ETR Metering

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

That is all nice and correct in theory, but that is not how most of us shoot in the field. I use manual exposure in auto ISO mode a lot to shoot wildlife. I base my exposure by the in-camera histogram in the viewfinder and the setting level for zebra- stripes in my Sony A1 viewfinder. I adjust shutter speed and aperture to " expose to the right" to maximize the exposure. So ETTR is important no matter what ISO you use. I think all the discussion that restricts ETTR to only base ISO just confuses people. It is valid at all ISOs as long as one remembers that the"signal" in signal/noise ratio (or the "Exposure" in ETTR) is dependent on shutter speed and aperture.
I know I can get away with more underexposure at base ISO, because I am looking at data that is not being artificially amplified by a higher ISO. However, significant underexposure when one happens to be necessarily shooting at a high ISO can be devastating. So to me that means ETTR is even more critical at high ISOs.

I do dabble a bit in real world wildlife photography too :)

Here's an example of what I wouldn't do. Let's say I had a deer in a field, all middle tone, and it was dim with a proper exposure of 1/250th at F/4 at ISO 6400. This is as slow a shutter speed as I am willing to use and my lens is wide open - I've let in as much light as I can. If I shot it at those values, when I got back home, I wouldn't adjust the brightness at all ( I kind of get the impression you think I'm advocating underexposure and I'm not). In that scenario, I'd probably have to shoot at ISO 25,600 to push my histogram all the way to the right. I wouldn't do it. I'd shoot at ISO 6400 and get the proper level of brightness. If I shot at 25K and brought it back down at home by two stops, the result would look the same as had I just shot it at 6400 in the field, BUT I'm flirting with clipping and I my color fidelity may not be as good either. I guess my question is, in that scenario, would you shoot 6400 or 25K?

I do agree that if I have something in the frame that should be close to white that I'll adjust the histogram / zebras to a point just before clipping, but that's not really ETTR, it's just exposing properly. Sometimes a proper exposure has the histogram to the right. On the other hand, ETTR is generally done by deliberately overexposing* to increase dynamic range and then bring the brightness back down in post. If you already have all the light coming in that you can muster using shutter speed and F/stop, there's really no good reason to crank up ISO to make the image overly bright and then bring it back down. It's exposure with extra steps. IMO, it's better to just shoot at the proper ISO so no brightness adjustment is needed later on.

*Technically, ETTR is just exposing the brightest highlight areas you want to preserve to just before clipping, so there are times an underexposure is necessary for ETTR. However, most of the time an ETTR image looks too bright
 
while i agree it has to be there for iso to amplify, the file size is finite and so basically the file only captures a “window” of the data. if the data falls outside of that window it will not be recorded, thus lost forever.

I mean, I don't know. I've never seen it demonstrated where information from the RAW file was saved though ISO amplification. Do you have any examples or links?
 
That is all nice and correct in theory, but that is not how most of us shoot in the field. I use manual exposure in auto ISO mode a lot to shoot wildlife. I base my exposure by the in-camera histogram in the viewfinder and the setting level for zebra- stripes in my Sony A1 viewfinder. I adjust shutter speed and aperture to " expose to the right" to maximize the exposure. So ETTR is important no matter what ISO you use. I think all the discussion that restricts ETTR to only base ISO just confuses people. It is valid at all ISOs as long as one remembers that the"signal" in signal/noise ratio (or the "Exposure" in ETTR) is dependent on shutter speed and aperture.
I know I can get away with more underexposure at base ISO, because I am looking at data that is not being artificially amplified by a higher ISO. However, significant underexposure when one happens to be necessarily shooting at a high ISO can be devastating. So to me that means ETTR is even more critical at high ISOs.
I believe that ETTR is commonly mentioned at base ISO because what the technique does is flood the sensor with as much data up to the point where you have chosen to lose your highlights. Then, when you are post processing an image, you have more data to work with, and a lot of folks prefer recovering highlights to opening up shadows. As mentioned above, ISO does nothing to help flood the sensor with additional data; only changes in amount of light hitting the sensor can do that.

How/if we incorporate that technique into our shooting styles certainly depends on balancing a lot of other factors as well. But, if you are interested in capturing as much data on the sensor as possible, and that is of importance to the image, then it may become a technique that is more important to implement than other techniques.

--Ken
 
I mean, I don't know. I've never seen it demonstrated where information from the RAW file was saved though ISO amplification. Do you have any examples or links?

just trying to leverage what i think i understand about the technology.

i'm probably not explaining well.

so a raw file has a bit depth. basically how many bits of data are stored for an individual pixel. so if you have a bit depth of 16, that means you can store 65,535 values.

16-bits.jpg
You can only see EXIF info for this image if you are logged in.


i don't know enough about raw formats, but i assume that data has to represent values for red, green and blue for each pixel. so if that assumption is correct, you could store 21,845 values for each color.

so basically the bit depth is the dynamic range the raw file format is capable of recording. we can pause our discussion to simply consider that you can't fit 20-bits of dynamic range into a 16-bit format.

three.jpg
You can only see EXIF info for this image if you are logged in.


in reality, it doesn't matter if the math is right, the point here is the amount of data that can be stored for the color value of an individual pixel is finite.

so, my understanding is ISO in digital cameras is basically amplification (gain) applied to the signal from the sensor. my belief is this is an analog process at this point, before the analog signal is converted to digital. my belief is after you apply the iso, you convert the signal to digital, using the bit depth of the raw file.

sensor ---> gain ---> digital/analog conversion ---> raw file

so, if we look at the below histogram and consider the blue bracket is a sample of 16 bits of data captured with the gain of iso 10,000 applied, and the orange bracket is 16 bits of data captured with a gain of 100 iso applied, we can see we can lose data either at either end of the spectrum.

middle-iso.jpg
You can only see EXIF info for this image if you are logged in.


in many cases, we can see the data of our scene fits inside of the raw file because we can see the lumps of data slope off before they hit the edges of the histogram.

in these cases, the data is present, so it's simply a question of if the gain in the camera is better than brightening the image in post, and, here i think i agree with you it _usually_ doesn't matter (much?), but caveat, it's probably camera dependent.

so a (raw, flat-neutral) histogram basically provides a visual representation of what data is in the raw file. we know that if we crank the iso and cause the sky to blow out, that the data is simply not present in the file and we can see it in the histogram.

same for shadows. if the shadows go below the left side of the histogram, they simply aren't in the file.

my kingdom for more dynamic range.
 
while i agree it has to be there for iso to amplify, the file size is finite and so basically the file only captures a “window” of the data. if the data falls outside of that window it will not be recorded, thus lost forever.

As I understand it for the one kind of noise photons even in steady light vary a little bit, so one pixel might be 'correct' but the next door neighbor in the same light might randomly end up being hit with too many or too few. Low exposure is like fliping a coin ten times, you might get strange results but adequate exposure is like flipping it 10,000 times where a clear pattern emerges. If you let in enough photons the correct amount emerges. If not then there is uncertainty in telling what the correct value is. Like a solar collector the photons knock electrons loose in each pixel/photosites which the camera converts to voltage and then converts the voltage to a number, adding the gain from the iso after exposure but either before or after the digital conversion.

My main point being that the exposure is well in the past by the time the gain from the iso is applied, so high or low, iso doesn't have anything to do with exposure, but if the initial exposure lacked enough photons for a clear pattern to emerge iso would add gain to the incorrect as well as the correct.
 
just trying to leverage what i think i understand about the technology.

i'm probably not explaining well.

so a raw file has a bit depth. basically how many bits of data are stored for an individual pixel. so if you have a bit depth of 16, that means you can store 65,535 values.

View attachment 30623

i don't know enough about raw formats, but i assume that data has to represent values for red, green and blue for each pixel. so if that assumption is correct, you could store 21,845 values for each color.

so basically the bit depth is the dynamic range the raw file format is capable of recording. we can pause our discussion to simply consider that you can't fit 20-bits of dynamic range into a 16-bit format.

View attachment 30624

in reality, it doesn't matter if the math is right, the point here is the amount of data that can be stored for the color value of an individual pixel is finite.

so, my understanding is ISO in digital cameras is basically amplification (gain) applied to the signal from the sensor. my belief is this is an analog process at this point, before the analog signal is converted to digital. my belief is after you apply the iso, you convert the signal to digital, using the bit depth of the raw file.

sensor ---> gain ---> digital/analog conversion ---> raw file

so, if we look at the below histogram and consider the blue bracket is a sample of 16 bits of data captured with the gain of iso 10,000 applied, and the orange bracket is 16 bits of data captured with a gain of 100 iso applied, we can see we can lose data either at either end of the spectrum.

View attachment 30622

in many cases, we can see the data of our scene fits inside of the raw file because we can see the lumps of data slope off before they hit the edges of the histogram.

in these cases, the data is present, so it's simply a question of if the gain in the camera is better than brightening the image in post, and, here i think i agree with you it _usually_ doesn't matter (much?), but caveat, it's probably camera dependent.

so a (raw, flat-neutral) histogram basically provides a visual representation of what data is in the raw file. we know that if we crank the iso and cause the sky to blow out, that the data is simply not present in the file and we can see it in the histogram.

same for shadows. if the shadows go below the left side of the histogram, they simply aren't in the file.

my kingdom for more dynamic range.

Nice post. To your one question, each pixel/photosite filters for only one color, red or green or blue. In Lightroom or whatever the software looks at the other colors in the neigboring pixels to pick the other 2 colors for that pixel. So the bit depth in camera gets to use all 14 or 16 bits for just the one color.then after the raw is converted the file size triples with the addition of the remaining two colors for each pixel.

Rawdigger is pretty cool if you like that sort of thing.
 
Last edited:
As I understand it for the one kind of noise photons even in steady light vary a little bit, so one pixel might be 'correct' but the next door neighbor in the same light might randomly end up being hit with too many or too few. Low exposure is like fliping a coin ten times, you might get strange results but adequate exposure is like flipping it 10,000 times where a clear pattern emerges. If you let in enough photons the correct amount emerges. If not then there is uncertainty in telling what the correct value is. Like a solar collector the photons knock electrons loose in each pixel/photosites which the camera converts to voltage and then converts the voltage to a number, adding the gain from the iso after exposure but either before or after the digital conversion.

My main point being that the exposure is well in the past by the time the gain from the iso is applied, so high or low, iso doesn't have anything to do with exposure, but if the initial exposure lacked enough photons for a clear pattern to emerge iso would add gain to the incorrect as well as the correct.
That is a great explanation!
 
just trying to leverage what i think i understand about the technology.

i'm probably not explaining well.

so a raw file has a bit depth. basically how many bits of data are stored for an individual pixel. so if you have a bit depth of 16, that means you can store 65,535 values.

View attachment 30623

i don't know enough about raw formats, but i assume that data has to represent values for red, green and blue for each pixel. so if that assumption is correct, you could store 21,845 values for each color.

so basically the bit depth is the dynamic range the raw file format is capable of recording. we can pause our discussion to simply consider that you can't fit 20-bits of dynamic range into a 16-bit format.

View attachment 30624

in reality, it doesn't matter if the math is right, the point here is the amount of data that can be stored for the color value of an individual pixel is finite.

so, my understanding is ISO in digital cameras is basically amplification (gain) applied to the signal from the sensor. my belief is this is an analog process at this point, before the analog signal is converted to digital. my belief is after you apply the iso, you convert the signal to digital, using the bit depth of the raw file.

sensor ---> gain ---> digital/analog conversion ---> raw file

so, if we look at the below histogram and consider the blue bracket is a sample of 16 bits of data captured with the gain of iso 10,000 applied, and the orange bracket is 16 bits of data captured with a gain of 100 iso applied, we can see we can lose data either at either end of the spectrum.

View attachment 30622

in many cases, we can see the data of our scene fits inside of the raw file because we can see the lumps of data slope off before they hit the edges of the histogram.

in these cases, the data is present, so it's simply a question of if the gain in the camera is better than brightening the image in post, and, here i think i agree with you it _usually_ doesn't matter (much?), but caveat, it's probably camera dependent.

so a (raw, flat-neutral) histogram basically provides a visual representation of what data is in the raw file. we know that if we crank the iso and cause the sky to blow out, that the data is simply not present in the file and we can see it in the histogram.

same for shadows. if the shadows go below the left side of the histogram, they simply aren't in the file.

my kingdom for more dynamic range.

All very good points and I think I see where you're going now. :)

I think where this will all make sense / come together is when you think about useable information. As the SNR decreases (lower light levels), the information in the shadows is overwhelmed by noise (with modern cameras, mostly photon noise). Basically, the information that is there is useless because the noise has obscured it. If you amplify the brightness, you're amplifying the noise as well. So, if the noise obscures it at lower brightness levels, that doesn't change when it gets brighter. Basically, you have brighter noise, but not any more data.

In fact, you've maybe heard me mention in some of my materials that once you get past ISO 400 ~800 range on most cameras, there's very little need for 14 bit files anymore. The shadow noise has decreased the DR to such an extent that those extra two bits are only storing noise and not useful information anymore. (And I have tested this - you get nicer noise, but you still can't discern medium or fine detail - however - 14 bit DOES hold an advantage in super-deep shadows under that 400-800 ISO range.)
 
The issue of ETTR is really a method of trying to have more light/photons captured by the photosite in midtone and dark areas. There is plenty of data captured in bright areas and you are not going to have a problem with manipulating those areas in software or dealing with noise. At the same exposure for dark areas, you may have 1/1000 the amount of data captured. By Exposing To The Right as base ISO, you can double or quadruple the amount of data you are capturing, and in darker areas that makes a big difference in shadow detail. Even if you bring down the exposure in post processing by those stops, you have captured the data and can pull out detail and minimize any noise.

As Steve described, it's better to raise ISO in camera than in post processing. The image data captured is the same, but modern cameras use Dual Gain to add amplification and noise management when the RAW file is created. With mirrorless it makes it easier to see and focus if you have a bright enough EVF, so you are more likely to make a good image with precise focus.

In this example from Photonstophotos.net you can see dual gain applied at ISO 500 and above. This is an effort to amplify the signal and reduce noise at the RAW level. If you simply shot at ISO 200 rather than ISO 1600, you give up this benefit. As far as we know, it is better to use the actual ISO required to make the image or at least a level of ISO 640 or higher to take advantage of the dual gain. Some other cameras have dual gain applied at different levels or more than one time, so this is camera specific. So assuming the camera is ISO invariant, it's usually within a given range and not for all ISO levels.


In the case of Nikon files, if you use the camera settings in LR, the software can apply ISO specific sharpening and noise combinations that are embedded by Nikon in the RAW file. Adobe reads these values and applies them, and you can still continue to adjust to your taste.

As we have discussed before, cropping needs to be part of the thought process for ISO and adjustments in post processing. Every time you reduce image size by 1/3, you are effectively magnifying the noise by the equivalent of a full stop. And when you are resizing the image to a smaller image size, you are effectively reducing the size of noise and potentially averaging pixels with noise with pixels that don't have noise.
 
I believe that ETTR is commonly mentioned at base ISO because what the technique does is flood the sensor with as much data up to the point where you have chosen to lose your highlights. Then, when you are post processing an image, you have more data to work with, and a lot of folks prefer recovering highlights to opening up shadows. As mentioned above, ISO does nothing to help flood the sensor with additional data; only changes in amount of light hitting the sensor can do that.

--Ken

But wildlife photographers don't often shoot at base ISO. i cant shoot a bird in flight at 1/2000 of a second at ISO 100 and know whether or not I have all the data I can capture (can you?) I'm going to have an ISO ceiling that I will use and then adjust shutter speed and aperture to capture as much data as I can.
 
Last edited:
All that talk makes my head spin! I realize that in most situations the correct exposure is all that is needed. But if the light is falling, your subject is sitting in a great spot, you've maxed out your aperture and shutter speed, then ISO is the only thing left. At that point it may be better to increase the ISO beyond a correct exposure and make sure your histogram hits the right side just a bit. I raised the question because I have a D850 and they are not known to have great high ISO performance. But since I read the article by Jared Lloyd I've experimented with the technique and have had mixed results. I've attached an examples. In the past, if I was maxed out in shutter and aperture, I would never push that camera past 3200 ISO because the noise would render the photo useless. The photo of the nips was taken at 10000 ISO F4 1/13 ( Hand held! ) Other photos I'd like to add are too large and for some reason the "file setting" drop down is missing from my export dialogue box. I can't reduce their size. If anyone can tell me how to fix that I will add other photo's taken at high ISO with the D850!
Hi ISO-3688.jpg
You can only see EXIF info for this image if you are logged in.
 
I do dabble a bit in real world wildlife photography too :)

Here's an example of what I wouldn't do. Let's say I had a deer in a field, all middle tone, and it was dim with a proper exposure of 1/250th at F/4 at ISO 6400. This is as slow a shutter speed as I am willing to use and my lens is wide open - I've let in as much light as I can. If I shot it at those values, when I got back home, I wouldn't adjust the brightness at all ( I kind of get the impression you think I'm advocating underexposure and I'm not). In that scenario, I'd probably have to shoot at ISO 25,600 to push my histogram all the way to the right. I wouldn't do it. I'd shoot at ISO 6400 and get the proper level of brightness. If I shot at 25K and brought it back down at home by two stops, the result would look the same as had I just shot it at 6400 in the field.
In the situation you describe I would do exactly as you do. But if you happened to be shooting at 1/500 of a shutter speed and there was histogram room to the right, I would decrease the shutter speed to 1/250 of a second and capture more data. The histogram , as you know is based on a processed jpeg and one may actually process it quite differently in Lightroom and photoshop.. I personally use a flat jpeg profile. I dont think of raising the ISO as ETTR, but I use shutter speed and aperture to push the histogram as far to the right as possiible without clipping highlights. This is so much easier to do now with mirrorless cameras.
I reponded to this post because I thought your initial LOL response was misleading to the concept ETTR. Jared Lloyd is an extremely talented nature and wildlife photographer. His images are impressive!
 
Last edited:
In the situation you describe I would do exactly as you do. But if you happened to be shooting at 1/500 of a shutter speed and there was histogram room to the right, I would decrease the shutter speed to 1/250 of a second and capture more data. The histogram , as you know is based on a processed jpeg and you may process it differently in Lightroom and photoshop.. I personally use a flat jpeg profile. I dont think of raising the ISO as ETTR, but I use shutter speed and aperture to push the histogram as far to the right as possiible without clipping highlights. This is so much easier to do now with mirrorless cameras.
I reponded to this post because I thought your initial LOL response was misleading to the concept ETTR. Jared Lloyd is an extremely talented nature and wildlife photographer. His images are impressive!

I kind of think what you're calling ETTR might be what I'm calling just a correct exposure. :)

I too would drop to 1/250th if I could from 1/500th to increase the SNR. No question. What I'm saying is that if I'm not at base ISO, I would not deliberately overexpose the image, I'd expose it correctly. So, if at 1/500th I had the correct brightness for an image at ISO 6400, when I dropped to 1/250th, I'd drop the ISO to 3200 rather than letting the image overexpose by a stop and then pulling it back. There's no benefit to it.

On the other hand, if I'm underexposed at ISO 6400 and 1/500th, then when I drop to 1/250th I would leave the ISO at 6400 so I had the correct level of brightness.

Also, you're right, that LOL was dismissive and I edited the post. My apologies for being offensive.
 
I kind of think what you're calling ETTR might be what I'm calling just a correct exposure. :)

I too would drop to 1/250th if I could from 1/500th to increase the SNR. No question. What I'm saying is that if I'm not at base ISO, I would not deliberately overexpose the image, I'd expose it correctly. So, if at 1/500th I had the correct brightness for an image at ISO 6400, when I dropped to 1/250th, I'd drop the ISO to 3200 rather than letting the image overexpose by a stop and then pulling it back. There's no benefit to it.

On the other hand, if I'm underexposed at ISO 6400 and 1/500th, then when I drop to 1/250th I would leave the ISO at 6400 so I had the correct level of brightness.

Also, you're right, that LOL was dismissive and I edited the post. My apologies for being offensive.
I think you are correct in that we are we are really talking about the same exposure. Because the in camera histogram is so small and I am afraid I may miss smaller areas of clipped highlights, i actually tend to rely more on the zebra stripes for adjusting exposure. And if you are using the zebra stripe method, then it seems to me that you truly are using an ETTR technique.. By the way., you've got great images also.:)
 
Seems there are at least two sources of noise: Photon statistical and instrumental. Both are stochastic (random).

For a given ISO, assume the average instrumental noise comes at a steady rate.

Exposure is determined by three things: Incident light intensity, aperture and shutter duration.

To exaggerate, start with a light intensity giving equal contributions to the camera's signal from light and noise...A SNR (signal-to-noise ratio) of 1:1.

If the intensity of light is doubled, the SNR increases to 2:1.

If area of the aperture is doubled, the SNR increases to 2:1.

But, if the duration of the shutter is doubled, the SNR is unchanged at 1:1!

Has anyone observed this in practice?

:unsure:
 
But wildlife photographers don't often shoot at base ISO. i cant shoot a bird in flight at 1/2000 of a second at ISO 100 and know whether or not I have all the data I can capture (can you?) I'm going to have an ISO ceiling that I will use and then adjust shutter speed and aperture to capture as much data as I can.
I believe that what I intended to convey and what you have taken away are not the same, so let me try to rephrase what I said in my second paragraph which I am quoting below:

How/if we incorporate that technique into our shooting styles certainly depends on balancing a lot of other factors as well. But, if you are interested in capturing as much data on the sensor as possible, and that is of importance to the image, then it may become a technique that is more important to implement than other techniques.

First, I live in the PNW, like to shoot BIF in winter, and it is always a struggle to get enough light under our often gray skies, so I understand about needing to crank up the ISO and keep shutter speeds high. And this helps make my point. I am prioritizing the need to properly capture BIF first and foremost. And that means picking shutter speeds that will give me crisp images and hopefully eliminate subject blur. If I was lucky to be shooing on a very sunny day, I might not need to raise my ISO, and could also incorporate ETTR if that did not conflict with my settings for BIF. But if the day is cloudy, like I often see, I still need my shutter speed set high to avoid subject blur, and my 200-500 is wide open at f/5.6, that leaves me with no option but to raise my ISO or address an underexposed image in post processing. The former is generally preferred, but results can be sensor dependent. ETTR, in this case is somewhat moot since you have higher priority needs and shooting conditions that do not allow you to do a base level ISO ETTR exposure. In any event, you still need a proper exposure, unless you plan to correct in post processing.

tl;dr ETTR is not a technique that can be universally applied in all types of shooting situations.

Hope this helps,

--Ken
 
I believe that what I intended to convey and what you have taken away are not the same, so let me try to rephrase what I said in my second paragraph which I am quoting below:



First, I live in the PNW, like to shoot BIF in winter, and it is always a struggle to get enough light under our often gray skies, so I understand about needing to crank up the ISO and keep shutter speeds high. And this helps make my point. I am prioritizing the need to properly capture BIF first and foremost. And that means picking shutter speeds that will give me crisp images and hopefully eliminate subject blur. If I was lucky to be shooing on a very sunny day, I might not need to raise my ISO, and could also incorporate ETTR if that did not conflict with my settings for BIF. But if the day is cloudy, like I often see, I still need my shutter speed set high to avoid subject blur, and my 200-500 is wide open at f/5.6, that leaves me with no option but to raise my ISO or address an underexposed image in post processing. The former is generally preferred, but results can be sensor dependent. ETTR, in this case is somewhat moot since you have higher priority needs and shooting conditions that do not allow you to do a base level ISO ETTR exposure. In any event, you still need a proper exposure, unless you plan to correct in post processing.

tl;dr ETTR is not a technique that can be universally applied in all types of shooting situations.

Hope this helps,

--Ken

What you say makes a lot of sense. I still am not sure about something in my own shooting. My camera is not invariant so the shadow noise does improve with in-camera iso as the iso increases, relative to just brightening in Lightroom. So in that case, does it pay to pursue an ettr strategy at other than base iso? I think I am certain that if the camera was pretty invariant it would not pay, but if the camera is not invariant I'm thinking there might be a small benefit?
 
What you say makes a lot of sense. I still am not sure about something in my own shooting. My camera is not invariant so the shadow noise does improve with in-camera iso as the iso increases, relative to just brightening in Lightroom. So in that case, does it pay to pursue an ettr strategy at other than base iso? I think I am certain that if the camera was pretty invariant it would not pay, but if the camera is not invariant I'm thinking there might be a small benefit?
Good questions, and I am not certain I can give you any kind of definitive opinion. My take is that once you are beyond native ISO, you are not really doing ETTR as some of us have narrowly defined above. At that point, I think that you have to focus on good exposure for the subject matter that is most important. As to the question of weather it is better to raise ISO in the camera or do it in a program like LR Classic, I have seen reputable answers on both sides of the coin and think that the best answer may be situation (and skills) specific. And I am mostly okay with that as long as folks understand the choices and know that one size may not fit all situations or personal styles. I once read a piece from a very famous photographer who defended underexposing his images. I questioned the logic, but the images looked fine, and he was quite famous, so whatever he was doing seemed to work for him. Whether he could have produced even better images with a different technique is certainly up for discussion, but I do try to remember that photography is an art, and our work reflects our style, approach and choices. I do love Cook's Illustrated and have learned a lot from them, but there are some dishes that I have learned to cook the way I want to get the results that I want. It is good to have beginner's mind and continue to learn to improve, but at some point we also have to own our styles. Balancing the two is what makes hobbies/interests fun and engaging. I am curious to hear what others have to say about your specific situation.

--Ken
 
What you say makes a lot of sense. I still am not sure about something in my own shooting. My camera is not invariant so the shadow noise does improve with in-camera iso as the iso increases, relative to just brightening in Lightroom. So in that case, does it pay to pursue an ettr strategy at other than base iso? I think I am certain that if the camera was pretty invariant it would not pay, but if the camera is not invariant I'm thinking there might be a small benefit?
This is a good distinction. I remember years ago Nikon Capture NX2 did a horrible job of changing exposure more than a stop, but Lightroom could handle a couple of stops without an issue because it was reprocessing the RAW file. Lightroom later evolved into a product that could make much larger adjustments to exposure, and Nikon NX Studio handles multiple stops of exposure adjustment very well.

I looked back at some images from the past few years and found one with a D800E, 600mm lens, f/4.5 for DOF, 1/200 sec, and ISO 800. It was an image intentionally exposed to the right at ISO 800, but my shutter speed and aperture had reached the maximum I could safely use. The image - or a male silverback gorilla in a zoo - was midtones and very dark tones - and ultimately converted to B&W. I exposed to the right at capture to have detail in the shadows of the face and the dark hair of the gorilla, but I still ended up pulling back the exposure in the brighter areas - particularly the face.

Here are the two images, and you can see how the original capture was intentionally over exposed by a full stop or more, but pulled back for the final image. You can see how dark the face would be without overexposure.

Zoo Atlanta_9-17-2013_130878.jpg
You can only see EXIF info for this image if you are logged in.


The key to this image is the eyes with the rest of the face being darkened from the original. By the way - my wife would not let me hang this in our house. :)
Zoo Atlanta_9-17-2013_130878-b_1.jpg
You can only see EXIF info for this image if you are logged in.
 
This is a good distinction. I remember years ago Nikon Capture NX2 did a horrible job of changing exposure more than a stop, but Lightroom could handle a couple of stops without an issue because it was reprocessing the RAW file. Lightroom later evolved into a product that could make much larger adjustments to exposure, and Nikon NX Studio handles multiple stops of exposure adjustment very well.

I looked back at some images from the past few years and found one with a D800E, 600mm lens, f/4.5 for DOF, 1/200 sec, and ISO 800. It was an image intentionally exposed to the right at ISO 800, but my shutter speed and aperture had reached the maximum I could safely use. The image - or a male silverback gorilla in a zoo - was midtones and very dark tones - and ultimately converted to B&W. I exposed to the right at capture to have detail in the shadows of the face and the dark hair of the gorilla, but I still ended up pulling back the exposure in the brighter areas - particularly the face.

Here are the two images, and you can see how the original capture was intentionally over exposed by a full stop or more, but pulled back for the final image. You can see how dark the face would be without overexposure.

View attachment 30651

The key to this image is the eyes with the rest of the face being darkened from the original. By the way - my wife would not let me hang this in our house. :)
View attachment 30652
That is quite an amazingly expressive image. Nice work!

--Ken
 
This is a good distinction. I remember years ago Nikon Capture NX2 did a horrible job of changing exposure more than a stop, but Lightroom could handle a couple of stops without an issue because it was reprocessing the RAW file. Lightroom later evolved into a product that could make much larger adjustments to exposure, and Nikon NX Studio handles multiple stops of exposure adjustment very well.

I looked back at some images from the past few years and found one with a D800E, 600mm lens, f/4.5 for DOF, 1/200 sec, and ISO 800. It was an image intentionally exposed to the right at ISO 800, but my shutter speed and aperture had reached the maximum I could safely use. The image - or a male silverback gorilla in a zoo - was midtones and very dark tones - and ultimately converted to B&W. I exposed to the right at capture to have detail in the shadows of the face and the dark hair of the gorilla, but I still ended up pulling back the exposure in the brighter areas - particularly the face.

Here are the two images, and you can see how the original capture was intentionally over exposed by a full stop or more, but pulled back for the final image. You can see how dark the face would be without overexposure.

View attachment 30651

The key to this image is the eyes with the rest of the face being darkened from the original. By the way - my wife would not let me hang this in our house. :)
View attachment 30652

Wonderful image!

Your post has me thinking that we all mean a different thing with ETTR. :)

In the example you posted, you'd need another stop, probably two, to qualify for what I think of as an ETTR. IMO, what you did was just smart metering :)

I believe a "proper" ETTR (if there is such a thing) is normally done by identifying the brightest area in the image and exposing to just a bit before clipping. Not clipping in the RAW file, clipping in the JPEG preview. The typical method used is to test your camera and see how far you can go past where it shows clipping for whatever picture profile you use, then add that amount of exposure to the image once you have the exposure there.

For instance, I can actually shoot my D850 1.7 stops past the point where I see clipping (I've tested it). So, if I see clipping and want to do an ETTR, I'd set my exposure so it's just at that edge of clipping. That's my starting point. Then I'd add 1.7 stops. Once home, I'd pull back in Lightroom.

I think what most are describing in this thread is simply adding a little extra exposure as needed, not ETTR as I understand it. I think that's why there are so many differing thoughts :)
 
I think of ETTR in practical terms that I use in real world shooting. Sony Ambasdador Mark Galer obviously does the same. Here are a few quotes from his Sony A1 ebook:

"The optimum exposure when shooting in the Raw file format is to Expose to the Right (ETTR). Adjust the exposure compensation dial until the highlights are bright, but not overexposed. Exposing to the Right will provide the maximum image quality to the shadows in the file. You can then fine-tune the exposure in your Raw editing software ... Establish a manual exposure where the brightest white clouds in the sky are bright but not overexposed. When the Zebras appear, lower the ISO setting until the highlight tones are protected."

He describes a method of determining the amount of highlight recovery you can achieve with your camera. You then set the zebras level to this setting ( zebras are set to 109 on my A1). One thing to note is that in some modes all exposure composition is doing is adjusting the ISO. In those cases you need to make sure you compensate with aperture or shutter speed adjustments. This ensures that you are capturing maximum amount of image data ( or "signal in signal/noise ratio). ETTR is not simply a base ISO concept.
 
Last edited:
This is a good distinction. I remember years ago Nikon Capture NX2 did a horrible job of changing exposure more than a stop, but Lightroom could handle a couple of stops without an issue because it was reprocessing the RAW file. Lightroom later evolved into a product that could make much larger adjustments to exposure, and Nikon NX Studio handles multiple stops of exposure adjustment very well.

I looked back at some images from the past few years and found one with a D800E, 600mm lens, f/4.5 for DOF, 1/200 sec, and ISO 800. It was an image intentionally exposed to the right at ISO 800, but my shutter speed and aperture had reached the maximum I could safely use. The image - or a male silverback gorilla in a zoo - was midtones and very dark tones - and ultimately converted to B&W. I exposed to the right at capture to have detail in the shadows of the face and the dark hair of the gorilla, but I still ended up pulling back the exposure in the brighter areas - particularly the face.

Here are the two images, and you can see how the original capture was intentionally over exposed by a full stop or more, but pulled back for the final image. You can see how dark the face would be without overexposure.

View attachment 30651

The key to this image is the eyes with the rest of the face being darkened from the original. By the way - my wife would not let me hang this in our house. :)
View attachment 30652
Thank you for posting these - they are wonderful Eric! And I would hang them on my wall.

I think you answered my questions - more real world application for us creative types vs. techy is much appreciated. I have read many of the posts on this thread twice trying to understand how it works and have learned a lot on how to think in the field. I was shooting Grizzly in the snow in bright light. Images are in my Spring 2021 Tetons album. I used spot metering on the.face to achieve the exposure and this seemed to produce good files in post. The files in playback looked over exposed even on the bears. Snow was blinking but I was able to bring back detail here in most cases. I did experiment with bringing exposure down a bit to reduce (not eradicate) blinkies in the snow This was on the D850 or D5, 500mm pf or 600mm f4, lowest aperture and shutter speed I could mange and then adjusted ISO. Sometimes they moved into shadow so I stayed with exposure for the face. Thanks for any additional insights or correction to my thinking.

I have recently moved to Sony and watched Mark Gayler. Zebras (set at 109) seem to work really well for landscape and wildlife shooting but I have no idea how my above example would work in this system - in my very limited experience so far. Any input would be gratefully received. I am headed to San Diego to shoot Grebes rushing next week with new to me gear.
 
Last edited:
I think of ETTR in practical terms that I use in real world shooting. Sony Ambasdador Mark Galer obviously does the same. Here are a few quotes from his Sony A1 ebook:

"The optimum exposure when shooting in the Raw file format is to Expose to the Right (ETTR). Adjust the exposure compensation dial until the highlights are bright, but not overexposed. Exposing to the Right will provide the maximum image quality to the shadows in the file. You can then fine-tune the exposure in your Raw editing software ... Establish a manual exposure where the brightest white clouds in the sky are bright but not overexposed. When the Zebras appear, lower the ISO setting until the highlight tones are protected."

He describes a method of determining the amount of highlight recovery you can achieve with your camera. You then set the zebras level to this setting ( zebras are set to 109 on my A1). One thing to note is that in some modes all exposure composition is doing is adjusting the ISO. In those cases you need to make sure you compensate with aperture or shutter speed adjustments. This ensures that you are capturing maximum amount of image data ( or "signal in signal/noise ratio). ETTR is not simply a base ISO concept.

The Sony A1 is a good example. If you look at the way shadows improve with in camera iso (versus brightening in lightoom) using the chart linked below, it seems that it has to be a dual gain system. No improvement to 400, then a one stop leap at 500 that stays flat until 12800, then a tiny improvement until 102000.

So if one had that camera and plenty of time to think about ettr exposure, to me it seems it would always be useful to use shutter speed and aperture to push to the right, and if a normal capture has you at 100 to 400 then using iso to push to the right can give one more stop of help, but if a normal capture already is at 500 or above, there is nothing more for the system to give by cranking up the iso.

Plus that camera already is showing more than 11 stops of dynamic range anyway.


 
Last edited:
Wonderful image!

Your post has me thinking that we all mean a different thing with ETTR. :)

In the example you posted, you'd need another stop, probably two, to qualify for what I think of as an ETTR. IMO, what you did was just smart metering :)

I believe a "proper" ETTR (if there is such a thing) is normally done by identifying the brightest area in the image and exposing to just a bit before clipping. Not clipping in the RAW file, clipping in the JPEG preview. The typical method used is to test your camera and see how far you can go past where it shows clipping for whatever picture profile you use, then add that amount of exposure to the image once you have the exposure there.

For instance, I can actually shoot my D850 1.7 stops past the point where I see clipping (I've tested it). So, if I see clipping and want to do an ETTR, I'd set my exposure so it's just at that edge of clipping. That's my starting point. Then I'd add 1.7 stops. Once home, I'd pull back in Lightroom.

I think what most are describing in this thread is simply adding a little extra exposure as needed, not ETTR as I understand it. I think that's why there are so many differing thoughts :)
Of course Steve is right - I just did not have any images like that handy and I went through 8 years of images. :) This particular image was from 2013 - a time when the Exposure Slider did not provide the latitude it does today. The other complication was with wildlife, you don't have the ability to slow the shutter enough to be at base ISO so it's another compromise.

For a better ETTR example image, I was looking for a midtone image at base ISO with a shutter or aperture adjustment to brighten the image with the expectation I would have more data captured and pull the image back down to a correct exposure in post. Even if it does not approach clipping, it's brighter than a correct exposure.

What I found out is I don't actually use ETTR very often when I take a photo. :)
 
Back
Top