What is the best exposure for your Olympus? (ETTR and other techniques)

mfturner

Mu-43 Regular
Joined
Aug 6, 2019
Messages
69
I've done a few experiments too, with both the m10.3 and the old PM1.

First lesson learned, ISO brkt on the PM1 appears to use trickery, you get one shutter and sensor reading for all 3 files, and although OWS acts like the 3 raw files are different, ufraw and dark table (which may use the same raw conversion engine, I don't know) acts like they are identical. Almost identical, anyway, jpegs saved from them and compared using imagemagick show very slight differences, so who knows.

On the other hand, using manual mode on either seems to work fine, and my very early thought is that a couple of stops under exposed isn't as big a penalty as underexposing the old Canon sensor's pattern noise if you are saving raw. Too much underexposure shifts shadow colors as you would expect. Whereas overexposing by more than a stop progressively blows out highlights, as you would expect.

But even the ancient PM1 has more latitude for underexposure than I expected. The m10 has about a stop less high ISO noise, which matches the dpreview pictures I compared before buying the PM1. So I'm pretty happy...
 

ibd

Mu-43 Veteran
Joined
Jun 5, 2016
Messages
416
At first, I was intrigued by the idea to bias the metering system to overexpose, then use the in-camera picture controls to automatically bring the exposure back down to normal levels. In theory, it could be an "automatic ETTR" mode with usable JPEGs right out of the camera. I experimented with a bias of +1EV, which I then brought back down to 0EV apparent output exposure using in-camera tools. All on Olympus E-M1 II.

Then I realized that it doesn't really add much. Let's go through it mode by mode:
- P mode (ISO fixed): Just lower ISO by 1 stop to achieve the same benefit as the "automatic ETTR" above. A & S will adjust to let in 1 more stop to compensate the lower ISO.
- A/S modes (ISO fixed): Same as above, but either A or S can be fixed by you.
- M mode (ISO fixed): Same as above, but you'll have to adjust A/S yourself.

Now, it changes a bit when choosing "auto ISO" in P/A/S modes. Indeed, when the metering is pushed to let in one more stop of light, the camera will sometimes choose to lower the ISO accordingly. However, this control is indirect at best, and unpredictable at worst. As soon as you hit the limit of what the camera is willing to adjust in A & S, it will instead increase ISO to give you the desired (over-)exposure. This is always the case in M mode with auto ISO -- there you don't gain anything except, undesirably, earlier highlight clipping (assuming an ISO-invariant sensor, which seems to be the case for the E-M1 II according to your article).

Finally, ETTR can be helpful if you want the lowest possible noise and your sensor's dynamic range exceeds the scene's. Then you can go to the base ISO, do ETTR, then bring it back down in post. Again, however, Olympus already solves this problem by providing ISO 100 and ISO 64 (approx. 1 and 2 stops below the sensor's base ISO, respectively) as a camera setting, achieving exactly the same as ETTR (lower noise at the cost of less head room in the highlights).

In conclusion, what you propose doesn't help me in any photographic scenario I've encountered. I'd rather try to remember "keep ISO fixed at the lowest you can get away with" than to fiddle with nudging auto ISO to where I want it.

Still, I appreciate your efforts and the inspiration it gives. :)
 

Matt Drown

Mu-43 Veteran
Joined
Dec 9, 2015
Messages
324
Location
California, USA
Again, however, Olympus already solves this problem by providing ISO 100 and ISO 64 (approx. 1 and 2 stops below the sensor's base ISO, respectively) as a camera setting, achieving exactly the same as ETTR (lower noise at the cost of less head room in the highlights).
Just an FYI when you are using 100/64, these are just olympus "pulling" the image, and are shot at iso ~200 (base iso) and in-camera pulled down to 100/64. You lose dynamic range when you do this.

My takeaway mirrors yours (no pun intended), "Keep the ISO as low as possible for the type of picture you are taking".
 

ibd

Mu-43 Veteran
Joined
Jun 5, 2016
Messages
416
Just an FYI when you are using 100/64, these are just olympus "pulling" the image, and are shot at iso ~200 (base iso) and in-camera pulled down to 100/64. You lose dynamic range when you do this.

My takeaway mirrors yours (no pun intended), "Keep the ISO as low as possible for the type of picture you are taking".
That's exactly what I wrote. My "less head room in the highlights" equals your "you lose dynamic range".

One more thing I'd like to add: I think it's wrong to claim every sensor has a "sweet spot". No such thing. Give a higher exposure to a sensor and you will get less noise, period. In practice, there are rare exceptions for dual gain sensors where you might just push the automatic system to jump to a slightly higher ISO which still has better noise characteristics.
 
Joined
Nov 21, 2015
Messages
272
Location
The Netherlands
Real Name
Roel
Finally, ETTR can be helpful if you want the lowest possible noise and your sensor's dynamic range exceeds the scene's. Then you can go to the base ISO, do ETTR, then bring it back down in post. Again, however, Olympus already solves this problem by providing ISO 100 and ISO 64 (approx. 1 and 2 stops below the sensor's base ISO, respectively) as a camera setting, achieving exactly the same as ETTR (lower noise at the cost of less head room in the highlights).
Cool to read your tests and efforts to see if "automatic" ETTR. To bad that it's not the case as it would be a nice benefit for people wanting jpeg SOOC.

I would add that ETTR is not only focussed on reducing noise but can also increase the tonal values you can capture in the final image. Overall noise reduction, as well as any effects on tonal values, is based on the specific scene. If the histogram with regular exposure has a fairly broad spread there is little to gain in ETTR. If your histogram is skewed to the left (and can push the image quite a bit with ETTR without clipping) it will be more beneficial.

The extended "low" setting is basically emulating this to an extent. A lower then base ISO is increasing the exposure time by lowering the voltage to the sensor reducing both noise but also signal. The manufacturer does not consider this the best setting for sensor performance (otherwise this would have been "base ISO"). You could regard this as "semi-automatic ETTR" (but at the expense of highlight headroom), which does work but depending on the scene manual ETTR and manual post-processing the RAW file can archive better results.

In conclusion, what you propose doesn't help me in any photographic scenario I've encountered. I'd rather try to remember "keep ISO fixed at the lowest you can get away with" than to fiddle with nudging auto ISO to where I want it.
Agreed, ETTR does have a specific use-case and in my opinion, is certainly not a solution for all photography situations. It's just a technique that helps in some situations the same as for instance underexposing a snow-covered scene.

I use it quite often for my landscape photography when the camera is on a tripod and I have lots of time to set up the shot. For regular photography (portraits/travel/street) I don't bother.
 
Last edited:

Hannety

Mu-43 Regular
Joined
Aug 15, 2019
Messages
108
One more thing I'd like to add: I think it's wrong to claim every sensor has a "sweet spot". No such thing.
True. Cameras may well have a 'sweet' ISO spot which when forced to underexpose further increase does not reduce noise, only decrease dynamic range.

I would like to know if/what that is on a G9.
 
Joined
Mar 31, 2015
Messages
216
Location
Switzerland
Real Name
Siegfried
Gents I appreciate each reply, thank you. Its easier for me to give one reply. I will try summarize all the comments...

First
Here are the basics from the article:
  1. The image sensor is an electronic component & has an optimum operating range (sweet spot)
  2. Inside this sweet spot the sensor performs at its best (ISO, tonal, IQ, DR, image details)
  3. Outside the sweet spot things will go wrong (this vary from sensor to sensor)
  4. Aperture and shutter speed controls exposure (ISO only adjust image brightness)
  5. Aperture and shutter speed are in-camera controls, they do not control the external environment like a flash would do
  6. It is NOT important how the ISO setting is connected in the camera or why the noise increase when the ISO increase (It's interesting for those who likes to know)
  7. What is important - When the ISO setting increase the noise will increase
Is this a major revelation? Depends on the reader, some already apply this knowledge for years and for others it's completely new. My aim is the second group,

How do we apply this knowledge:-

I described these steps:-
  1. Understand the problem and what we can do to solve it
  2. Master your cameras features and how to apply them (histogram, highlight/shadow blink, IBIS, ISO min/max settings)
  3. Use or apply an Exposure Profile (practical how-to solution)
  4. Techniques like ETTR targeting shutter speed & aperture
  5. Software Solutions to reduce image noise
I did not discuss any external exposure modifications like using a flash...

General comments:-
  • Cameras are different - test your camera with under exposing and over exposing (-2EV, -1EV, 0, +1EV, +2EV) (use shutter or aperture, KEEP the ISO fixed!!)
  • The G9 will also benefit - I will have a G9 for a few days :) - If I recall the G9 curves does not adjust the mid tones?
  • You will see from the test images the G9 response was good - try do the above EV test - do it at different ISO's
  • My wife has an EM10 III and asked me about "noise" in her images? I lowered the auto ISO high limit to 1000 and now she is happy. I could also ad a (0.5EV) exposure lift?
  • Should I use ETTR with high ISO's? - Absolutely, you will always see benefits having the sensor operating in its "sweet spot", also at high ISO's
  • Ibt said he lower the ISO with 1 stop to get the same result - True & false - all you did was to reduce the high ISO noise effect
  • Ibt if you want to see the effect on the sensor only use shutter speed & aperture
  • Ibt your feedback is important as it confirms why I wrote the article - thank you
  • Using curves to have a OOC ready jpeg - absolutely
  • If you leave your ISO setting at default - you will see how quickly the camera increase the ISO
  • In most daylight situations a max of ISO400 is OK - if left to the camera you will see ISO800 or ISO1000 - that's why its important to always check the ISO
Final comments:-

- You in charge - I use a day to day exposure profile, it just makes my life easy - but I also use the histogram and highlight shadow blink to always check before I take an image.
- I use Aperture Mode for my day to day photography (80% of the time)
- The rest of the time I use Shutter Priority (night or blue hour) or Manual Mode (in manual I use all manual including ISO) - I will apply some level of ETTR ranging from 0,5EV to 2,5EV
- I keep the auto ISO setting low in my camera as I learned to rely more on IBIS
- Now there are times that I will up my ISO to as high as ISO6400

Have fun, I will write shorter practical example in the future - currently its cold over here and raining

Siegfried
 
Last edited:

mfturner

Mu-43 Regular
Joined
Aug 6, 2019
Messages
69
@BushmanOrig, I have been convinced for some time that increasing the actual exposure (e.g. luminance, aperture and SS) increases the signal, thus improving the SNR. So we are in agreement here I think.

My typical scenario is: the scene is dark and I cannot do anything about lighting (e.g. flash), my aperture is wide open, and my SS is as slow as either the subject or my shaky hands allow. What is the best ISO setting with these constraints? So my experiments have been focused on best ISO setting given fixed aperture and SS.

Additionally, while I'm not afraid of raw, I would like to avoid raw if possible when I travel on vacation, so that I don't have to bring my laptop along (part of my overall gear down-sizing for travel). I would happily tune the camera's jpeg settings if this helps to get useful photos sooc for webviewing (for printing I can do more work on the image after the trip at home if I wish). Much of this depends on your own tolerance of noise detail vs blurring with noise reduction SW.

Playing in Darktable with some high-ISO raw images of my obliging dog and cats, I noticed that the base curve used for converting raw to jpeg sets the local contrast, which in turn can make noise more apparent where you don't want it. I can affect the "chunkiness" of the noise depending on where the steepest part of the base curve is (e.g. the regions of highest tonal contrast). Many base curves (including Darktable's Olympus ones) are quite steep in the darkest couple of stops, although it can look unnatural if you modify it too far. There is also the idea of "crushing the blacks" that most people dislike, but I do think it can be helpful to reduce the contrast for a few of the darkest f-stops.

In order to emulate this in-camera, especially for the ancient PM1 which does not have a "mids" adjustment but only "highlight" and "shadow" adjustments, I started experimenting with the "gradation" setting. Not knowing exactly what they do, I have previously left it set to "Normal" on both cameras. I reasoned that the "low key" and "high key" might push the steepest part of the S-curve left and right in the range of values, and this appears to be so. It appear to have less chunky noise in the darker sections when I set it on low key, similar to when I softened the initial climbing of the base curve in darktable. I did not try the "AUTO" setting, which appears to extend the base curve and reduce overall contrast like a psuedo HDR photo. This seems like the opposite of what I want at high-ISO.

A baseline ISO 10000 photo with some of these modifications from my ancient PM1 gives this at the 1600 px forum resolution:
Subscribe to see EXIF info for this image (if available)


Pushing ISO up (ETTR in that sense) simply blew out more of her forehead and foreleg highlights. Pushing ISO down (ETTL) gave color shifting of the dark background as one of the color channels started zero-ing before the others. Up to 2 stops of ETTL did not significantly increase the noise at 1x magnification. If you scan between the two images you should be able to see that the shadow background is magenta shifted below, and green shifted above. You should also see more detail in her forehead and foreleg below.

Subscribe to see EXIF info for this image (if available)


This for a 1x crop of the dog's head. Not something I would try to use at 1x, and you might not want to use any of these for anything, but for me these are OK for web sharing at reduced resolution, despite some of the crazy color shifting. I could depress the color shifting with a saturation adjustment easily.

Subscribe to see EXIF info for this image (if available)
 

pdk42

One of the "Eh?" team
Joined
Jan 11, 2013
Messages
6,707
Location
Leamington Spa, UK
Siegfried - I applaud your efforts here and producing the article you have.

But - it's a long article and IMHO the point of ETTR is really quite simple, and very well documented. It simply means increasing the exposure to the point where you are almost, but not quite, saturating any of the pixel sites. If that results in an image that looks bright when you view it, then reduce the brightness in post processing - and the result will be an image where noise is reduced in the darker areas because reducing the brightness also reduces the "brightness" of the noise. Simple!

Of course, ETTR only makes sense at base ISO. If you can increase exposure to move the histogram right with impunity (keeping a safe shutter speed etc), then you can reduce ISO and that will improve things much more effectively (because if you're increasing ISO you're doing so to amplify a signal from the sensor that hasn't received enough light and since amplification introduces noise, you'd be best advised to fill the sensor first - i.e. get more light on it!).

Achieving that in practice is harder than it seems, because:
  • Every production camera I know shows you the histogram/blinkies/zebras etc based on the "jpeg" interpretation not the raw. You can see this easily since changing the JPEG tone curve will change the histogram without changing the exposure (and therefore without changing the raw).

  • The histogram is usually too crude to show small areas of highlight peaking. These may be important in your image.

  • Blinkies/zebras only show an RGB composite view, so you might actually be blowing highlights in, say, the red channel but the merged view will mask it.

The only real solution is to experiment and get to know your camera as best you can - and if in doubt, do an exposure bracket. Or shoot a sequence and stack later.

Then it's also worth pointing out a few semi-obvious things about m43 cameras:

  • The pixel sites are like buckets for collecting photons (the "full well photon count") and m43 sensors have smaller buckets than larger sensors. This means that it's easier to blow highlights on m43 sensors so the technique is less effective than it might be. Blown highlights are, IMHO, about the ugliest thing you can do (unless you are artistically trying to blow out an area on purpose).

  • Every m43 camera I've owned (Olympus and Panasonic) is cautious about over-exposure and therefore they all tend to underexpose. It's good to know this. Personally, I always ignore the camera's metering and just use exp comp and the blinkies to set the right exposure.

  • The blinkies are conservative (and driven off the JPEG). I think setting the threshold to less than 255 is counterproductive since it'll get you to think you're blowing highlights when you're not.

Finally, when all's said and done, these technical aspects of photography never fully make up for getting the basics right - subject, composition, lighting, focus, DOF, no motion/camera-shake blur!
 
Last edited:
Joined
Mar 31, 2015
Messages
216
Location
Switzerland
Real Name
Siegfried
@BushmanOrig, I have been convinced for some time that increasing the actual exposure (e.g. luminance, aperture and SS) increases the signal, thus improving the SNR. So we are in agreement here I think.

My typical scenario is: the scene is dark and I cannot do anything about lighting (e.g. flash), my aperture is wide open, and my SS is as slow as either the subject or my shaky hands allow. What is the best ISO setting with these constraints? So my experiments have been focused on best ISO setting given fixed aperture and SS.

Additionally, while I'm not afraid of raw, I would like to avoid raw if possible when I travel on vacation, so that I don't have to bring my laptop along (part of my overall gear down-sizing for travel). I would happily tune the camera's jpeg settings if this helps to get useful photos sooc for webviewing (for printing I can do more work on the image after the trip at home if I wish). Much of this depends on your own tolerance of noise detail vs blurring with noise reduction SW.

Playing in Darktable with some high-ISO raw images of my obliging dog and cats, I noticed that the base curve used for converting raw to jpeg sets the local contrast, which in turn can make noise more apparent where you don't want it. I can affect the "chunkiness" of the noise depending on where the steepest part of the base curve is (e.g. the regions of highest tonal contrast). Many base curves (including Darktable's Olympus ones) are quite steep in the darkest couple of stops, although it can look unnatural if you modify it too far. There is also the idea of "crushing the blacks" that most people dislike, but I do think it can be helpful to reduce the contrast for a few of the darkest f-stops.

In order to emulate this in-camera, especially for the ancient PM1 which does not have a "mids" adjustment but only "highlight" and "shadow" adjustments, I started experimenting with the "gradation" setting. Not knowing exactly what they do, I have previously left it set to "Normal" on both cameras. I reasoned that the "low key" and "high key" might push the steepest part of the S-curve left and right in the range of values, and this appears to be so. It appear to have less chunky noise in the darker sections when I set it on low key, similar to when I softened the initial climbing of the base curve in darktable. I did not try the "AUTO" setting, which appears to extend the base curve and reduce overall contrast like a psuedo HDR photo. This seems like the opposite of what I want at high-ISO.

A baseline ISO 10000 photo with some of these modifications from my ancient PM1 gives this at the 1600 px forum resolution:
View attachment 800379

Pushing ISO up (ETTR in that sense) simply blew out more of her forehead and foreleg highlights. Pushing ISO down (ETTL) gave color shifting of the dark background as one of the color channels started zero-ing before the others. Up to 2 stops of ETTL did not significantly increase the noise at 1x magnification. If you scan between the two images you should be able to see that the shadow background is magenta shifted below, and green shifted above. You should also see more detail in her forehead and foreleg below.

View attachment 800383

This for a 1x crop of the dog's head. Not something I would try to use at 1x, and you might not want to use any of these for anything, but for me these are OK for web sharing at reduced resolution, despite some of the crazy color shifting. I could depress the color shifting with a saturation adjustment easily.

View attachment 800380
Wow Mftuner this is interesting, one would never have expected to see a PM1 (EP1) image at ISO1000 in the past unheard of....? One does see the noise in the last image you posted but still very acceptable...

I think you demonstrated exactly what it's all about. What I wanted to point out in the article is the sensor is the one component that can really make a difference. That is why a good understanding of exposure (shutter & aperture) are so important and then separately how ISO adjust the final image brightness but also at a price (noise)

But there are so much more.... And it is this "so much more" you pointed out so well... I was in two minds to spend more time and add more to the article but it would just be too much - What you did with this reply is therefore very good. If I may dream.....my dream would be - instead of useless sensor size talk, change the subject and we all master our gear....perfect world -:(

You touched on a few interesting concepts:-
- Use a flat exposure profile
- Correct the profile in the camera for out the camera ready images

Flat exposure profile:
- I agree with you that in-camera settings like contrast and sharpness will increase image noise. I have seen the same thing.
- Obviously you need to experiment - what if you create your own "flat" profile which you can reverse again in the camera or using darktable online (you did try part of this)
- The one thing you did not mention is using the camera editing functions to edit the raw file or reverse the flat profile.
- Interesting to experiment with gradation - have not tried that myself...
- I have an old PM2 and I quickly tried in camera editing - not much one can do. I wonder if when one expose the same image different and then overly them if that would do anything - only thinking out loud...
- The picture you show is also that extreme 5%, right? This is not day to day stuff?

Correct the profile in the camera:
- I think there is an important concept one could try. It is editing in the camera and that has two parts
Part 1
- Edit while taking the image (curves, art or scn filters - in effect you editing the jpeg file)
Part 2
- Edit in the edit menu under the image review menu (the older PM1 maybe has very little options)

Another option is to get something newer like the EP5... I think we so spoiled with great older cams today, some at really give-away prices - maybe one should list the best older models versus price?

One last comment - seem you exposed for the complete image? What of you use center weighted exposure and focus on the dog's head, that way you could drop the ISO one or two stops and ettr with a half or one stop? You could not ettr in this image because part of the image was already blown out...

Then I saw the old sensors are sometimes more challenging in shadows than newer sensors...

I sometimes call what you talk about as tolerance.....being creative..:)

Apologize for ma long reply - your image and post inspired me..

Siegfried
 

SojiOkita

Mu-43 All-Pro
Joined
Feb 23, 2014
Messages
1,523
Location
France
Achieving that in practice is harder than it seems, because:
(...)
I never saw a digital camera having a "ETTR" exposure mode (based on RAW data, of course).
I don't know if it exists. In theory, this seems not so difficult to do... maybe it is...
 
Joined
Mar 31, 2015
Messages
216
Location
Switzerland
Real Name
Siegfried
Siegfried - I applaud your efforts here and producing the article you have.

But - it's a long article and IMHO the point of ETTR is really quite simple, and very well documented. It simply means increasing the exposure to the point where you are almost, but not quite, saturating any of the pixel sites. If that results in an image that looks bright when you view it, then reduce the brightness in post processing - and the result will be an image where noise is reduced in the darker areas because reducing the brightness also reduces the "brightness" of the noise. Simple!

Of course, ETTR only makes sense at base ISO. If you can increase exposure to move the histogram right with impunity (keeping a safe shutter speed etc), then you can reduce ISO and that will improve things much more effectively (because if you're increasing ISO you're doing so to amplify a signal from the sensor that hasn't received enough light and since amplification introduces noise, you'd be best advised to fill the sensor first - i.e. get more light on it!).

Achieving that in practice is harder than it seems, because:
  • Every production camera I know shows you the histogram/blinkies/zebras etc based on the "jpeg" interpretation not the raw. You can see this easily since changing the JPEG tone curve will change the histogram without changing the exposure (and therefore without changing the raw).

  • The histogram is usually too crude to show small areas of highlight peaking. These may be important in your image.

  • Blinkies/zebras only show an RGB composite view, so you might actually be blowing highlights in, say, the red channel but the merged view will mask it.

The only real solution is to experiment and get to know your camera as best you can - and if in doubt, do an exposure bracket. Or shoot a sequence and stack later.

Then it's also worth pointing out a few semi-obvious things about m43 cameras:

  • The pixel sites are like buckets for collecting photons (the "full well photon count") and m43 sensors have smaller buckets than larger sensors. This means that it's easier to blow highlights on m43 sensors so the technique is less effective than it might be. Blown highlights are, IMHO, about the ugliest thing you can do (unless you are artistically trying to blow out an area on purpose).

  • Every m43 camera I've owned (Olympus and Panasonic) is cautious about over-exposure and therefore they all tend to underexpose. It's good to know this. Personally, I always ignore the camera's metering and just use exp comp and the blinkies to set the right exposure.

  • The blinkies are conservative (and driven off the JPEG). I think setting the threshold to less than 255 is counterproductive since it'll get you to think you're blowing highlights when you're not.

Finally, when all's said and done, these technical aspects of photography never fully make up for getting the basics right - subject, composition, lighting, focus, DOF, no motion/camera-shake blur!
Hello Paul
Thanks. I think your feedback is very relevant....its comes back to know your camera, right? Even if the histogram display image brightness you can still see the rgb channels and if you practice and know your camera it would not matter that much. It will be really nice if Olympus could give us a luminance histogram....

Then I agree nothing is new.... :cool: That said I was so surprised when I searched ettr and I saw how many youtubers explain ettr and using ISO - That just showed me....we could never talk enough about this?

Take the example with the post with the dog at ISO10000 with a PM1 - I think we could learn so much from others....

Best
Siegfried
 

mfturner

Mu-43 Regular
Joined
Aug 6, 2019
Messages
69
@BushmanOrig, Yes, this is only an experimental photo, I have plenty of ISO 200 photos that are well lit and well exposed of this dog. Your PM2 should have the same sensor as my M10.3, so maybe 1/2 stop better low light performance than this.

I may experiment some more to see if I am correct that I can get something like this SOOC, and to see if I can get something "web sharable" at ISO 25600 with the M10.3. A lot depends on the image processing order, I don't yet know if the in-camera processing order is the same as with OWS. The sun has to come out to give me a high dynamic range image though, it's snowing right now LOL.
 
Joined
Nov 21, 2015
Messages
272
Location
The Netherlands
Real Name
Roel
I never saw a digital camera having a "ETTR" exposure mode (based on RAW data, of course).
I don't know if it exists. In theory, this seems not so difficult to do... maybe it is...
It is quite hard to implement as the camera can not really "see" the scene.

To put it very simplistically (not the exact scientific explanation but it will do for the argument): the camera has to go on the interpretation of the brightness of the scene and try to get it as balanced as possible (= least chance of blown highlights and severe underexposed areas).
This approach results in proper exposure in most situations as most scenes combine highlights with shadows.

But if, for example, you shoot a very bright scene e.g. lots of snow, the camera simply doesn't know you are shooting snow and will underexpose the scene quite a bit as it is still trying to balance the exposure. That's why some cameras include a scene selection option so you can tell the camera that it's a high key photograph you want to take.

I expect that with faster processors cameras can implement (more) machine learning algorithms that enable the camera to see (interpret) what you try to photograph and change its settings based on the subject (and not only some crude information around exposure). So if you are shooting in an overly white environment it would know this and set the proper exposure.
 

pdk42

One of the "Eh?" team
Joined
Jan 11, 2013
Messages
6,707
Location
Leamington Spa, UK
I never saw a digital camera having a "ETTR" exposure mode (based on RAW data, of course).
I don't know if it exists. In theory, this seems not so difficult to do... maybe it is...
I've pondered that a lot of times. I assume that the rendering of the raw data into something you can see is performed in LSI circuitry in the camera. That being the case, it's possible that getting at the raw values to produce a raw histogram is not a trivial matter. However, that's all speculation.
 
Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Mu-43 is a fan site and not associated with Olympus, Panasonic, or other manufacturers mentioned on this site.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Copyright © 2009-2019 Amin Forums, LLC
Top Bottom