TL;DR Shoot the brightest exposure you can without risking overexposure given the time and takes you can afford for the shots you are performing. More light=more data to use in post.
ETTR is useful for aps-c cameras and micro four thirds because it gives you an end result with less noise if you know what you are doing. Full frame raw’s naturally have greater dynamic range and a higher signal to noise ratio at same ISO value compared to smaller sensors, so for those the difference is marginal as demonstrated in this video. As far as clipping highlights, this is an artistic and/or practical choice, and does not necessarily follow ETTR unless the artist chooses it.
I have often argued with friends when i say that I often UNDEREXPOSED by at least 1/3 stop in certain lighting situations. In Greenport, Long Island, NY, two of us tested our ideas shooting at a passing sailboat. He showed me how nice his shots were and said mine were too dark. Then I had him zoom in to the sails. His were blown out while mine showed seams in the sails. Too dark? All you have to do in post processing is to lighten the image.
As I said, you don’t want to go too far. But just because it’s overexposed on the back of the camera doesn’t mean too much since that’s an in-camera processed jpeg. What matters is if he has detail he can recover in the raw file.
In my opinion, ETTR used to make sense when we were all shooting films. The film was so forgiving in terms of blown-out highlights, that some photographers would simply overexpose all of their images. In today's modern digital photography it makes the most sense to go for the mid-tones. Personally, When the scene is beyond my sensor's dynamic range I would rather move to the left and restore in post thus not to blow the highlights. My goal is to avoid my histogram touching any side or at least the right one.
The main take away for all photographers is that the LCD display preview of the shots you take are JPEG and have a lower dynamic range then the raw file.Thus, if you are hitting either end of the histogram and you are shooting in raw, there is no problem with recovering the data. however, it’s no problem if you’re a little bit over exposed or under exposed. The degree to which you can recover data will depend on your camera and the sensor. In all cases, if you want the highest dynamic range, you must shoot in raw.
Great video, very informative. I can agree with everything stated but want to point out that you are comparing images shot on a pretty modern camera sensor and image processor. I have been shooting night photography since 2007 and went from canon 30D to 6D. When I didn't use ETTR method on my first camera I was getting a lot of pattern noise and banding. When I finally found out about the ETTR it totally changed my world and had to shoot some images again. What I want to emphasize is that maybe it is not that relevant anymore, but it sure makes a difference when you don't have enough money to buy latest and greatest. Keep up the great videos. Cheers.
A much better tool than the histogram for setting exposure to the right is if your camera has highlight "blinkies" or "zebra stripes" on the LCD display to indicate clipping. Increase the exposure until just a few small insignificant highlight areas are blinking, no further.
I’m no pro but there are a few things he forgot to mention: 1) that modern cameras are not only quite good at recovering shadows, they are comparatively much much worse at recovering highlights. This means any attempt to use ETTr has a very high risk of permanently losing highlight detail whereas not using it carries little to no risk of losing shadow detail, and the fact that it’s a jpg you are looking at makes it even harder to make this estimation. 2) blowing out highlight is more nuanced that blowing out skis as you can blow out highlights on a per color basis for areas that are vibrant but don’t appear to be “white” on your image. The “overall” histogram combines all colors and won’t necessarily reveal the specific colors or that a specific color channel is being blown out unless you look for it. The reviewer is Correct, the technique is over rated, but I would also call it stupid, unless you not only fall into the group he refers to, and always take bracketed shots.
In film, you expose for the shadows and develop for the highlights. Digital, in my experience, is just the opposite. I normally shoot 1/3 to 2/3 stop under because like you said: shadows are easier to recover.
Thank you David and Brad for asking the question. Like you David I have tried ETTR and found it of little value. I shoot Raw and find the latitude usually adequate for a good exposure. However I do pay more attention to my histogram then I have in the past because of my awareness of ETTR. I am more careful of achieving a more balanced histogram.
The “big picture” context of digital exposure is the sensors have a fixed range which many scenes exceed. The Histogram is a tool to evaluate the scene range. For optimally exposed highlights the right edge of the histogram representing the brightest highlights needs to barely touch the right side of the graph. It is only at the point of exposure adjustment what occurs on the left side will reveal whether or not the scene range is shorter than the sensor range, matches it exactly, or exceeds it. In lighting which exceeds the sensor range (e.g. sunny cross-lighting) one must decide whether to preserve highlight detail at the expense of shadow detail or let the brightest highlights blow out to record more detail in the shadows during exposure. Because of the way our eye respond to brightness in a scene by restricting pupils to keep the highlights exposed correctly (at the expense of shadow detail awareness) we react to photos similarly, more likely to notice flat blown out highlights more than loss of detail in shadows. Perceptually we sense 3D shape in 2D images via shadows and the specular highlights on objects. If shadows are underexposed the specular highlights on them will still be visible and shape will be inferred. But if photographing a object like a wedding dress with satin and beadwork or a white car where the main clue of 3D shape come from the separation of the specular 255 highlights from the 250 solid white any overexposure will “kill” the separation of solid vs specular highlights and the perception of 3D shape. When shooting faces be aware that because of the tone of the skin the RED channel will usually clip first, even before the histogram indicates overexposure. Clipping the RED channel in skin highlights results in a yellow waxy appearance around any specular highlights on the face which can’t be remedied in post processing. The best approach is to bracket exposures whenever possible and/or err on the side of preserving the highlight detail
I agree David! However, most camera sensors are not the same size, back-lit, or same pixel count and handle light and Dynamic range completely different! For example, take the Panasonic GH5 vs the Panasonic S1 series. The GH5 I found that I always ETTL as it's sensor likes to underexpose, then brought back in the shadows for the best Image. GH5 doesn't like or do well with any overexposure. Whereas the S series is ETTR is the way to go! With it's wide dynamic range you can push it 3 stops to the right and still get the best image.
With your image, there’s a balance between better details in the shadows vs blowing out the highlights in the skies. I guess you can ETTR but protect the highlight.
Agree to an extent that this is for pixel peeping, however consider this; the difference using ETTR on a drone (most of which currently have only 1 inch sensors) is going to be far larger than using a DSLR, particularly in low light or dull conditions. I am an aerial landscape photographer and my clients purchase up to A1 size prints which are enormous, and to cater to that size with a small sensor is hard. Recently, using ETTR, there is a noticeable difference in noise and quality and the slight expense of contrast (which can be added in anyway).
If you are going to use the technique, using it in JPEG is pointless. In RAW, however it can be useful. It requires a deliberate approach not to different than what it takes to make a plate film exposure. By that I mean taking your time. If that isn’t possible, then just bracket the shot and call it a day. Modern cameras can recover about 1 stop of over exposure. The noise in the shadows may be able to be reduced, but you have to work at it.
I expose correctly most times, because highlights blow out more quickly than info is lost in shadows. And, unless shadows are brought up, noise is negligible. Shadows shouldn't be too bright anyway. That said, I do use an adaptation of Ansel Adam's zone system to expose very contrasty images. You actually are correct in saying not to go too far in exposing to the right, because blown out highlights turn a nasty gray if brought down, and there's no detail in them. Contrast can be controlled well with a perfectly exposed image with Adam's "Zone 5" as mid gray.
Adams was keen on placing the important part of the image on a particular zone, and allowed the rest to fall where they would. He corrected in developing the negative and further in dodging and burning the print. Shooting digital is much like shooting slide film, in properly exposing for the highlights. At least we can recover the shadows to a point. Like many things in the photography world, ETTR is one of many techniques that can be used to make an image.
If only stills cameras would start using waveforms. It's so much more detailed and shows exactly which spots are clipping instead of just overall exposure. In some lighting situations (low-key), histograms are almost entirely useless when 50-75% of my frame is underexposed on purpose but my histogram is all the way on the left without telling me if the skin is exposed correctly.
I too face this same dilemma when trying to expose for skin tones in particularly stylistic lighting; I don’t feel like ETTR with a histogram is very practical in these cases. Do you have any strategies or go-to workarounds for these scenarios?
Let’s say for portraits, you could use spot metering and put it on the model’s face. Of course, there’s a greater risk of clipping background highlights this way than when using multi metering. Clipping highlights is not necessarily a bad thing though, just one of many artistic choices the photographer has to make.
I've had this 'discussion' a few times. It seems modern cameras with their very good sensors render ETTR somewhat obsolete. However, what ETTR does (in simple terms) is it captures more light because you extend your shutter times as you expose to the right. For digital capture, the more light (without clipping) the better. You get a stronger electrical signal - which is how digital starts out, before ADC. This is also why we like larger sensors, they capture more light because of the area they cover.
You’re absolutely correct about the digital signal. This works with audio too. Recording the “hottest” signal you can gives you a better signal to noise ratio.
You are so involving speaking about a field that you know so much about, I am a photographer, but you are bringing me back to school man, love it, thanks. Keep it up!
As ETTR has come from the film days it has more benefit in film shooting. Film shooters don't have the flexibility to retain shadow details as much as recovering highlights details, even though it's too little amount. But modern digital camera is quite opposite because in digital camera we can retail shadow details more than highlights details. That means it's possible to recover shadow details from 3 or 4 or sometimes more stops of an underexposed image in exchange with noise. But it's impossible to to retain 3 or 4 stops of blown out highlights detail.
You can usually recover at least a stop or two at the top and bottom. But there’s still much less data in the shadows so it has potential to bring in more noise than the highlights. As I said, though, I don’t think it makes much of a difference in most real world scenarios. And as for film, it really depended what type and brand of film you were shooting. Negatives gave us much more latitude than slide film. By today’s digital standards, Kodachrome had a very small dynamic range. It looked best when underexposed a bit so you got those deep, saturated colors. But shadows were gone and that contrast - along with the warm color tones - were what gave Kodachrome it’s charm. :)
It also depends what you consider a blown out image. It can look blown out on your screen, but as long as you don't touch the right side you will be able to get those details back, no matter is it a sky or a bride's dress...
I think the key is shooting at base ISO when one wants to shoot ETTR. So much is mentioned about ETTR but they leave out the at base ISO part, otherwise ETTR is kind of over talked
I've read up on this a bit, and the consensus I have found is that IF you have shadow areas in your photo that are important to you (ie. you don't want to clip the shadows) then ETTR can be useful. But if not, then a normal exposure would be fine. As with anything in photography, it's really going to depend on what you're doing and what you want to accomplish. An example would be in a slot Canyon may be important and so in the slot Canyon, overexposing a bit (or using ETTR if there isn't a huge level of DR that exeeds say 5 stops) may be useful because if you want to capture detail in both the shadows and highlights, aside from resorting to HDR/bracketing, ETTR would be your only other option in that scenario. But on the other hand, shadows under a bridge on a lake may not be that important and using ETTR may not be as critical (or needed). I mean to a extent, more highlight data is beneficial (for shadows and mid-tones and even highlights if you aren't blowing them out) as you can always "throw away" this data later, but it's harder to get back something you don't have (lost shadows). But ETTR isn't necessarily going to be the best approach in all situations especially in high-dynamic range scenarios where you may have more than 5-stops of DR (or about +/-3 stops on either end of the scale), then bracketing may be needed.
We should all get better at shooting the proper exposure to begin with. Shooting jpeg is great to practice this, then shoot raw for jobs just in case. :) Now, I understand that the example he presented of the back-lit building. I would probably just go back to the location at a time of day when it has better lighting. There are gonna be drawbacks whether you shoot ETTR or ETTL when you are trying to pull the detail out of those strained images.
Might be semantics, but if the idea is to push to the right and then lower exposure in post, then your original capture is definitely over exposed from where you want it at the end.
I thought ETTR was only going as far right as you’re willing to go knowing you might be losing some highlights. My friends who shoot landscape and architecture will bracket but they’re always editing in post. Whereas I hardly edit in post unless there’s a paycheck attached to my efforts 😂
@@DavidBergmanPhoto Pressing the shutter button is merely a data collection exercise; ETTR done correctly maximises the amount of data and takes no longer than exposing to any other standard. Similarly, post processing is an essential part of the image making process, who doesn't want to show their pictures at their best, so optimising the exposure in post makes no longer with an ETTR image than any other. ETTR gives you a few more stops of dynamic range.
Thanks for answering this David. We were having a chat about it in Seth's Discord channel and decided to ask someone way more knowledgeable than us. Good advice! Keep it coming!
Good explanation, but link this explanation to shooting with ISOless sensors, so shooting with the base ISO of a camera and then stretching the exposure to the right in post. So, for example, the camera needs 100, f4 and iso 3200 and then you roll your iso back to base. You get a dark photo but you can then correct it in post with the exposure sliders, wouldn't you have less noise in your photo?
I think you need to experiment, partly because cameras are different. I shoot with micro four-thirds and also with a Nikon Z7. I consistently find that my Nikon camera tends to push exposures to the left, and the images can get muddy in post, even though I shoot raw, and even though the Nikon has arguably better dynamic range than my m. four-thirds cameras. When I meter the same image in my Olympus cameras, the histogram tends to be more spread out, with less tendency to bunch the pixels up on the left. The jpeg images on the view finder may look identical, but the histograms are not identical, and the raw files are not identical . So I think different cameras may require different exposure techniques. Also, using a flash may also affect things, since your viewfinder jpeg is also more problematic. In this case, using a hand-held incident meter might be a better option if practical . So your point is well taken. I don't think those who advocate always exposing to the right are correct, even though the theory seems sound. Also, as you point out, it depends on the scene. If you blow out the details on the bride's wedding dress, you are not going to have a happy bride. Protect the dress at all cost might be a better rule here. So a better rule than always exposing to the right might be, expose to the right when the equipment and the scene require it.
I have to do the complete opposite. I shoot minor hockey. The arena lighting is sufficient for the ice. I shoot manual mode. I have an entry level camera with a Nikkor 70-300nm f:/4.5-6.3 ED VR lens. So it’s variable aperture. Normal settings throws the exposure off due to reflection. So my home arena usually is ISO 3200 1/320 sec. my EV is +1.7 and my camera is set for 4.5 which changes as I zoom out. Because of the ice. If my histogram is blown too far to the right I’ll adjust my EV back to +1.3 or +1.0 which I do when I zoom out or figure out where lighting in another arena. I want to see skin tones. See faces, and proper uniform colours. I want to see the puck in the shot
You’re still technically “overexposing,” event ar +1 otherwise the ice would be grey and faces would be too dark. But of course you’re not doing it with the intent of lowering exposure in post. You just want it to look good. :)
It does look very good. I have to clarify when I adjust the EV. I lose light because of the variable aperture. My aperture drops to 5.6 around 200mm and 6.3 at 300mm. I have to pull light in from somewhere. I have to boost the iso to 6400 when I zoom out because of the loss of light. This causes the histogram and the exposure further to the right. Over exposing the image. So I drop to +1.3EV. If that doesn’t quite do it, I’ll shoot at 1/400 second instead of 1/320. It all depends. It all changes as I zoom in and out. As I said, I want to see the puck. I want details on the ice. I want proper skin tones. I want the team uniforms the right colours. I want to see expressions. I want proper whites and proper blacks I know how my equipment reacts. I really like your videos. I also learned quite a bit from Gavin Hoey Daniel Norton and Mark Wallace.
this technique is essentially outdated because of modern higher quality cameras. Blow out the highlights and they are gone, no getting them back, tons of data left in the shadows and even if there's a tiny amount of noise its easy to remove in post. Thoroughly disagree with the notion that a landscape photographer would ever want to push their exposure and blow out the highlights in any way. Details in the clouds/sky are often what makes the picture
Thanks for this video. I'm just a hobbyist photographer who shoots RAW+JPEG and normally just uses the JPEGs unless I want to edit a photo (which I normally don't want to do). I try to properly expose my images as best as I can and use the histogram to make sure I'm not clipping at either end.
Another way of saying this (ETTR) is expose for the shadows, correct? Also, my understanding is it depends on the camera- for Canon I believe I heard it’s best to ETTL and raise shadows in post, with Sony, Nikon, vice versa (I.e., ETTR). On my Fuji x100 I find it best to shoot 2/3 stop under and raise shadows in post as needed
Not sure you can say to always just "expose for the shadows." If you have a very contrasty image, if you only pay attention to your shadow exposure, then you might overexpose too much and lose your highlights. Unless you're bracketing and planning to combine in post, exposure is always a give-and-take on exposure for shadow and highlight and it could vary on each scene. And yes - different brand handle noise and over/under exposure recovery differently. Not sure I'd lump all brands into one category or another. I'd run test on your own gear - and that includes the RAW processing software you're using because they can have an impact as well.
ETTR is honestly best reserved for scenes where there's a very bright area and an extremely dark area together (in both of which you want to retain detail and care about it), because then noise and inaccurate colors are a guarantee in some of your image. If your camera doesn't offer zebras, the gamble could be really damn high. And even then, you'll have to do some tests with different zebra values to determine when and by how much you should over-expose. If you're shooting in daylight and there's a very bright sky like this but everything else is well-lit there's no point risking your sky over some barely noticeable noise. Same goes for snowy scenes or perhaps water foam etc. I don't think it's overrated, people just keep forgetting they're given tools, not magic solutions. Everything in photography should be balanced or it goes the opposite way and creates some other issue in your workflow / image. ETTR truly can save an image if you're, say, photographing a dark indoor scene with windows, it's definitely powerful knowledge, but that doesn't mean it can solve all your noise issues. It's certainly revolutionary to me because my camera is a very old and cheap canon dslr that adores noise. Learned about ETTR today actually. I never really sat down and wondered if there's a system to calculate the absolute limit of the highlights of cameras to overall reduce noise, going past warnings like zebra patterns and the like.
At 1min 50sec I think you didn't fully explain to people what the vertical axis of the histogram represents. It helps to know that the height of the graph for a particular tonal value on the horizontal axis is showing the proportion of the total number of pixels in the image that have that tonal value. Thus the area under the graph is fixed and equals the total number of pixels. It is then clearer to see how exposing so that more of the graph is to the right of the graph results in loss of information. In the extreme case where every pixel in the image has the highest value (massive overexposure), all the area under the graph is at that right hand edge. All information is clearly lost. The same is true with an extreme underexposure, everything goes off to the left hand edge of the graph and once again all information is lost.
ETTR is perfect in dark environments without much contrast. You have so much room to push before clipping (if you even have any whites at all). Ettr is useless in strong contrast or bright scenarios like a sunset or backlight tho.
Nowadays you expose to the right using zebras, not the histogram. As Mark Galer suggested for Sony Cameras, you put them at +109 in order not to clipp highlights. This makes ETTR easy to implement without the risks that David mentioned. Although it's true that cameras are better today recovering shadows, the science behind ETTR is sound and undisputed: with more light you have more data to work with. Since digital sensor are inherently linear, it's only logical to capture more data by overexposing your image. Besides, if you open a Raw file with a dedicated software like RawDigger, you'll realize that it's always underexposed at the suggested camera meter.
In principle I agree with the use of ETTR and often use the technique myself. The only issue I have with it is exactly what you described here. It's ideal if you have plenty of time to set your exposure just a tad over but in such circumstances I'd prefer to bracket my shots and combine them in post for greater dynamic range anyway. As you alluded to, in event situations it's much easier to screw up attempts at ETTR for the reason that your light is changing constantly. If you get ETTR wrong you will get unrecoverable highlight information that is lost. I tend to slightly underexpose in run and gun situations simply because even if I do have to raise shadows in post it's way easier to just use noise reduction to help improve the image quality. There is no such option for highlights, once they are gone, they are gone for good! So whilst ETTR is a viable and useful technique I think for most, the ability to combine multiple exposures in post rather than trying to achieve maximum DR via one single image has rendered it almost obsolete.
That's always going to be the case, there's always going to be tricky lighting that requires full manual exposure control. True exposure to the right has never meant that you'd expose so bright that you'd be blowing a bunch of highlights and areas. You might blow some minor highlights if there's unimportant detail, like glare on a car hood, but if it's significant, then you're not exposing to the right, you're just overexposing and hoping that you can save it in post. Legitimate expose to the right photos aren't going to have a significant spike to the histogram on the right side, if you do, then you've over-exposed.
An alternative is to use extend low ISO. For instance, some canon cameras have base ISO of 100, but you can extend it to 50. It's basically the same as shooting in ISO 100, overexpose one stop, and in post process reduce one stop.
I just go as far to the right where I’m comfortable with what I might be losing. But from what I’ve learned on Fujifilm, the raw file most of the time doesn’t have blown out highlights when I do that, and Fuji raw files tend to be at least a stop under what the jpg is... When I’m doing paid work, I’m more careful with what I do in camera because I have to edit the raw. But when it’s shooting just for me, I care how the jpg looks and still do raw + jpg just in case there’s an image I really really want to take into post.
One thing to remember, the metering area! Most shoot wide metering but there will be times of spot (maybe for a bright spot). Also many will move the spot around to a dark area then a light area, this is how the ETTR can be seen. For those with zebras active you adjust the exposure dial neg till they go away if too bright you need to bracket, say for a sunrise with the sun visible. For night shots in manual you have M.M. at the bottom, use Full Frame metering and if a city or town watch the zebras but for those dark places say for Milky Ways with f/wide and SS set so no star tailing adjust ISO for M.M. of .7 will be very close to 3/4 to the right. With sensors today shadows with detail easy to bring out BUT those Highlights can never be. Night shots in the city watch the zebras but stay above ISO 800 noise worse below. Also try ISO Auto yes can get high and maybe noisy (not so much on ISO invariant sensors) but as of 2020 Lr and other Post Programs will clear up without softening.
I like to let the highlights go above a little when shooting in like golden hour and the light is shining behind the subject but usually under expose. But recently learned about HDR and it’s a pretty unique tool to use to get best of everything.
If you've got a static scene to shoot and you've got a tripod or reasonably stable hands, that can work. But, it all depends on the software or your skill in aligning things to get a good result.
If you expose for middle grey, then everything will be fine for most photos. If you have too much white like in weddings and snow, then change exposure to favor highlights. If photographing a dark subject, then exposure to favor details in the dark areas. If subject is not moving and you are using a tripod, you can take the shot at multiple exposures and combine the images to get what you want. Most non photographers do not pay attention to shadow details or even slightly blown highlights, but care more about the overall picture. The quality of the light and color is most important as well as the composition. Some of the most famous photos historically have poor exposure.
Sometimes if you’re over-zealous pushing the highlights when you try recovering in post processing you find the though you have tone and texture you’ve lost true color.
I wouldn't argue ETTR on a sensor that is not ISO invariant such as many Canon sensors (with your guidance about not going overboard as a caveat). Seems a lot of Canon folks are the ones that pitch ETTR. But once you bring in an ISO invariant sensor (for example - Sony and Fuji) you won't see a difference between setting the exposure at the base ISO and jacking the exposure up versus setting it at the ISO you need for the correct exposure at a given shutter/aperture combo. Sony/Nikon/Fuji/Pentax users - give it a try. Put your camera on a tripod and manually focus on a subject. Set your camera to an ISO that is 3 stops above base ISO for your camera and adjust your shutter speed and aperture to get a correct exposure. RAW images only. No JPG and no picture effects. Take the shot. Change only your ISO to the camera's base ISO leaving the shutter speed and aperture the same and retake. Bring both images into your favorite editor and do some side by side pixel peeping. What do you see? Why do it? Well - when you underexpose the shadows you also underexpose the highlights and end up capturing some amazing color and shadow detail in the highlight sections of the photo that are much easier to work with than an ETTR exposed image. And you will have the same level of noise in the dark areas with both settings. Cool! Should you do it for every picture. Good lord no. But images where you have a looming sunrise or perhaps some streetlights in an otherwise dark street can give you some really nice results. Part of the fun of this hobby is experimenting and this is a good one to understand. All this said - your points about getting the exposure correct for once in a lifetime moments and if you are going to do ETTR not going overboard are spot on. As a wildlife photographer ETTR gets pitched like it is the only way to do things and I have seen the disaster that results when the background and subject are dramatically different in terms of exposure values and people push to the right and then find they can't get back fine shadow details on white birds. Not saying to underexpose in these situations (absolutely do not do so) - but keep as much detail as you can in the center of the histogram. Really appreciate your videos David. Well thought out and presented. Looking forward to more.
MovieBuff Texas, ETTR has nothing to do with ISO invariance. The point of ETTR is to keep the ISO constant, usually at base, while increase the exposure (the amount of light hitting the sensor) as much as possible without saturating important highlight details. More light will benefit any camera, regardless if it has ISO invariant sensor or not.
MovieBuff Texas, ETTR has nothing to do with ISO invariance. The point of ETTR is to keep the ISO constant, usually at base, while increase the exposure (the amount of light hitting the sensor) as much as possible without saturating important highlight details. More light will benefit any camera, regardless if it has ISO invariant sensor or not.
So am I correct that if you were to expose at night a cowboys closeup face dimly light by a campfire way out of frame you would ETR to get detail in the mid-tones assuming that high-tones are nonexistent? If darkening is preferred it can be adjusted in post.
A very good video with arguably poor and even wrong explanations. You made a case where the exposure time on one image was 4 times as that of the other and hence the noise is obviously lower because it received 4 times as light. What you should, in principle, do, is to set exposure and aperture equal, and capture at different ISOs. That way the only difference would be how well the camera handles A2D (Analogue to digital) conversion. The noise would be the same. And in case you're saying oh no, higher ISO makes the image grainier you should go ahead and watch Tony and Chelsea's video on ISO. ETTR and ETTL actually depends on the camera you use... For Sony, you'd definitely better expose to the left.
A fair point, but In the real world, I’d still want to keep my ISO as low as possible and would rather slow my shutter speed (on a tripod, of course). Lots of controversy on the Northrup video and many disagree with his thoughts on base ISO.
@@DavidBergmanPhoto Correct. Once you expose a picture CORRECTLY, ISO measures the noise. Otherwise it doesn't. I think that's the controversy behind Northrup video. But in this video, he's not doing this. So, the Northrup argument applies here. On my Sony cameras, I have found that exposing to the left gives me better results. This is especially the case if I'm shooting Jpeg. For Raw, I think there might be a case where ETTR is good, but I should do some more experiments.
Changing the ISO is changing the gain not the exposure. The sensor behaves the same all the time if you change the gain, you are just amplifying the output of the sensor. The best way to do the comparison is by under exposing and over exposing to increase and decrease the amount of light per photo-site. ETTR is about exposure after all.
Thanks for your video. But I still don't see the point of histograms. Why look at an abstract histogram when I can just see the photo itself? I can see the exposure via the camera. Using the histogram is one step removed from looking at the screen or view finder. I don't see their point. It's like looking at the tubes of paint that painters use instead of looking at what the painter painted.
Sometimes your screen or viewfinder can be too bright or too dark. Also Sony mirrorless camera screens are the worse they can overexpose or clip highlights on screen. many camera screens or viewfinder cannot be fully trusted. Histogram gives you a perfect range of your jpg image exposure, so you know in raw you have alittle more room.
Around 4', "the fact that it's linear... it means it's exponential" 😀. Jokes aside, nice video and a lot of passion in the way you explain things. Thanks !
I think I got what he want to mean. It´s that for instance - for a dark area you have kinda 0-30 brightness - and a light one 180-255 - So, multiply by 3 - RGB - It means that enven in a linear gather, you would have much more information as you go to the right. :) Maybe a simple term as - the amount of data (details) - grows as you go to the brighter side - So, more information to the software to render the image. :D Luke, I´m your father!
Lots to learn David!! thanks a lot. I have one question that I hope you can help me with in the future! :) I recently bought a Lap Top that runs windows and want to set up Capture ONE PRO 20 to work with Capture Pilot (I have an Iphone XR). I have tried to do it but has not worked... it would be amazing if you can help me with this!! Thanks in advance!
ETTR is often for video were videographers aren’t recording raw an possibly are shooting raw, A lot of raw curves (e.g. Panasonic’s V-Log) allow for much more detail in the upper range of the curve - use don’t exceed te maximum range in the highlights. Also, for photography, adjusting in post to raise the brightness only works for cameras with ISO-invariant sensors
Maybe it’s semantics, and it doesn’t mean exposure so far that you blow out / lose highlights. But the whole point is to expose a bit brighter than you want your final image. To me, that is overexposing.
It’s yet another fad. There will be a lot of people doing it because an influencer said it’s a good idea. As you have shown in the majority of instances it adds nothing. Generally aim not to burn out highlights and create darkness in post. Before long manufacturers will be adding an ETTR mode on the set up menu and, people will bore on about how a shot was ruined because the photographer should have used ETTR. Good cameras have huge dynamic range and this renders ETTR pointless as a concept. Maybe the micro 4/3 shooters can occasionally get a little more mileage but, post production diminishes the majority of problems in exposure ever more effectively.
What you said is true with one exception. Budget shooters. The way m4/3 sensors curves are you get way better texture and details with ETTR 1 stop if you can afford it in your circumstances. Lets say that stops quality is the best for the first Native ISO MG set by developers, lets say for ISO 400 and WB 5600 for BMPCC4k. Thats the mode you shoot on most if the time. Exposing for ISO 200 get you better results, but you risk blowing out highlights which isnt worth it in any digital camera. Budget cameras butcher contrast and hues when underexposing, so you can get -1 stop at best realistically. I have external plugins for denoising made for porfessionals, but they cant salvage bad image texture. Its not grainy, more like muddy.
About ETTR being over rated -- I could not agree more. Thank you for lending your considerable expertise to what I have already known to be unnecessary work.
off topic, recently bought a manual macro lens, the laowa 65mm, love the magnification but want even more, would you recommend putting macro extension tube on top for more magnification or one of those raynox front lens attachment?
So in terms of noise is it better to shoot an image at a lower iso and have to raise the exposure in post or get as close to perfect exposure even if it means shooting at a much higher iso?
Really depends on the camera as some of them handle higher ISO much better than others. With my Canon 1dx3, I won’t hesitate to push the ISO to 8000 or more. But I wouldn’t consider than with a camera from 10 years ago...
S Burgos, each camera has an ISO value after which it becomes ISO invariant, so further increasing ISO no longer improves the noise compared to just shooting at a lower ISO, with the same shutter speed and aperture, and brightening the image in post. In fact, if we just look at the image quality and ignore the inconvenience of post processing, it is better to never increase ISO value past that point, because you are just clipping the highlights, for no benefit. At what ISO value camera becomes ISO invariant is highly dependent on the camera model.
S Burgos, each camera has an ISO value after which it becomes ISO invariant, so further increasing ISO no longer improves the noise compared to just shooting at a lower ISO, with the same shutter speed and aperture, and brightening the image in post. In fact, if we just look at the image quality and ignore the inconvenience of post processing, it is better to never increase ISO value past that point, because you are just clipping the highlights, for no benefit. At what ISO value camera becomes ISO invariant is highly dependent on the camera model.
Thanks. ETTR, just like Bokeh, is just a buzz word / phrases I hear from people new (newer) to photography, as if they are the panacea to bad images. You are spot on about not being able to see real differences when comparing correctly exposed images, using different settings. Anytime I am asked, or read a post about any buzz work or phrase in photography, I ask the questioner if the panacea they are discussing is more about a short-cut to getting it right in camera, or the belief that post-production with save the image.
Overexposed isn't the right word. Overexposed means blown. To me, ETTR just means to the right hand side of the range that is properly exposed. If you start blowing the highlights, it's not ETTR, it's overexposed.
If you really want to push your exposure as much to the right as possible without clipping, search for UniWB. By setting a custom white balance you get a more „accurate“ histogram. The downside is that this white balance will make your preview look almost completely green.
ETTR eas probably more important before. I think that there was a significant change from the 7D2 onwards. For all I know the change might be in DPP. Yes, I'm stone age!!!
Very helpful, and great explanation. Thank you. When you speak of the histogram being affected by picture profiles, are you referring to the histogram readings during image playback, or do the picture profiles and settings affect the histogram while shooting also? Thanks.
I “lean to the right” with my exposures... Not under, not WAY over...but the histogram leans slightly (2/3-1 stop over)... Histogram is representation of computer data... David seems to be about the ONLY professional shooter understands that a camera is a very specialized one purpose computer!!!...🤓 It needs data to function correctly, but giving it too much corrupts what data is there... “Lean to the right”
When I've produced coffee table books, the CMYK conversions are done as the very last step and and then carefully proofed and adjusted (by the printer) to match my originals.
It’s a pleasure to listen to a well informed individual. The subject matter is enhanced by the gentleman’s commutative skills.
Excellent
TL;DR Shoot the brightest exposure you can without risking overexposure given the time and takes you can afford for the shots you are performing. More light=more data to use in post.
Perfectly summarized! Maybe my video should have been shorter. :)
ETTR is useful for aps-c cameras and micro four thirds because it gives you an end result with less noise if you know what you are doing. Full frame raw’s naturally have greater dynamic range and a higher signal to noise ratio at same ISO value compared to smaller sensors, so for those the difference is marginal as demonstrated in this video. As far as clipping highlights, this is an artistic and/or practical choice, and does not necessarily follow ETTR unless the artist chooses it.
I have often argued with friends when i say that I often UNDEREXPOSED by at least 1/3 stop in certain lighting situations. In Greenport, Long Island, NY, two of us tested our ideas shooting at a passing sailboat. He showed me how nice his shots were and said mine were too dark. Then I had him zoom in to the sails. His were blown out while mine showed seams in the sails. Too dark? All you have to do in post processing is to lighten the image.
As I said, you don’t want to go too far. But just because it’s overexposed on the back of the camera doesn’t mean too much since that’s an in-camera processed jpeg. What matters is if he has detail he can recover in the raw file.
In my opinion, ETTR used to make sense when we were all shooting films. The film was so forgiving in terms of blown-out highlights, that some photographers would simply overexpose all of their images. In today's modern digital photography it makes the most sense to go for the mid-tones. Personally, When the scene is beyond my sensor's dynamic range I would rather move to the left and restore in post thus not to blow the highlights. My goal is to avoid my histogram touching any side or at least the right one.
The main take away for all photographers is that the LCD display preview of the shots you take are JPEG and have a lower dynamic range then the raw file.Thus, if you are hitting either end of the histogram and you are shooting in raw, there is no problem with recovering the data. however, it’s no problem if you’re a little bit over exposed or under exposed. The degree to which you can recover data will depend on your camera and the sensor. In all cases, if you want the highest dynamic range, you must shoot in raw.
Well said!
I love listening to and viewing your tutorials. Great info presently clearly and fully, by a very likeable guy.
Exposure bracket when needed. Problem solved....and you have many options in post for ETTR and HDR if needed.
IllumiGraphic Particularly when you're shooting a landscape with a tripod.
Great video, very informative. I can agree with everything stated but want to point out that you are comparing images shot on a pretty modern camera sensor and image processor. I have been shooting night photography since 2007 and went from canon 30D to 6D. When I didn't use ETTR method on my first camera I was getting a lot of pattern noise and banding. When I finally found out about the ETTR it totally changed my world and had to shoot some images again. What I want to emphasize is that maybe it is not that relevant anymore, but it sure makes a difference when you don't have enough money to buy latest and greatest. Keep up the great videos. Cheers.
A much better tool than the histogram for setting exposure to the right is if your camera has highlight "blinkies" or "zebra stripes" on the LCD display to indicate clipping. Increase the exposure until just a few small insignificant highlight areas are blinking, no further.
5:21 - Hollywood's technique, protecting highlights, except they use on-set lights to boost the shadows, rather than doing it in post.
he doing street shots without lights
I’m no pro but there are a few things he forgot to mention: 1) that modern cameras are not only quite good at recovering shadows, they are comparatively much much worse at recovering highlights. This means any attempt to use ETTr has a very high risk of permanently losing highlight detail whereas not using it carries little to no risk of losing shadow detail, and the fact that it’s a jpg you are looking at makes it even harder to make this estimation. 2) blowing out highlight is more nuanced that blowing out skis as you can blow out highlights on a per color basis for areas that are vibrant but don’t appear to be “white” on your image. The “overall” histogram combines all colors and won’t necessarily reveal the specific colors or that a specific color channel is being blown out unless you look for it. The reviewer is Correct, the technique is over rated, but I would also call it stupid, unless you not only fall into the group he refers to, and always take bracketed shots.
In film, you expose for the shadows and develop for the highlights. Digital, in my experience, is just the opposite. I normally shoot 1/3 to 2/3 stop under because like you said: shadows are easier to recover.
Thank you David and Brad for asking the question. Like you David I have tried ETTR and found it of little value. I shoot Raw and find the latitude usually adequate for a good exposure. However I do pay more attention to my histogram then I have in the past because of my awareness of ETTR. I am more careful of achieving a more balanced histogram.
For me where I shot older equipment, it is still relevant to ETTR
It depends on what the focus is and which camera you're using. Generally, with Sony, it's easier to recover highlights, so I tend to underexpose.
It can be more beneficial with micro 4/3rds as you are dealing with more noise in the darker areas of the image than a full frame.
The “big picture” context of digital exposure is the sensors have a fixed range which many scenes exceed. The Histogram is a tool to evaluate the scene range. For optimally exposed highlights the right edge of the histogram representing the brightest highlights needs to barely touch the right side of the graph. It is only at the point of exposure adjustment what occurs on the left side will reveal whether or not the scene range is shorter than the sensor range, matches it exactly, or exceeds it.
In lighting which exceeds the sensor range (e.g. sunny cross-lighting) one must decide whether to preserve highlight detail at the expense of shadow detail or let the brightest highlights blow out to record more detail in the shadows during exposure.
Because of the way our eye respond to brightness in a scene by restricting pupils to keep the highlights exposed correctly (at the expense of shadow detail awareness) we react to photos similarly, more likely to notice flat blown out highlights more than loss of detail in shadows.
Perceptually we sense 3D shape in 2D images via shadows and the specular highlights on objects. If shadows are underexposed the specular highlights on them will still be visible and shape will be inferred. But if photographing a object like a wedding dress with satin and beadwork or a white car where the main clue of 3D shape come from the separation of the specular 255 highlights from the 250 solid white any overexposure will “kill” the separation of solid vs specular highlights and the perception of 3D shape.
When shooting faces be aware that because of the tone of the skin the RED channel will usually clip first, even before the histogram indicates overexposure. Clipping the RED channel in skin highlights results in a yellow waxy appearance around any specular highlights on the face which can’t be remedied in post processing.
The best approach is to bracket exposures whenever possible and/or err on the side of preserving the highlight detail
I agree David! However, most camera sensors are not the same size, back-lit, or same pixel count and handle light and Dynamic range completely different! For example, take the Panasonic GH5 vs the Panasonic S1 series. The GH5 I found that I always ETTL as it's sensor likes to underexpose, then brought back in the shadows for the best Image. GH5 doesn't like or do well with any overexposure. Whereas the S series is ETTR is the way to go! With it's wide dynamic range you can push it 3 stops to the right and still get the best image.
Very true. Lesson there is to test and learn your own equipment as you’ve done.
The histogram you see in-camera is of the jpeg. That is so important it should flash on screen when you are turning it on!
Good info David. For me I try to get it right in camera I use filters if needed I do not like to be on the computer for hours.
With your image, there’s a balance between better details in the shadows vs blowing out the highlights in the skies. I guess you can ETTR but protect the highlight.
I am reminded of learning sunset and moon Photography.
Agree to an extent that this is for pixel peeping, however consider this; the difference using ETTR on a drone (most of which currently have only 1 inch sensors) is going to be far larger than using a DSLR, particularly in low light or dull conditions. I am an aerial landscape photographer and my clients purchase up to A1 size prints which are enormous, and to cater to that size with a small sensor is hard. Recently, using ETTR, there is a noticeable difference in noise and quality and the slight expense of contrast (which can be added in anyway).
If you are going to use the technique, using it in JPEG is pointless. In RAW, however it can be useful. It requires a deliberate approach not to different than what it takes to make a plate film exposure. By that I mean taking your time. If that isn’t possible, then just bracket the shot and call it a day. Modern cameras can recover about 1 stop of over exposure. The noise in the shadows may be able to be reduced, but you have to work at it.
I expose correctly most times, because highlights blow out more quickly than info is lost in shadows. And, unless shadows are brought up, noise is negligible. Shadows shouldn't be too bright anyway. That said, I do use an adaptation of Ansel Adam's zone system to expose very contrasty images. You actually are correct in saying not to go too far in exposing to the right, because blown out highlights turn a nasty gray if brought down, and there's no detail in them. Contrast can be controlled well with a perfectly exposed image with Adam's "Zone 5" as mid gray.
Adams was keen on placing the important part of the image on a particular zone, and allowed the rest to fall where they would. He corrected in developing the negative and further in dodging and burning the print. Shooting digital is much like shooting slide film, in properly exposing for the highlights. At least we can recover the shadows to a point. Like many things in the photography world, ETTR is one of many techniques that can be used to make an image.
If only stills cameras would start using waveforms. It's so much more detailed and shows exactly which spots are clipping instead of just overall exposure.
In some lighting situations (low-key), histograms are almost entirely useless when 50-75% of my frame is underexposed on purpose but my histogram is all the way on the left without telling me if the skin is exposed correctly.
I too face this same dilemma when trying to expose for skin tones in particularly stylistic lighting; I don’t feel like ETTR with a histogram is very practical in these cases. Do you have any strategies or go-to workarounds for these scenarios?
Let’s say for portraits, you could use spot metering and put it on the model’s face. Of course, there’s a greater risk of clipping background highlights this way than when using multi metering. Clipping highlights is not necessarily a bad thing though, just one of many artistic choices the photographer has to make.
Couldn’t you use spot metering and use on your focus point?
I've had this 'discussion' a few times. It seems modern cameras with their very good sensors render ETTR somewhat obsolete. However, what ETTR does (in simple terms) is it captures more light because you extend your shutter times as you expose to the right. For digital capture, the more light (without clipping) the better. You get a stronger electrical signal - which is how digital starts out, before ADC.
This is also why we like larger sensors, they capture more light because of the area they cover.
You’re absolutely correct about the digital signal. This works with audio too. Recording the “hottest” signal you can gives you a better signal to noise ratio.
@@DavidBergmanPhoto indeed.. I used to be a studio engineer
You are so involving speaking about a field that you know so much about, I am a photographer, but you are bringing me back to school man, love it, thanks. Keep it up!
Thanks!
As ETTR has come from the film days it has more benefit in film shooting. Film shooters don't have the flexibility to retain shadow details as much as recovering highlights details, even though it's too little amount. But modern digital camera is quite opposite because in digital camera we can retail shadow details more than highlights details. That means it's possible to recover shadow details from 3 or 4 or sometimes more stops of an underexposed image in exchange with noise. But it's impossible to to retain 3 or 4 stops of blown out highlights detail.
You can usually recover at least a stop or two at the top and bottom. But there’s still much less data in the shadows so it has potential to bring in more noise than the highlights. As I said, though, I don’t think it makes much of a difference in most real world scenarios.
And as for film, it really depended what type and brand of film you were shooting. Negatives gave us much more latitude than slide film. By today’s digital standards, Kodachrome had a very small dynamic range. It looked best when underexposed a bit so you got those deep, saturated colors. But shadows were gone and that contrast - along with the warm color tones - were what gave Kodachrome it’s charm. :)
It also depends what you consider a blown out image. It can look blown out on your screen, but as long as you don't touch the right side you will be able to get those details back, no matter is it a sky or a bride's dress...
I think the key is shooting at base ISO when one wants to shoot ETTR. So much is mentioned about ETTR but they leave out the at base ISO part, otherwise ETTR is kind of over talked
Great explanation, David. I understood the general concept before watching your video but your explanation helped fill in some technical gaps.
I've read up on this a bit, and the consensus I have found is that IF you have shadow areas in your photo that are important to you (ie. you don't want to clip the shadows) then ETTR can be useful. But if not, then a normal exposure would be fine. As with anything in photography, it's really going to depend on what you're doing and what you want to accomplish. An example would be in a slot Canyon may be important and so in the slot Canyon, overexposing a bit (or using ETTR if there isn't a huge level of DR that exeeds say 5 stops) may be useful because if you want to capture detail in both the shadows and highlights, aside from resorting to HDR/bracketing, ETTR would be your only other option in that scenario. But on the other hand, shadows under a bridge on a lake may not be that important and using ETTR may not be as critical (or needed). I mean to a extent, more highlight data is beneficial (for shadows and mid-tones and even highlights if you aren't blowing them out) as you can always "throw away" this data later, but it's harder to get back something you don't have (lost shadows). But ETTR isn't necessarily going to be the best approach in all situations especially in high-dynamic range scenarios where you may have more than 5-stops of DR (or about +/-3 stops on either end of the scale), then bracketing may be needed.
We should all get better at shooting the proper exposure to begin with. Shooting jpeg is great to practice this, then shoot raw for jobs just in case. :) Now, I understand that the example he presented of the back-lit building. I would probably just go back to the location at a time of day when it has better lighting. There are gonna be drawbacks whether you shoot ETTR or ETTL when you are trying to pull the detail out of those strained images.
ETR does not equal over expose!
Might be semantics, but if the idea is to push to the right and then lower exposure in post, then your original capture is definitely over exposed from where you want it at the end.
I thought ETTR was only going as far right as you’re willing to go knowing you might be losing some highlights. My friends who shoot landscape and architecture will bracket but they’re always editing in post. Whereas I hardly edit in post unless there’s a paycheck attached to my efforts 😂
Shows how much I know about camera lingo lol.
@@DavidBergmanPhoto Pressing the shutter button is merely a data collection exercise; ETTR done correctly maximises the amount of data and takes no longer than exposing to any other standard. Similarly, post processing is an essential part of the image making process, who doesn't want to show their pictures at their best, so optimising the exposure in post makes no longer with an ETTR image than any other. ETTR gives you a few more stops of dynamic range.
Thanks for answering this David. We were having a chat about it in Seth's Discord channel and decided to ask someone way more knowledgeable than us. Good advice! Keep it coming!
Thanks and keep those questions coming!
Good explanation, but link this explanation to shooting with ISOless sensors, so shooting with the base ISO of a camera and then stretching the exposure to the right in post. So, for example, the camera needs 100, f4 and iso 3200 and then you roll your iso back to base. You get a dark photo but you can then correct it in post with the exposure sliders, wouldn't you have less noise in your photo?
Thank you for another excellent video! Very grateful.
You’re welcome! Thanks for watching!
I use the ettr all the time. It's so easy to postproduce it... but I don't go over....
Yes I typically only go a third over
I think you need to experiment, partly because cameras are different. I shoot with micro four-thirds and also with a Nikon Z7. I consistently find that my Nikon camera tends to push exposures to the left, and the images can get muddy in post, even though I shoot raw, and even though the Nikon has arguably better dynamic range than my m. four-thirds cameras. When I meter the same image in my Olympus cameras, the histogram tends to be more spread out, with less tendency to bunch the pixels up on the left. The jpeg images on the view finder may look identical, but the histograms are not identical, and the raw files are not identical . So I think different cameras may require different exposure techniques. Also, using a flash may also affect things, since your viewfinder jpeg is also more problematic. In this case, using a hand-held incident meter might be a better option if practical . So your point is well taken. I don't think those who advocate always exposing to the right are correct, even though the theory seems sound. Also, as you point out, it depends on the scene. If you blow out the details on the bride's wedding dress, you are not going to have a happy bride. Protect the dress at all cost might be a better rule here. So a better rule than always exposing to the right might be, expose to the right when the equipment and the scene require it.
On a full frame camera it's not such an issue.
Smaller sensors, like your phone really benefit from ETTR.
Certainly some sensors hold shadow detail better then others.
Loving your videos by the way. I'm learning so much.
Get the exposure right in the camera. YES!
Great! You have a way of putting peoples minds at rest, thanks 🇬🇧
I have to do the complete opposite. I shoot minor hockey. The arena lighting is sufficient for the ice. I shoot manual mode. I have an entry level camera with a Nikkor 70-300nm f:/4.5-6.3 ED VR lens. So it’s variable aperture. Normal settings throws the exposure off due to reflection. So my home arena usually is ISO 3200 1/320 sec. my EV is +1.7 and my camera is set for 4.5 which changes as I zoom out. Because of the ice. If my histogram is blown too far to the right I’ll adjust my EV back to +1.3 or +1.0 which I do when I zoom out or figure out where lighting in another arena. I want to see skin tones. See faces, and proper uniform colours. I want to see the puck in the shot
Andrew Riddell same as overexposing when shooting in white snow.
You’re still technically “overexposing,” event ar +1 otherwise the ice would be grey and faces would be too dark. But of course you’re not doing it with the intent of lowering exposure in post. You just want it to look good. :)
It does look very good. I have to clarify when I adjust the EV. I lose light because of the variable aperture. My aperture drops to 5.6 around 200mm and 6.3 at 300mm. I have to pull light in from somewhere. I have to boost the iso to 6400 when I zoom out because of the loss of light. This causes the histogram and the exposure further to the right. Over exposing the image. So I drop to +1.3EV. If that doesn’t quite do it, I’ll shoot at 1/400 second instead of 1/320. It all depends. It all changes as I zoom in and out.
As I said, I want to see the puck. I want details on the ice. I want proper skin tones. I want the team uniforms the right colours. I want to see expressions. I want proper whites and proper blacks
I know how my equipment reacts.
I really like your videos. I also learned quite a bit from Gavin Hoey Daniel Norton and Mark Wallace.
this technique is essentially outdated because of modern higher quality cameras. Blow out the highlights and they are gone, no getting them back, tons of data left in the shadows and even if there's a tiny amount of noise its easy to remove in post. Thoroughly disagree with the notion that a landscape photographer would ever want to push their exposure and blow out the highlights in any way. Details in the clouds/sky are often what makes the picture
Again - you don’t want to go too far when you lose detail in the raw file. But a little extra signal to noise won’t hurt.
Thanks for this video. I'm just a hobbyist photographer who shoots RAW+JPEG and normally just uses the JPEGs unless I want to edit a photo (which I normally don't want to do). I try to properly expose my images as best as I can and use the histogram to make sure I'm not clipping at either end.
That’s usually your best bet for exposure. Although I’d still prefer to use raw files for everything. :)
Another way of saying this (ETTR) is expose for the shadows, correct?
Also, my understanding is it depends on the camera- for Canon I believe I heard it’s best to ETTL and raise shadows in post, with Sony, Nikon, vice versa (I.e., ETTR).
On my Fuji x100 I find it best to shoot 2/3 stop under and raise shadows in post as needed
Not sure you can say to always just "expose for the shadows." If you have a very contrasty image, if you only pay attention to your shadow exposure, then you might overexpose too much and lose your highlights. Unless you're bracketing and planning to combine in post, exposure is always a give-and-take on exposure for shadow and highlight and it could vary on each scene.
And yes - different brand handle noise and over/under exposure recovery differently. Not sure I'd lump all brands into one category or another. I'd run test on your own gear - and that includes the RAW processing software you're using because they can have an impact as well.
ETTR is honestly best reserved for scenes where there's a very bright area and an extremely dark area together (in both of which you want to retain detail and care about it), because then noise and inaccurate colors are a guarantee in some of your image. If your camera doesn't offer zebras, the gamble could be really damn high. And even then, you'll have to do some tests with different zebra values to determine when and by how much you should over-expose.
If you're shooting in daylight and there's a very bright sky like this but everything else is well-lit there's no point risking your sky over some barely noticeable noise. Same goes for snowy scenes or perhaps water foam etc.
I don't think it's overrated, people just keep forgetting they're given tools, not magic solutions. Everything in photography should be balanced or it goes the opposite way and creates some other issue in your workflow / image. ETTR truly can save an image if you're, say, photographing a dark indoor scene with windows, it's definitely powerful knowledge, but that doesn't mean it can solve all your noise issues. It's certainly revolutionary to me because my camera is a very old and cheap canon dslr that adores noise. Learned about ETTR today actually. I never really sat down and wondered if there's a system to calculate the absolute limit of the highlights of cameras to overall reduce noise, going past warnings like zebra patterns and the like.
At 1min 50sec I think you didn't fully explain to people what the vertical axis of the histogram represents. It helps to know that the height of the graph for a particular tonal value on the horizontal axis is showing the proportion of the total number of pixels in the image that have that tonal value. Thus the area under the graph is fixed and equals the total number of pixels. It is then clearer to see how exposing so that more of the graph is to the right of the graph results in loss of information. In the extreme case where every pixel in the image has the highest value (massive overexposure), all the area under the graph is at that right hand edge. All information is clearly lost. The same is true with an extreme underexposure, everything goes off to the left hand edge of the graph and once again all information is lost.
Great video! Learning so much from you. 🙏🏻
ETTR is perfect in dark environments without much contrast. You have so much room to push before clipping (if you even have any whites at all). Ettr is useless in strong contrast or bright scenarios like a sunset or backlight tho.
ETR is almost essential if we are not shooting RAW.
I often stack in HDR mode use luminance to combine.
Nowadays you expose to the right using zebras, not the histogram. As Mark Galer suggested for Sony Cameras, you put them at +109 in order not to clipp highlights. This makes ETTR easy to implement without the risks that David mentioned.
Although it's true that cameras are better today recovering shadows, the science behind ETTR is sound and undisputed: with more light you have more data to work with. Since digital sensor are inherently linear, it's only logical to capture more data by overexposing your image. Besides, if you open a Raw file with a dedicated software like RawDigger, you'll realize that it's always underexposed at the suggested camera meter.
In principle I agree with the use of ETTR and often use the technique myself. The only issue I have with it is exactly what you described here. It's ideal if you have plenty of time to set your exposure just a tad over but in such circumstances I'd prefer to bracket my shots and combine them in post for greater dynamic range anyway. As you alluded to, in event situations it's much easier to screw up attempts at ETTR for the reason that your light is changing constantly. If you get ETTR wrong you will get unrecoverable highlight information that is lost. I tend to slightly underexpose in run and gun situations simply because even if I do have to raise shadows in post it's way easier to just use noise reduction to help improve the image quality. There is no such option for highlights, once they are gone, they are gone for good! So whilst ETTR is a viable and useful technique I think for most, the ability to combine multiple exposures in post rather than trying to achieve maximum DR via one single image has rendered it almost obsolete.
That's always going to be the case, there's always going to be tricky lighting that requires full manual exposure control. True exposure to the right has never meant that you'd expose so bright that you'd be blowing a bunch of highlights and areas. You might blow some minor highlights if there's unimportant detail, like glare on a car hood, but if it's significant, then you're not exposing to the right, you're just overexposing and hoping that you can save it in post.
Legitimate expose to the right photos aren't going to have a significant spike to the histogram on the right side, if you do, then you've over-exposed.
An alternative is to use extend low ISO. For instance, some canon cameras have base ISO of 100, but you can extend it to 50. It's basically the same as shooting in ISO 100, overexpose one stop, and in post process reduce one stop.
I shoot slightly to the left, but I’m not really looking for detail in the darks.
I just go as far to the right where I’m comfortable with what I might be losing. But from what I’ve learned on Fujifilm, the raw file most of the time doesn’t have blown out highlights when I do that, and Fuji raw files tend to be at least a stop under what the jpg is...
When I’m doing paid work, I’m more careful with what I do in camera because I have to edit the raw. But when it’s shooting just for me, I care how the jpg looks and still do raw + jpg just in case there’s an image I really really want to take into post.
Thanks, David. That takes a load off my shoulders. 😄
One thing to remember, the metering area! Most shoot wide metering but there will be times of spot (maybe for a bright spot). Also many will move the spot around to a dark area then a light area, this is how the ETTR can be seen. For those with zebras active you adjust the exposure dial neg till they go away if too bright you need to bracket, say for a sunrise with the sun visible. For night shots in manual you have M.M. at the bottom, use Full Frame metering and if a city or town watch the zebras but for those dark places say for Milky Ways with f/wide and SS set so no star tailing adjust ISO for M.M. of .7 will be very close to 3/4 to the right. With sensors today shadows with detail easy to bring out BUT those Highlights can never be. Night shots in the city watch the zebras but stay above ISO 800 noise worse below. Also try ISO Auto yes can get high and maybe noisy (not so much on ISO invariant sensors) but as of 2020 Lr and other Post Programs will clear up without softening.
Very useful explanation. Thank you for posting this.
I like to let the highlights go above a little when shooting in like golden hour and the light is shining behind the subject but usually under expose. But recently learned about HDR and it’s a pretty unique tool to use to get best of everything.
If you've got a static scene to shoot and you've got a tripod or reasonably stable hands, that can work. But, it all depends on the software or your skill in aligning things to get a good result.
Great video - glad you actually looked at the prints! 😎🙏
Canon and adobe suggest to underexpose in doubt! Little overexposure was the way to go in old times on film.
Great point to view the print from a high quality printer. That's the ultimate histogram for me.
If you expose for middle grey, then everything will be fine for most photos. If you have too much white like in weddings and snow, then change exposure to favor highlights. If photographing a dark subject, then exposure to favor details in the dark areas. If subject is not moving and you are using a tripod, you can take the shot at multiple exposures and combine the images to get what you want.
Most non photographers do not pay attention to shadow details or even slightly blown highlights, but care more about the overall picture. The quality of the light and color is most important as well as the composition. Some of the most famous photos historically have poor exposure.
Sometimes if you’re over-zealous pushing the highlights when you try recovering in post processing you find the though you have tone and texture you’ve lost true color.
Yup. Gotta be careful not to go too far. That’s why I think you’re usually better off just nailing the exposure the first time.
Loved it. I’m subscribing!
I wouldn't argue ETTR on a sensor that is not ISO invariant such as many Canon sensors (with your guidance about not going overboard as a caveat). Seems a lot of Canon folks are the ones that pitch ETTR. But once you bring in an ISO invariant sensor (for example - Sony and Fuji) you won't see a difference between setting the exposure at the base ISO and jacking the exposure up versus setting it at the ISO you need for the correct exposure at a given shutter/aperture combo.
Sony/Nikon/Fuji/Pentax users - give it a try. Put your camera on a tripod and manually focus on a subject. Set your camera to an ISO that is 3 stops above base ISO for your camera and adjust your shutter speed and aperture to get a correct exposure. RAW images only. No JPG and no picture effects. Take the shot. Change only your ISO to the camera's base ISO leaving the shutter speed and aperture the same and retake. Bring both images into your favorite editor and do some side by side pixel peeping. What do you see? Why do it? Well - when you underexpose the shadows you also underexpose the highlights and end up capturing some amazing color and shadow detail in the highlight sections of the photo that are much easier to work with than an ETTR exposed image. And you will have the same level of noise in the dark areas with both settings. Cool! Should you do it for every picture. Good lord no. But images where you have a looming sunrise or perhaps some streetlights in an otherwise dark street can give you some really nice results. Part of the fun of this hobby is experimenting and this is a good one to understand.
All this said - your points about getting the exposure correct for once in a lifetime moments and if you are going to do ETTR not going overboard are spot on. As a wildlife photographer ETTR gets pitched like it is the only way to do things and I have seen the disaster that results when the background and subject are dramatically different in terms of exposure values and people push to the right and then find they can't get back fine shadow details on white birds. Not saying to underexpose in these situations (absolutely do not do so) - but keep as much detail as you can in the center of the histogram.
Really appreciate your videos David. Well thought out and presented. Looking forward to more.
So bottom line - get it right the first time. Hehe. And thanks!
MovieBuff Texas, ETTR has nothing to do with ISO invariance. The point of ETTR is to keep the ISO constant, usually at base, while increase the exposure (the amount of light hitting the sensor) as much as possible without saturating important highlight details. More light will benefit any camera, regardless if it has ISO invariant sensor or not.
MovieBuff Texas, ETTR has nothing to do with ISO invariance. The point of ETTR is to keep the ISO constant, usually at base, while increase the exposure (the amount of light hitting the sensor) as much as possible without saturating important highlight details. More light will benefit any camera, regardless if it has ISO invariant sensor or not.
Thanks for video David it is a real light on moment
So am I correct that if you were to expose at night a cowboys closeup face dimly light by a campfire way out of frame you would ETR to get detail in the mid-tones assuming that high-tones are nonexistent? If darkening is preferred it can be adjusted in post.
Digital darkroom talk is filled with useful information. Enjoyed it brought back memories of working in a pro lab. 👍
Very well explained, love this series and channel. 👍🏻👍🏻👍🏻
A very good video with arguably poor and even wrong explanations.
You made a case where the exposure time on one image was 4 times as that of the other and hence the noise is obviously lower because it received 4 times as light.
What you should, in principle, do, is to set exposure and aperture equal, and capture at different ISOs. That way the only difference would be how well the camera handles A2D (Analogue to digital) conversion. The noise would be the same.
And in case you're saying oh no, higher ISO makes the image grainier you should go ahead and watch Tony and Chelsea's video on ISO.
ETTR and ETTL actually depends on the camera you use... For Sony, you'd definitely better expose to the left.
A fair point, but In the real world, I’d still want to keep my ISO as low as possible and would rather slow my shutter speed (on a tripod, of course). Lots of controversy on the Northrup video and many disagree with his thoughts on base ISO.
@@DavidBergmanPhoto Correct.
Once you expose a picture CORRECTLY, ISO measures the noise. Otherwise it doesn't. I think that's the controversy behind Northrup video.
But in this video, he's not doing this. So, the Northrup argument applies here.
On my Sony cameras, I have found that exposing to the left gives me better results. This is especially the case if I'm shooting Jpeg. For Raw, I think there might be a case where ETTR is good, but I should do some more experiments.
Changing the ISO is changing the gain not the exposure. The sensor behaves the same all the time if you change the gain, you are just amplifying the output of the sensor. The best way to do the comparison is by under exposing and over exposing to increase and decrease the amount of light per photo-site. ETTR is about exposure after all.
ETTR doesn't mean over exposing images. It's just we tryna get maximum brightness possible in camera without blowing out the highlights.
Pretty much what I said. :)
This guy is great love the channel, very informative and keeps it simple for guys like me! :)
Thanks for your video. But I still don't see the point of histograms. Why look at an abstract histogram when I can just see the photo itself? I can see the exposure via the camera. Using the histogram is one step removed from looking at the screen or view finder. I don't see their point. It's like looking at the tubes of paint that painters use instead of looking at what the painter painted.
Sometimes your screen or viewfinder can be too bright or too dark. Also Sony mirrorless camera screens are the worse they can overexpose or clip highlights on screen. many camera screens or viewfinder cannot be fully trusted. Histogram gives you a perfect range of your jpg image exposure, so you know in raw you have alittle more room.
Thank you for TEACHING....LOVE YOUR VIDEOS
Around 4', "the fact that it's linear... it means it's exponential" 😀. Jokes aside, nice video and a lot of passion in the way you explain things. Thanks !
I think I got what he want to mean. It´s that for instance - for a dark area you have kinda 0-30 brightness - and a light one 180-255 - So, multiply by 3 - RGB - It means that enven in a linear gather, you would have much more information as you go to the right. :)
Maybe a simple term as - the amount of data (details) - grows as you go to the brighter side - So, more information to the software to render the image. :D
Luke, I´m your father!
Lots to learn David!! thanks a lot. I have one question that I hope you can help me with in the future! :) I recently bought a Lap Top that runs windows and want to set up Capture ONE PRO 20 to work with Capture Pilot (I have an Iphone XR). I have tried to do it but has not worked... it would be amazing if you can help me with this!! Thanks in advance!
Hmmm I’m not a windows user, but it should work fine. You might need to contact Phase One tech support so they can troubleshoot for you.
ETTR is often for video were videographers aren’t recording raw an possibly are shooting raw, A lot of raw curves (e.g. Panasonic’s V-Log) allow for much more detail in the upper range of the curve - use don’t exceed te maximum range in the highlights. Also, for photography, adjusting in post to raise the brightness only works for cameras with ISO-invariant sensors
Excellent video, David. So, when all is said and done it's really "Much ado about nothing". I can certainly follow that!
Uhm, since when has ETTR meant *overexpose* ? (Hint: It has NEVER meant that)
Maybe it’s semantics, and it doesn’t mean exposure so far that you blow out / lose highlights. But the whole point is to expose a bit brighter than you want your final image. To me, that is overexposing.
Thank you! Very helpful!
Well explained!
It’s yet another fad. There will be a lot of people doing it because an influencer said it’s a good idea. As you have shown in the majority of instances it adds nothing. Generally aim not to burn out highlights and create darkness in post. Before long manufacturers will be adding an ETTR mode on the set up menu and, people will bore on about how a shot was ruined because the photographer should have used ETTR. Good cameras have huge dynamic range and this renders ETTR pointless as a concept. Maybe the micro 4/3 shooters can occasionally get a little more mileage but, post production diminishes the majority of problems in exposure ever more effectively.
What you said is true with one exception. Budget shooters. The way m4/3 sensors curves are you get way better texture and details with ETTR 1 stop if you can afford it in your circumstances. Lets say that stops quality is the best for the first Native ISO MG set by developers, lets say for ISO 400 and WB 5600 for BMPCC4k. Thats the mode you shoot on most if the time. Exposing for ISO 200 get you better results, but you risk blowing out highlights which isnt worth it in any digital camera. Budget cameras butcher contrast and hues when underexposing, so you can get -1 stop at best realistically. I have external plugins for denoising made for porfessionals, but they cant salvage bad image texture. Its not grainy, more like muddy.
The audio in these past videos hurt the ear. Its a bit tinny and thin. I use broadcast heaphones.
Sorry about that! Been working on my audio. Check my latest videos and I think you’ll be happier. :)
So it's basically "The Zone System" for the Digital world :-)
Kinda. :)
About ETTR being over rated -- I could not agree more. Thank you for lending your considerable expertise to what I have already known to be unnecessary work.
Helpful!!!
off topic, recently bought a manual macro lens, the laowa 65mm, love the magnification but want even more, would you recommend putting macro extension tube on top for more magnification or one of those raynox front lens attachment?
Have never used those attachments. Tubes work pretty well in most cases.
Loving these thumbnails 😉
So in terms of noise is it better to shoot an image at a lower iso and have to raise the exposure in post or get as close to perfect exposure even if it means shooting at a much higher iso?
Really depends on the camera as some of them handle higher ISO much better than others. With my Canon 1dx3, I won’t hesitate to push the ISO to 8000 or more. But I wouldn’t consider than with a camera from 10 years ago...
S Burgos, each camera has an ISO value after which it becomes ISO invariant, so further increasing ISO no longer improves the noise compared to just shooting at a lower ISO, with the same shutter speed and aperture, and brightening the image in post. In fact, if we just look at the image quality and ignore the inconvenience of post processing, it is better to never increase ISO value past that point, because you are just clipping the highlights, for no benefit. At what ISO value camera becomes ISO invariant is highly dependent on the camera model.
S Burgos, each camera has an ISO value after which it becomes ISO invariant, so further increasing ISO no longer improves the noise compared to just shooting at a lower ISO, with the same shutter speed and aperture, and brightening the image in post. In fact, if we just look at the image quality and ignore the inconvenience of post processing, it is better to never increase ISO value past that point, because you are just clipping the highlights, for no benefit. At what ISO value camera becomes ISO invariant is highly dependent on the camera model.
Does this apply to both DSLRs and Mirrorless?
Always kinda hear to favor darker than light with mirrorless.
Mirror or no mirror is irrelevant. It’s the sensor that matters. Some hold shadow detail better than others.
Thanks. ETTR, just like Bokeh, is just a buzz word / phrases I hear from people new (newer) to photography, as if they are the panacea to bad images. You are spot on about not being able to see real differences when comparing correctly exposed images, using different settings. Anytime I am asked, or read a post about any buzz work or phrase in photography, I ask the questioner if the panacea they are discussing is more about a short-cut to getting it right in camera, or the belief that post-production with save the image.
Overexposed isn't the right word. Overexposed means blown. To me, ETTR just means to the right hand side of the range that is properly exposed. If you start blowing the highlights, it's not ETTR, it's overexposed.
Hi, is this applicable to when filming in log?
Furthermore I found that on some camera's one colour can be overexposed, but not showing in the main histogram. So ETTR will enlarge that risk.
If you really want to push your exposure as much to the right as possible without clipping, search for UniWB. By setting a custom white balance you get a more „accurate“ histogram. The downside is that this white balance will make your preview look almost completely green.
@@andrecanis4894 Thanks, always nice to learn something new :)
ETTR eas probably more important before.
I think that there was a significant change from the 7D2 onwards. For all I know the change might be in DPP. Yes, I'm stone age!!!
Definitely as dynamic range improves, it becomes even less of an issue with each camera generation.
I like images with contrast. That means the Kodachrome look.
So, I'm not entirely enthusiastic about a wide dynamic range.
Very helpful, and great explanation. Thank you. When you speak of the histogram being affected by picture profiles, are you referring to the histogram readings during image playback, or do the picture profiles and settings affect the histogram while shooting also? Thanks.
Never understood the concept. Seems it would lead to blowing out the highlights. Another excellent video.
I “lean to the right” with my exposures...
Not under, not WAY over...but the histogram leans slightly (2/3-1 stop over)...
Histogram is representation of computer data...
David seems to be about the ONLY professional shooter understands that a camera is a very specialized one purpose computer!!!...🤓
It needs data to function correctly, but giving it too much corrupts what data is there...
“Lean to the right”
I like that. Maybe it should be called LTTR. :)
Hello David. When you convert your pictures from RGB to CMYK, which file do you use - raw or jpeg? thanks
I would go onto his site and ask the question there. He may get to it in a video.
When I've produced coffee table books, the CMYK conversions are done as the very last step and and then carefully proofed and adjusted (by the printer) to match my originals.