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

[BushmanOrig]

@relic, @Rob Trek, @Machi and other.....

Few Points:-

1. Whatever theoretical explanation you make, please add the following info to your conclusion:
- Add a how-to apply this new information practically with your camera
- What benefits/improvements can we expect from this technical know-how

If this basic requirement is not met, then the technical explanations, other than interesting, is just wind and useless.... Nobody can claim noise is only photons or only this or only that. It always depends on who created the narrative. I included a paper in my article by the sensor manufacturer Photometric that explains there are more than a photon theory. I explained why I believe Canon picked the photon theory at the time.... Sony later trampled all over this narrative and proofed there are more than photons when they launched the A7 III? So again I never said the Photon theory is false - my question just is, is the magnitude of the photon theory really that important?

2. Some of you commented saying - So what is new? Saying we all know this, we do not need to discuss this again.
Absolutely, I see it's clear from all the replies....everybody is familiar with the basics and on the same page? No need to talk about this ?

3. ETTR is only a workable solution at base ISO. If my camera base ISO is 200 then ETTR will ONLY work with images at ISO200. Then a long theory of photons, full and half full buckets... That's all perfectly OK if you believe that is important and if you ONLY use ETTR at base ISO.
My advice to readers is, go try ETTR for yourself with your camera and find the solution (ISO and EV) that will make a difference and that will work for you. It really is that simple....

3. Rob, thank you for your feedback - your question, why using a fixed shift exposure? Here my reasoning

• I noticed some of my Olympus cameras generally under expose a little in Auto mode.
• It does not really matter if you program +0.5EV or +1EV as an exposure shift, you can always pull it back with exposure compensation
• Used together with the histogram and the over expose screen warning, those trying the profile will quickly develop a good feel for ETTR
• It also depends on what exposure measurement method you use - ESP, spot or center weighted
• The more I try this myself, I probably will end up with a 2/6 setting for my camera exposure shift setting
• Another possible option is to always bracket, -0.5 / 1 / +1.5EV just looks like a lot of images. But will be great for HDR enthusiasts

4. I am not going to get into technical debates because it helps nobody - especially if you did not take the time to study my article. I spend weeks preparing it, studying photons and all the other theories. The final information I put together and presented will give those interested a more solid base to work from, it answers in-depth questions BUT most important my conclusion has a workable how-to proposal people can try and work with. If you studied my info and you believe I made a fundamental mistake, PLEASE point it out and I am more than happy to correct mistakes.....

Here are 3 more examples:

Example 1

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

This was my thinking process when taking this image:-

1. I knew I had to up the ISO so I selected an ISO that would work for me and the image - I decided for ISO800
2. Selecting the ISO - I considered things like IBIS (how it would help me), next the increase in noise as I up the ISO (try keep the ISO low)
3. I also know ISO has nothing to do with exposing the sensor so my next step was selecting the shutter speed & aperture to expose the sensor
4. We know that its only the shutter speed and the aperture that will expose the sensor, the sensor has a optimum operating (exposure) range
5. My camera was set to Aperture Mode and it automatically selected the shutter speed for the ISO (Image brightness) and the aperture I selected
6. My first image was taken with the camera auto selecting the exposure setting.
7. My second image was exposed to the right (ETTR) at +1EV
8. The ETTR image exposed the sensor better and had less noise in the shadows.
9. I would not normally take my image like this, I prefer a tripod and bracketing to later convert into an HDR image, but this was handheld and OK

Example 2

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

This was my thinking process taking this image:-

1. ISO was 200, my camera exposure set at "center weighted" and the camera in Aperture Mode
2. The camera screen over exposure warning showed me the buildings in the back was close to over exposing
3. I then took a -0.5EV image, normal exposure and +0.5EV
4. The image you see is the +0.5EV (ETTR) image
5. The EM1 II dynamic range was perfectly OK to pull back the highlights and up the shadows.
6. The shadow quality in the +0.5EV image was better than the -0.5EV image

Example 3

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

Few comments on this image:-

1. This image was edited for Instagram, so the sky is a little overdone....but
2. Same thing, camera at ISO200, I used Aperture Mode and I tried different exposures
3. Camera auto exposure and +0.5EV and +1EV (ETTR) (Only changing the shutter speed NOT the ISO)
4. I used the +1EV image above
5. Again the EM1 II dynamic range was more than enough to pull back the highlights in RAW and have clean shadows

Conclusion:-

Guys its basic. Whatever theory, Photons, sensor, gain amplifier, ADC, dark currents, read noise.......does not matter what. Every time you apply ETTR the shadows improve. Add to this the know how that it's only the shutter speed and aperture that will expose the sensor. We also know image sensors are at their best when exposed correctly or a little over exposed. Master these exposure skills and ETTR and you will have a winning new exposure technique.

ISO is a little more complex to explain. That said, all you need to know is, ISO has nothing to do with exposure (exposing the sensor), ISO adjust image brightness and generally the higher the ISO the more noise we see in the image (test this with your camera) - the final solution is obvious? (Use IBIS to help you keep your day to day ISO low - do not allow the camera to go wild using Auto ISO)

You do not need to go buy a full frame camera or an APC camera to get better IQ - You will probably get better results from applying good exposure techniques using your MFT camera. Changing your camera for a FF or APC camera is NO guarantee for getting better IQ, especially when you never mastered the basics on exposure.....

And this is the reason why repeating this "simple and basic" advice again and again, is important.... it will ALWAYS benefit MFT and Olympus owners.... this is also why it is BIG news for many many Olympus owners..... not that difficult to understand, right?

Best

Siegfried

 

[PakkyT]

4. I am not going to get into general technical debates because it helps nobody - especially if you did not take the time to study my article. I spend weeks preparing it, studying photons and other and the final info I have put together and presented I feel will give those interested a good base to work from, it answers deeper questions BUT most important my conclusion has a workable how-to proposal. If you studied my info and you believe I made a fundamental mistake, please point that out and I am more than happy to correct any mistakes.....

With all due respect, this is a cop-out. First line basically says, 'I don't care if I am right or wrong, I spent a lot of time putting this together, therefore you should just accept what I wrote.' Several of us did study your article and even re-read it on following days to see if we maybe misread or missed things the first time. Just because we don't agree with some of your "findings" doesn't mean we are automatically incorrect or didn't read it nor that we shouldn't feel free to comment in a thread you started to present your article and I assume to receive feedback.

I have no issue with you presenting some methods of shooting different scenes in a manner that works well for you and that others might want to try. But for something that you say is so basic, your article (and again this is meant to be helpful, not hurtful) is long, is often hard to follow (it jumps around at time to different ideas than the one currently being discusses in a given section), and makes some claims that are not accurate. If your intent starting this thread for for constructive criticism, then that is what we are presenting. If you really didn't want people to point out issues, then it was a mistake presenting it here and you shouldn't shoot the messengers.

 

[BushmanOrig]

With all due respect, this is a cop-out. First line basically says, 'I don't care if I am right or wrong, I spent a lot of time putting this together, therefore you should just accept what I wrote.' Several of us did study your article and even re-read it on following days to see if we maybe misread or missed things the first time. Just because we don't agree with some of your "findings" doesn't mean we are automatically incorrect or didn't read it nor that we shouldn't feel free to comment in a thread you started to present your article and I assume to receive feedback.

I have no issue with you presenting some methods of shooting different scenes in a manner that works well for you and that others might want to try. But for something that you say is so basic, your article (and again this is meant to be helpful, not hurtful) is long, is often hard to follow (it jumps around at time to different ideas than the one currently being discusses in a given section), and makes some claims that are not accurate. If your intent starting this thread for for constructive criticism, then that is what we are presenting. If you really didn't want people to point out issues, then it was a mistake presenting it here and you shouldn't shoot the messengers.

What are you talking about? Its now the third or fourth time you baiting me? You quote me and then claim I said the opposite I wrote.... is this normal for you...?
Do me a favor, stop the senseless talking and use real examples. Let me help you....write no other words than for example:-

"Siegfried I disagree with you, I do not believe ISO is a camera function"

I can take that and work with that.....but all the other confusing words and telling me what is my mind does not work for me - I can do nothing with that please....

 

[PakkyT]

OK I can see you are not open to help or constructive criticism, so I won't comment further other than to mention your made up quote of mine is not anything I said nor implied. If you would like specific examples of where I think you can improve your article, I would be happy to make suggestions, but not if you are just going to get angry about them.

 

[relic]

Many thanks to @mfturner, @pdk42, @Matt Drown, @Machi, and @BushmanOrig, for your replies and for taking the time to try to educate me. Unfortunately my brain is too dense to absorb all that information: I tried reading your responses several times but I think that I really need many, many more neurons... But I think I understand @pdk42's posts as they are what my understanding has been. Things were rather simpler in the days of push-processing Tri-X Thanks again to everyone.

 

[SojiOkita]

I'm not sure about that.
Let's take an example and say that you're histogram above is : 1/100 s - f/4 - ISO 200.
If we use the first method, you could achieve ETTR by choosing, let's say : 1/50 s - f/4 - ISO 200.
(that means that with 1/40 s, your sensor is overexposed).

Increasing the ISO only (1/100 - f/4 - ISO 400) is useless for ETTR.
However, what happens if you choose (1/50 - f/4 - ISO 400) or even (1/50 - f/4 - ISO 800).
You will have, on the JPEG, overexposed photos (+1 and +2IL) with clipped highlights.
But what about the RAWs? Will highlights be clipped?
When choosing respectively -1IL or -2IL exposure on you RAW software, will you recover the highlights?

The answer was probably obvious, but sometimes you have to test by yourself to be sure
So, yes, highlights are clipped on the RAWs when overexposed at high ISO.
(and they wouldn't have with the same shutter speed and aperture value, at base ISO).

So, ETTR has to be used at base ISO.

 

[BushmanOrig]

Many thanks to @mfturner, @pdk42, @Matt Drown, @Machi, and @BushmanOrig, for your replies and for taking the time to try to educate me. Unfortunately my brain is too dense to absorb all that information: I tried reading your responses several times but I think that I really need many, many more neurons... But I think I understand @pdk42's posts as they are what my understanding has been. Things were rather simpler in the days of push-processing Tri-X Thanks again to everyone.

Please do not give up. My English mixed with complex theoretical explanations on the forum all contribute to making something basic look complex. Each one contributing is trying to help.... When you focus on the following basics, you will be able to test for yourself what gives better or worse results:

- The camera ISO setting sets the light sensitivity (brightness) of the entire imaging system
- Shutter Speed and Aperture are the 2 camera controls that controls how much the sensor is exposed
- The 2 camera components that we can influence is the sensor and the ISO
- The sensor has a small optimum operating range & is controlled by the shutter & aperture
- When we up the ISO we generally see more image noise

The most basic and rewarding test you can do is:

- Keeping the ISO constant (set it to the base ISO and do another test at ISO3200)
- Then use Aperture Mode and try 5 different exposure settings from -2EV to +2EV (1/2 stops)

Study the RAW files.

You will quickly see if the noise improved in the shadow areas when you increased the exposure from +0EV to +2EV.... When you get the feel for this its practice and a process of visualizing what you do. Later you will also master the camera features designed to help you master ETTR....

 

[BushmanOrig]

Unfortunately I haven't much time now so just quickly. There isn't SNR noise, that's like saying signal-to-noise ratio noise. It's just SNR and it's exactly what it's saying it is - ratio of signal to the noise.
Noise has in photography many components but generally main of them are just two. Those caused by photon nature of the light (shot noise) and those caused by camera (read-out noise) which has many subcomponents (read noise, dark/thermal noise, pattern noise, etc, look at the Martinec' article).

Camera's noise is for the majority of practical application constant at given ISO (only for long exposures it will be larger by growing contribution of dark noise).
For modern cameras it's 1-10 electrons rms (~equal or less than 1 DN at base ISO).
DN means digital number and gives amount of output in the RAW file if opened linearly (without gamma correction). Exact number is dependent on ISO.
One can measure this in appropriate software (ImageJ, RAW digger?) for totally black image as standard deviation.

Photon shot noise is dependent on the amount of detected photons and it's equal to square root of number of detected photons. For 10 000 photons it's equal to 100, for 16 photons it's 4.
As photons are detected as electrons in the camera, it's easy to compare contributions.
For modern camera with standard deviation of read-out noise 5 electrons, photon shot noise has the same contribution to the overall noise at just 25 photons/electrons (square root of 25 is 5).

So read-out noise (noise inherent to the camera) will be visible only in cases of very low signal (in shadows).
In majority of cases, shot noise will be much larger.
Now magnitude of photon shot noise can be demonstrated also in the appropriate image of luminated homogenous target, for example white paper.

View attachment 800664

Here is quick demonstration for totally black frame and for white paper shot.
In the histogram look at standard deviation (STD) which measure variations (=noise) for given area.
As you can see, luminated target has much more noise (STD equal 10 vs 1 for black frame).
More noise in the shot of white paper is not so important because signal is much higher which means better SNR.
Better SNR means image which looks less noisy.

If one improves signal further by using ETTR technique it surprisingly leads to more noise in the image!
So why images taken by ETTR looks better?
Trick is in the fact that noise is growing by square root of increment and signal is growing linearly so SNR is growing too. For 2× times higher signal, noise will grow as square root of 2 which is equal to 1.41 and SNR will be better by factor 2/1.41 which is equal to 1.41 (square root of 2 again).
In fact for every increment of signal, SNR will be better by square root of it.
And that's why ETTR works, it improves SNR.

EDIT: I've deleted last part, where I wrongly stated that ETTR lowers noise. and I've also improved readability of text (hopefully).

I do the same....always try and improve the readability - not sure if it helps. Thanks for your info, I did study it and I do not fundamentally disagree. I went through my article again and did some updates, will have to implement a few more.... My aim was:
- To keep it as simple as possible (its still way to complex in terms of lingo and text)
- Develop a profile/model photographers can apply to improve IQ and image noise
- From statements we see on the web, image noise, ISO, technology and exposure are probably the most confused subjects on digital photography - that means 15 years of digital photography resulted in a generally poorly educated photography community - not a great report?

 

[roelwillems]

- From statements we see on the web, image noise, ISO, technology and exposure are probably the most confused subjects on digital photography - that means 15 years of digital photography resulted in a generally poorly educated photography community - not a great report?

It's still a battle for the number 1 spot vs "DoF as a property of sensor size"

 

[BushmanOrig]

Siegfried, I tried your settings in the Pen-F and they look good.

However (uh oh!), when I checked the ColorChecker's Black it is was crushed down into the RGB=30-40's and not around RGB=52 where I think it is expected. This was in sunlight. Could be raising the Shadow's curve to maybe a +3 might help, but haven't applied it to your settings yet to check (Seems all my cameras have some sort of + applied to the Shadows part of the curves as Olympus likes blacks buried which lose any shadow detail.). I will re-try it later as I have found my two Sekonic meters are disagreeing with each other for some reason (L-478D & L-858D) and need to re-cal them with the Exposure II chart.

Is there a way to apply all these settings to some memory in the Pen-F? I never use any of the memory settings as I forget what was in them.

Mack - see this info...might interest you

 

[relic]

Please do not give up. My English mixed with complex theoretical explanations on the forum all contribute to making something basic look complex. Each one contributing is trying to help.... When you focus on the following basics, you will be able to test for yourself what gives better or worse results:

- The camera ISO setting sets the light sensitivity (brightness) of the entire imaging system
- Shutter Speed and Aperture are the 2 camera controls that controls how much the sensor is exposed
- The 2 camera components that we can influence is the sensor and the ISO
- The sensor has a small optimum operating range & is controlled by the shutter & aperture
- When we up the ISO we generally see more image noise

The most basic and rewarding test you can do is:

- Keeping the ISO constant (set it to the base ISO and do another test at ISO3200)
- Then use Aperture Mode and try 5 different exposure settings from -2EV to +2EV (1/2 stops)

Study the RAW files.

You will quickly see if the noise improved in the shadow areas when you increased the exposure from +0EV to +2EV.... When you get the feel for this its practice and a process of visualizing what you do. Later you will also master the camera features designed to help you master ETTR....

Thank you. I'll also have to read your article another time (or, maybe, several more times) to try to understand a bit more-- a difficult subject for me..

 

[BushmanOrig]

Thank you. I'll also have to read your article another time (or, maybe, several more times) to try to understand a bit more-- a difficult subject for me..

Thanks....give me a day or two pls... I am doing a few updates....

 

[exakta]

In your mind he did it wrong because he adjusted ISO last where as you think you adjusted exposure last and therefore did the real ETTR.

In the example you gave, changing the ISO last was because there was no wider aperture available and (I assume) reducing the shutter speed was not an option either. Otherwise, why ever use a higher ISO?

 

[BushmanOrig]

I have done more research and studied several more references. What was always extremely difficult for me was to visualize how noise happens and then apply commonsense to create a simple and easy model I can apply when doing photography..... One read about photons, buckets half full and full or you presented with complex theories and graphs to find all is even more vague and complicated when done reading.... There has to be a simple way to explain it all.....

The people researching sensor behavior and writing about it are clever people and they quick to go into really complex calculations and explanations. I found two writers that is writing amazing articles. From what I learned, I was able to put together a simple model that is easy to explain and use on a day to day basis.

I am doing changes to my article and trust is will be done in a week latest. I am adding new info daily.

So again thank you to all who tried so hard to explain why they disagreed with me....

The interesting part is the basic exposure technique plus using Olympus exposure features does not change. ETTR is still a key element of the solution - the only difference is I am explaining it theoretically more textbook like.... Using ETTR at different ISO's is also a workable solution, in fact now more so.

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

In this graph shows the conversion from photons into electrons. Study the graph, its really self explanatory.... What is important:-

- The sensor is a linear device, this makes it easy to see sensor behavior as long as you stay below the saturation point
- You can see the sensor dynamic range on the image (dark noise to the saturation point)
- You also see the noise graph in RED (how noise happens)
- The usable signal to noise ratio is the difference between the blue and the red graphs looking at the horizontal axis
- You can see that the signal to noise ratio at the saturation point is at its best - As you move to the dark noise floor the usable signal decrease.....
- Not in the graph - the image dark areas are at the "temporal dark noise point" and the image highlights at the saturation point
- Now you can see its better to expose for highlights (before hitting the clipping) than it is exposing for the shadows (ETTR)
- The only difference at higher ISO's is the DR will be reduce - so more care should be taken not to clip the image....
- Why not clip? We cannot recover clipped info - we can recover shadow detail

Also see this interesting article See his other articles also - this is a great site to bookmark...

Siegfried

 

[Mack]

.....

Also see this interesting article See his other articles also - this is a great site to bookmark...

Siegfried

In that link, one of the comments below by DavidH says the Sony A7 seems to perform better when ETTL (Left.) and not ETTR. There is a Flicker link to the Sony image in it.

Been playing around with a large Styrofoam ball from art supply (JoAnn's Fabrics) for setting up and maximizing the ETTR.

Normally I use an 18% Gray Card but found using it in the shade where contrast is lower does not maximize the ETTR for exposure. This link explains some of it where the low-contrat shot as done by the camera can be improved by 1 2/3 more exposure: https://www.fastrawviewer.com/blog/how-to-use-the-full-dynamic-range-of-your-camera

Often the whites in shadows are not pushed to the right as much as if the shot was taken in bright sunlight where contrast was higher. Also, being white , the Styrofoam ball makes the meter more responsive and also seems to have less issues with glare and positioning the gray card correctly for a proper reading.

I found if I use the Olympus camera's Spot Meter setting on a rough-surfaced white Styrofoam ball and set the exposure so the meter reads +2.7 stops, the ETTR is maximized to the right over what the gray card normally will do - either in sunlight or shade. Also, the playback of the ball's image will show the red over-exposure blinkies on it in the ball's specular highlights telling me I am near the far right of what the camera's histogram shows, along with a bit of playroom left over in the RAW ORF file.

I can further verify the exposure in FastRawViewer (~$19 and great for culling photos!) where the white of the ball shows the FRV histogram's curve settling down around the +3 EV mark on its X-axis. If I engaged the FRV Over-Exposure button in the bottom menu, I can see if the highlights of the ball are still visible, or if they are blown out to a solid white meaning I have clipped the highlights.

So I now use EFTS (Expose For The Styrofoam) instead of ETTR, just apply a +2.7 in the camera's Spotmeter on the upper highlighted area of the ball. If image is nosiy, I'll use Topaz Denoise AI.

 

[PakkyT]

In the example you gave, changing the ISO last was because there was no wider aperture available and (I assume) reducing the shutter speed was not an option either. Otherwise, why ever use a higher ISO?

My point isn't that using a higher ISO is wrong. In both examples, the actual shooting settings ended up identical. But the article writer claims that if you raise the ISO you are not using ETTR even though in his examples he often starts up with a higher ISO, then claims he is using ETTR because he then adjusted the shutter & aperture last. I contend he is probably doing the same as many of the YouTube videos and ends up with the same settings anyway but states they do it wrong.

Take a look at the three photo example in his article. He writes "Next you should be able to point out which of the other two images I exposed to the right." First issue is none of them are exposed to the right. He starts at ISO 2000, then the only thing he actually did was realized the ISO setting was higher than it needed to be, drops it down, then readjusts the exposure on the other two images back to normal exposure by adjusting the shutter on one and aperture on the other. Neither of them are exposed to the right since ETTR is by definition taking what the camera meters as a normal exposure and pushing it to the right (overexposes without blowing out highlights). Second issue is proper ETTR technique would have been to start at base ISO, adjusted shutter and aperture to get the maximum exposure for what he considered the limits of each (considering DoF, camera shake, or subject movement concerns). If the image is still too dark, then sticking with ETTR-esque technique (since once you go above base ISO you are not truly in ETTR land anymore) you then raise the ISO to get proper exposure. After this demonstration he has a paragraph going on about how the YouTube videos he has viewed do it wrong because they use exposure compensation and bump the ISO. As he cites none of them as an example, we don't know what steps those videos actually did to get to the final settings but it could very well be they did exactly the same as his examples, just that they adjusted ISO last rather than first. If both sequence of steps end up with the same camera settings then how do you call one correct and the other wrong?

 

[pdk42]

OK, so let's try some controlled experiments.

Here's the rather messy scene in my home office. It's not a very high contrast scene, so definitely a candidate for ETTR.

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

The above shot, and all the following were shot on a Pen-F in raw with the Oly 17mm f1.8 @ f1.8; then processed in Lightroom using default settings apart from changes to exp comp and shadows as indicated later. No NR applied to any shots. Sharpening was at default value of 40.



1) This shot was taken at ISO 200 and used the camera's suggested exposure (it's the same as the above, I'm just repeating it for clarity):

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

2) Here's the same scene, shot this time with +1.5 stops of exp correction. The white on the binder on the right of the bottom shelf was just showing as overexposed on the blinkies display:

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

3) Here it is again, but pulled back by -1.5 stops in LR:

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

This now looks the same the first shot - so, we've done +1.5 stops of ETTR. Has it made a difference? It's hard to tell at this size (1024 pix), so let's try a bit harder. Let's add +100 to the shadows on them both and go peering into some dark areas at 200% in Lightroom. The ETTR shot is on the right:


4) ETTR comparison:

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

Wow - big difference. It's as if I've got 1.5 stops better noise - since I have!! That's ETTR. Now look at the exposures:

- Left shot (non-ETTR): 1/30 at f1.8.
- Right shot (ETTR): 1/10 at f1.8

So, more light in the right shot - so better filled wells, lower noise. The result matches the theory. Yay!!

Now let's try it at ISO 3200.


4) ISO 3200, camera's suggested exposure:

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

5) ISO 3200, +1.5 exp comp, pulled back -1.5 in LR:

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

Now let's do the same trick - raise shadows by +100 and go peering in the dark areas. Left is normal, right is "ETTR":


6) ETTR Comparison at ISO 3200

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

So - ETTR is working then?

No - it isn't. Look at the exposures in the two shots (shutter speed and aperture). The exposure of the "ETTR" shot is 1/180s at f1.8 at ISO 3200 (+1.5 EC in camera, -1.5 EC in LR) but only 1/500s at f1.8 in the non-ETTR shot . Now let's shoot the same shot at 1/180s, f1.8 at ISO 1250. This is now the same exposure as the "ETTR" ISO 3200 shot, but now shot at EC 0 according to the camera's meter. Let's push the shadows +100 again and go peering into the same dark area:


7) ETTR at ISO 3200 vs regular exposure at ISO 1250

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

So, virtually the same. The ISO 3200 is perhaps very, very marginally better - but the difference is negligible. We've won nothing by dialing in ISO 3200 and playing around with over exposure and pulling back in post processing.

Why? - because the exposure (1/180 at f1.8) is the same in both. The sensor has received the same total light, the wells are filled the same. All we've done on the ISO 3200 shot compared to the ISO 1250 shot is to add amplification in the camera and then reduce the amplification in LR. We've not made the slightest difference to the sensor's performance.

There may be some small differences due to the way that ISO amplification is done (determined by how electronic gain is being done vs ADC vs digital gain) - but whilst it may be helpful to know these details, the magnitude of the change will likely be small, and of course very camera dependent. It may even be the case that the order of noise effects will swap according to what strategy the camera is adopting (i.e. "ETTR" at higher ISOs might be worse). These are all second order effects compared to ETTR (at base ISO) which will make a big difference because we are giving the sensor more light.

 

[relic]

Q.E.D. Even I understand this. Thank you: now I am convinced. I only wonder, for the 2nd case, why, in a real situation, one would use unnecessarily high shutter speed when a lower one would do, so that one can then use a lower ISO. (Of course I often do so because either (1) I simply forgot to check my exposure or (2) I didn't think I had time to adjust things before clicking the shutter button.

However, I beg to differ very strongly with your characterization of the scene in your home office as "messy"!

 

[pdk42]

I beg to differ very strongly with your characterization of the scene in your home office as "messy"!

I admit to tidying it a little for the shot. But don't go looking at the dust on the printer at bottom right!

 

[pdk42]

Incidentally, some people say that using the extended low ISOs on Olympus cameras (64 and 100) is the same as over-exposing at ISO 200 and pulling back the raw file in post (as I've done here). I think in fact that this is not 100% accurate. I haven't been able to find definitive information, but it seems that when you use these extended low ISO values, there is some electronic or processing tweak being done to the sensor such that the sensor response becomes non-linear in the highlights. It tails off, giving a shoulder to the curve. The outcome of this is that highlight headroom is actually a little better at ISO 100 than at ISO 200. In effect, this probably means that you're losing only a half-stop of highlight capacity for the one stop gain in shadow capacity. Net, net - an extra half stop of DR. You can see this in fact in the DR curves on DxO - this one for the Pen F:

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

Half a stop ain't much and I don't usually bother with low ISO - but when I do use it, I'm usually quite surprised with the extra detail I get in the shadows.