Yeah this seems a little misconstrued. Using iso to boost your exposure in low light does increase noise esp in shadows... I don’t understand the point he’s trying to make. Like yeah it’s technically the lack of light that makes it noisy, but cleaner gain sensors don’t look as noisy in the shadows even when both are at native.
The logic of your argument is not fully correct here. Noise is intrinsic to the light itself, but the more light (exposure) you get, the less noise. So, the question is what gives more exposure. Low f-number and long exposure time (low shutter speed)! But often the problem is that the shutter speed is limited because of motion blur, and the f-number is limited because of the focus depth that we want. Now, to have less noise always the lowest ISO is the best. Because the higher the ISO, the faster we get to saturation level, thus we should end our exposure. This is unless for any of those limiting factors we can not fully expose our full dynamic range (exposing to the right of the histogram). ONLY THEN, we can increase the ISO to get actually less noise, and that is because it will reduce the part of the readout noise that lies after the ISO amplification circuitry. ISO amplification is just a like a multiplier, it does not increase the real exposure.
Shoot a photo at iso 52000! That is high ISO!
I actually just figured this out yesterday doing macro photography. It was a very cloudy and overcast day. I went outside and I was using a flash with a paper plate setup on my Canon 100mm f2.8 lens.. I was shooting at 1/200s of a second f16 with ISO 6400. Because I was using the flash to introduce extra light. When I checked my photos in Lightroom there was no extra noise. Originally I was upset I had to crank up my ISO to 6400 because I knew that the amount of noise being introduced with the aggravating. Or at least I thought so. But the end result came out phenomenal. After doing a Google search of this I could not find a definitive answer until your video Tony. Thank you again for the wonderful upload and I hope you and Chelsea have a blessed and wonderful week.
i hope that Tony and Jared can do a photo shootout collab someday. It'd be really interesting
This isn't a very productive video, for beginners watching this it will confuse and mislead them. I understand that this video is intended for pros and makes for a very clickbaitey title but I think overall it will confuse people who dont know any better. Its also very misleading since Tony uses 6400 ISO (A higher ISO) on all of the pictures and specifically chooses the under exposed images to bring back in post which will always introduce noise into the image. Bringing back exposure in post introduces noise and Tony knows this, you might know this, I know this but someone who is just beginning might now know this.
Read the pinned comment
So it's similar to gain, and noise caused by electronic (digital) amplification of sensor when there's not enough light
(Sorry but speedometer comparison was a bit confusing to me)
Yes. Exactly
Electronics inside the camera like the sensor and ADC this is what makes noise. The noise is just a constant and by making shorter exposures it will look like we have more noise (because the noise is the same all the time, though it will be a bit bigger in higher temperatures).
The “noise” Tony is referring to is his own voice. The solution is to turn off this RUclips video and watch someone who knows what they’re talking about. Hope that helps!!
Dead wrong as usual, Silver Fox. You’re raising the exposure of the ISO100 shot so much to get it to see details that are in your ISO6400 shot that it’s causing noise. You can’t push that far in post without getting noise in the shadows. That’s why you’re getting more noise in the ISO100 shot. And btw, the 6400 shot also has a lot of noise - because it’s taken at a high ISO. It looks better to you because the ISO100 shot that you DESTROYED in post by pushing it so far in the shadows has degraded. ISO is, has always been, and will always be applied gain in a DSLR or mirrorless camera. It’s applied AFTER the shot is captured to artificially apply more light to the shot. Anything done artificially in the camera (and even in your destruction of the ISO100 shot in post) to compensate for the light that should have been in the shot to begin with will degrade the image and add noise. Not opinion, it’s a fact. Maybe you are ignorant - nothing wrong with that. But you should (1) learn how a camera really works; and (2) apologize on a future video to all those you are trying to learn the same thing you should already know by now. We’re not perfect, but ou need to admit when you’ve made a mistake.
What a great concept, speedometer.
He does make some sense, I don't think his statements can be entirely dismissed. Think of bird or sports photography in daylight. We shoot with high ISO so we can get the shutter speed as low as possible. An ISO of 6400 in sunlight will definitely not show as much noise as 6400 in low light. So it is certainly acting as a speedometer in this case, directly correlating to faster shutter speed.
Spot on. Low iso's don't prevent noise in my milky way shots if I fail to gather enough light. I often need 3 or 4 minute long exposures of my foreground to prevent high noise when i bring the shadows up. Gathering light is like catching rain in a bucket; if you don't have enough you're screwed regardless of how good your sensor is.
Tony, As pro's, we are taught that "shadows create detail". That's fine in theory, but not always applicable in real world situations. For example, in my real estate photography business, I shoot a majority hand-held at a higher ISO's than I would like - this is dictated by pricing, and number of homes I shoot daily. I have found with my EOS R and 16-35L f4 IS, a comfortable ISO to be around 800. This means I am shooting a standard room 1/60 @ f8 with flash at 1/4 power (angled 45 degrees back over my head) . I am able to add enough fill to not create bounce shadows, while retaining detail shadows and not blowing out highlights (windows).
Ken is gonna love this
Interesting manipulation of information...not real, but nice.
ISO works like a booster for low light environment...it boosts everything....also the noise is provided by parts of sensor that are not excited by photons, so increase of ISO is the increase of sensitivity of this parts...
A good sensor can have reduced noise as the software can analyze and eliminate part of noise (wrong signals from photodiodes) by comparing pairs of them and reducing fake electric signals.
This is why so called "night shots with reduced noise" have smaller size, as the matrix of photodiodes will be used partially for canceling the noise induced.
when you adjust the exposure in post(shadows highlight ....), you virtually increase your iso, noise is only dependant of the iso value (in camera + post)
when you have a short stutter speed, you pull the post processing iso up, so more iso and more noise
technically in low light it's better to stay at the lowest iso where your camera is iso invariant, 400 or 800 iso most of the time, so you won't clip the highlight (in a concert for exemple) (also be carefull with the new mirrorless and banding effect)
You say "noise is only dependant of the ISO value (in camera + post)" and "dependant" is an interesting word. Noise correlates with the in-camera ISO value, but it's not a causal relationship. Adjusting exposure in post does not cause noise at all; all the noise is in the raw file at the time of capture.
"technically in low light it's better to stay at the lowest iso where your camera is iso invariant, 400 or 800 iso most of the time, so you won't clip the highlight (in a concert for exemple)"
@@TonyAndChelsea "If you move into a brighter scene (maybe they increase the room lighting) your camera will not be able to drop to a lower ISO" then you need to adjust. Most of the time I'm in auto iso, or base iso. Some rare case, I lock at 800 iso cause I don't want my camera to use 400 iso or 640 iso.
PS when I say noise I only care about final image
for the rest, It's just a matter of phrasing and yes the more light your sensor get, the less noise you image will get.
I'm really happy you posted this vid, I was starting to watch some of your material. I'm glad you stopped that quickly by revealing your level of competence.
But dont you bring up the "ISO" in post? Thats why there is always more noise....
Yeah, digital ISO is really "gain" which is what the exposure slider does. Typically, camera processors are more efficient than what your computer can do in post. The real takeaway from this is to shoot your image perfectly exposed or as bright as you can without clipping highlights (ETTR) then adjust it down in post. It is always better to shoot with a higher ISO to get the exposure where it needs to be than it is to underexpose and lift in post.
Christopher Goetting They aren‘t more efficient actually, it‘s only that the ISO control boosts the actual voltages before the Analog/Digital Conversion. As the picture on the Computer is digital, it has to boost the digital signal, which was already processed, even in RAW.
@@felixruppert4498 yeah I was simplifying
No, the noise is present in the file when it's captured. Post processing doesn't introduce noise.
@@TonyAndChelsea I feel like this way of framing the problem is overly complicated. When one brightens an image in post, that's really just an amplification of signal and noise, increasing their respective visibility to the human eye. Technically the signal to noise ratio doesn't change, so it's correct to say that the noise was always there. But brightening in post processing (or raising ISO in camera) is what makes the noise visible.
Therefore, I'd argue, a more intuitive way to frame this is to say that increasing the ISO or brightening in post increases the *visible* noise of an image.
[To clarify: This is about this particular comment thread. The overall point of the video seems valuable nonetheless.]
Easier said than done when shooting wildlife with telephoto lenses and the need to use fast shutter speeds to prevent motion blur! The light outdoors at any given place on any given day is fixed.
The beauty of larger sensors is that you can dial up the ISO more and still see less noise. ISO 6400 on my Panasonic G85 is nearly unusable. ISO 6400 on a Sony a7s II is VERY usable. And that can be crucial in a situation where you don't have much control over the light or as much flexibility to reduce shutter speeds. I have found sports and wildlife photography can really push a camera to its limits like no other type of photography.
Tony, You are Saying The Same Thing in a different WAY, Right ?
i.e.
Less Light Makes Noise !!!
Yeah, less light = more noise.
Less light is more noise is known but the misconception that one shouldn't raise the ISO in challenging situations is wrong... so the titles correct !
Less light doesn't cause noise, it causes a underexposed image or a low key image... however noise is produced when you force your sensor by increasing iso to be more receptive of the little light is there. If you had a 4 stops underexposed image and gain light by each of the settings the result would be very different. Shutter speed would cause movement, aperture would cause shallow depth of field and ISO WOULD CAUSE NOISE.
@SwitchRich he was clearing up your last sentence. But i guess you totally missed it.
@@eduardobianchi4564 No, watch another video from Tony. About ISO. If you correct underexposed RAW by 4 stops in post it's the same as raising ISO in camera before taking picture. Do You think this channel has millions of subscribers and don't know how exposure work?
Simple setup: take images with the lens cap on. Change ISO. Change Shutter speed. Change Aperture. Do the same for each setting of ISO. Look in Lightroom. Do not adjust exposure. Look at the difference. Then increase the exposure for EVERY image to the SAME increase of exposure (+5). Look at the difference. On my Canon 5D MK IV, at ISO 100, the images are black and stay black, regardless of shutter speed or aperture. At ISO 32000, the noise from the sensor can be seen in the image before exposure adjustment and is accentuated drastically when exposure is increased. This means there is, at least for Canon, a direct correlation to increased noise at higher ISO settings. (ISO100,1/60,F1.8;ISO16000,1/60,F1.8;ISO32000,1/60,F:1.8;ISO100,30",F16;etc.etc.etc.)
The point about low light is valid, but to say that ISO doesn't produce noise, at least for Canon, is pretty easy to show is not correct by doing the lens cap test.
Very interesting. Thank you for showing me how to think about ISO in a different way :)
Noise isn't caused by high iso but by the lack of light - Simon d'Entremont.
*sits and waits patiently for Ken Wheelers response 😳
Great video. But to nitpick, f/ stop and shutter speed and lack of light don’t *cause* noise. The noise is always in the sensor. What f/ stop and shutter speed and light affect is the signal to noise ratio (SNR). When you float the ISO and it reaches high levels, it’s an indicator that SNR is dropping.
The ISO in this cases is caused by the shadow lifting in your editing.
I think the correct way to compare in this case would be having the on image correctly exposed, and the other one over exposed, but correcting it with an ND so both would be at correct exposure. That would be a fair comparison
Why? The result would be identical. First decreasing the light by using an ND filter, then restoring it by lowering your shutter speed. The same amount of light hits the sensor. Are you implying that the duration of the exposure alone is a determining factor in noise? Interesting though, will test it tomorrow :)
TFM no. I’m saying that because he adjusted the dark image to be brighter, and he should have done the opposite. Adjusting the brighter image to be darker.
Because it’s obvious that by pushing the shadows it will come noise.
So I just mentioned ND’s so you are to get lower speeds without clipping and able to recover the highlights just that
@@GuillePozzi correcting with an ND reduces incoming light which is not the same as darkening the image in post. You'd reduce incoming light therefore increase noise.
@@tfm2934 I do understand, but what I'm saying is that he is matching exposure in post. Which will obviously cause noise when increasing the brightness/exposure.
So what he needs to do to correctly test this issue, is by matching exposure IN CAMERA, and the only way to do that (without altering other parameters) is by using an ND.
And I do understand what you said, but I've done this test before (properly, and no, ND does not increase noise, and yes ISO causes noise).
@@GuillePozzi I think we agree with eachother :) But I'd expect the ND filter to offset the addition of a light source so the reuslts would be identical. Same light, same ISO, same noise. Agreed?
Fair argument, except the iso isn't bound to shutter speed or f stop, you can set it manually. And regarding that, iso introduces noise!
because in video you can't change any other settings..
This is called SPS in engineering or paramater dependance in equations. velocity and acceleration are bound to each other as with the throttle/spedometer example. Tonys argument only makes sence with camera in shutter, auto iso or aperture priority
High ISO doesn't cause the noise to exist, but it does amplify it.
YES.... YES....
and YES....
Thats EXACTLY what happens. ISO is sensitivity of the image sensor to light, i.e sensitivity is increased / amplified electronically. If there is less or no light, then the output of increased sensitivity of the image processor would be noise because it has nothing to amplify.
Sadly. this video would only make sense to people who are new to photography.
Shoot something exposed correctly at base ISO and then shoot it again exposed correctly with the ISO cranked higher and there will be more noise...this is nonsense, its like saying improper shutter speed doesnt cause motion blur a fast moving object does
That's true! This is why I suggest watching the ISO like a speedometer. But the ISO is not causing the noise, the low light is. It's the difference between causation and correlation, and that's important to understand because correcting the problem cannot be done simply by changing the ISO.
@@TonyAndChelsea I have to disagree, iso is applied after the image is taken so its software induced digital noise
3:12 "change your settings to reduce your iso" then 3:50 "changing these settings doesnt change iso", does this not contradict itself?
i trust you (and have through other controversies), and have learned just about everything from yall (ive watched literally every video), this one seems wrong to me though (according to what i've learned from you). interested in seeing your response,
I'm speechless, can't digest the logic of this video, how the hell he compare black under exposed photo at 1/640 sec F2.8 ISO 6400 to slightly over exposure at 1/6 sec F2.8 ISO 6400 and then made conclusion that short shutter cause noise.
Lol wtf?
thats not what he said. You need more light to reduce noise. And you dont get more light by changing your iso.
@@Nicofoto whats wrong with that ? the only thing he changed was the shutter speed so obviously the short shutter caused the noise.
I watched your previous video about ISO long time ago, and I have to say you're right. The conclusion is, every photographer needs to get the correct amount of light before taking any photos.