SL2, ISO50 and overexposed JPEGs

[SrMi]

Advertisement (gone after registration)

I have encountered an issue with ISO 50 that Leica is not aware of. I wonder if others can reproduce the following bug on SL2 (latest firmware version, camera after full reset).

Test:
- Record DNG + JPEG (or just DNG)
- Set ISO 50, adjust exposure so that you have the blinkies just appear or disappear (ETTR)

Result: 
- DNG is properly exposed, JPEG is overexposed (same data in EXIF files, though).
- The problem is visible in camera review as well. While live-view image shows good histogram/blinkies, image review (JPEG based) shows overexposed histogram/blinkies.

Thank you.

 

[Photoworks]

I just did a test and the 2 images look identical .

The can be many factors, but most likely the computer and the editing software are reading the profiles differently.

I never use 50 ISO because the highlight clip early more than any iso setting.

Most like sensors are better underexposed, this way you preserve the highlights and lift the shadows in post. For this reason you probably would not use the jpg out of camera.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[SrMi]

1 hour ago, Photoworks said:

I just did a test and the 2 images look identical .

The can be many factors, but most likely the computer and the editing software are reading the profiles differently.

I never use 50 ISO because the highlight clip early more than any iso setting.

Most like sensors are better underexposed, this way you preserve the highlights and lift the shadows in post. For this reason you probably would not use the jpg out of camera.

Thank you for checking it. To confirm, did you expose manually, either with exposure compensation or in M mode, and used ETTR?

I use ISO 50 whenever I expose manually, i.e., using histogram/blinkies instead of automatic exposure. Of course, no issue with ISO50 blowing highlights in such an approach.

However, with ISO50, on my SL2, the image review is useless for checking exposure as the embedded JPEG is overexposed. Below are two images from rear LCD: left is live-view (and what LrC sees) and right is image-review (embedded JPEG). The black on the concrete surface is blinkies. You see also that the histogram shows clipped highlights, but they are not clipped in DNG.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

 

[Photoworks]

like i said, with Leica sensor in is not common to ETTR

In any case your meeter indicates you are 1 stop over. and like many cameras that changes on subject.

To me both images are both over, it will be hard to recover colors on over exposed images. Lightroom uses other channels to recover the burned channel of color.

That i know the JPG in camera don't reflect HDR options, Highlight and shadow compensation like other Mirrorless cameras

[BlackDoc]

vor 9 Stunden schrieb SrMi:

Thank you for checking it. To confirm, did you expose manually, either with exposure compensation or in M mode, and used ETTR?

I use ISO 50 whenever I expose manually, i.e., using histogram/blinkies instead of automatic exposure. Of course, no issue with ISO50 blowing highlights in such an approach.

However, with ISO50, on my SL2, the image review is useless for checking exposure as the embedded JPEG is overexposed. Below are two images from rear LCD: left is live-view (and what LrC sees) and right is image-review (embedded JPEG). The black on the concrete surface is blinkies. You see also that the histogram shows clipped highlights, but they are not clipped in DNG.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

 

Please be so kind:

Take a photo with the "same" setting but NOT using "A-mode" but "M mode"...

It might be a difference of the camera (automatic or "a" mode ) by showing the live view and on the other hand after this taking a photo as jpg. Further on: DNG is supposed to be able  to postprocess a pic. JPG pics are somehow "finnished" and compressed and by doing the postprocessing much more problematic.

To my point of view and my workflow it´s more a "nice to know why" problem but not a real existing problem. The dynamic range of taken DNG photos is so big that by taking in consideration "the blinkies" correctly  the pics always (!)  have a good exposure with postprocessing.
And i´ve taken thousands of pics with the SL-2 right at the moment.

\off topic on: additional jpg makes sense if i do want an additional safety or if you  have to work very fast- > e.g.  the jpg pics are to be used right out of the cam with minimal changes. 

Regards,
JOS

[jaapv]

14 hours ago, SrMi said:

I have encountered an issue with ISO 50 that Leica is not aware of. I wonder if others can reproduce the following bug on SL2 (latest firmware version, camera after full reset).

Test:
- Record DNG + JPEG (or just DNG)
- Set ISO 50, adjust exposure so that you have the blinkies just appear or disappear (ETTR)

Result: 
- DNG is properly exposed, JPEG is overexposed (same data in EXIF files, though).
- The problem is visible in camera review as well. While live-view image shows good histogram/blinkies, image review (JPEG based) shows overexposed histogram/blinkies.

Thank you.

 

The embedded jpg always has a more limited DR by about one stop than the DNG - reason why the clipping warnings in your EVF are one stop pessimistic, as they are based on the JPG. Except for the M9 Monochrom, which bases the review clipping warning on the DNG.

[BernardC]

Advertisement (gone after registration)

Question: why "ETTR" at all?

That technique was used in the first days of digital photography, to compensate for the behaviour of contemporary digital sensors (limited dynamic range, noisy shadows). The problem with ETTR, then and now, is that it leaves little room for manipulation after the fact, and it leads to very inconsistent exposures.  That second factor is an issue for photographers who want to produce multiple images of the same scene, it's very hard to make them match.

Do you still find that the benefits outweigh the limitations? What do you find are they main advantages of ETTR? Is it shadow noise, like it was a dozen years ago, or is it something else?

[SrMi]

4 hours ago, jaapv said:

The embedded jpg always has a more limited DR by about one stop than the DNG - reason why the clipping warnings in your EVF are one stop pessimistic, as they are based on the JPG. Except for the M9 Monochrom, which bases the review clipping warning on the DNG.

I am not looking at DR but at exposure. The issue occurs only at ISO 50 and not at ≥ISO100.

The issue is that the live view and image review histograms/blinkies differ a lot at ISO50 Both are based on JPEGs.

[SrMi]

1 hour ago, BernardC said:

Question: why "ETTR" at all?

That technique was used in the first days of digital photography, to compensate for the behaviour of contemporary digital sensors (limited dynamic range, noisy shadows). The problem with ETTR, then and now, is that it leaves little room for manipulation after the fact, and it leads to very inconsistent exposures.  That second factor is an issue for photographers who want to produce multiple images of the same scene, it's very hard to make them match.

Do you still find that the benefits outweigh the limitations? What do you find are they main advantages of ETTR? Is it shadow noise, like it was a dozen years ago, or is it something else?

The issue is that the image review shows overexposed histogram/blinkies when compared to live view, but only at ISO 50. The main reason I use ETTR in this example is because it best demonstrates the overexposure issue (change in surface of blinkies, clipping in histogram).

I am reluctant to comment on ETTR as it may hijack the thread :). Let's start a new thread.

Edited August 27, 2020 by SrMi
Expand on why use ETTR

[Photoworks]

51 minutes ago, SrMi said:

I am not looking at DR but at exposure. The issue occurs only at ISO 50 and not at ≥ISO100.

The issue is that the live view and image review histograms/blinkies differ a lot at ISO50 Both are based on JPEGs.

you don't need to start a new thread. it is already been talk about so much on this forum.

did you try to do a search. 50ISO is problematic on highlights.

Many cameras have different behaviors, I would suggest to lean to find the sweet-spots of your sensor

The base iso is 100 and there is where you find the most benefit.

 

[Dr. G]

33 minutes ago, Photoworks said:

you don't need to start a new thread. it is already been talk about so much on this forum.

did you try to do a search. 50ISO is problematic on highlights.

Many cameras have different behaviors, I would suggest to lean to find the sweet-spots of your sensor

The base iso is 100 and there is where you find the most benefit.

 

Thank you for clearing this up.  I thought the base ISO on the SL2 was 50.  Knowing that it's 100 makes much more sense considering what's happening at ISO 50.

[SrMi]

23 minutes ago, Photoworks said:

you don't need to start a new thread. it is already been talk about so much on this forum.

did you try to do a search. 50ISO is problematic on highlights.

Many cameras have different behaviors, I would suggest to lean to find the sweet-spots of your sensor

The base iso is 100 and there is where you find the most benefit.

 

Yes, there is lots of talk on this forum about ETTR, but apparently, some are still unclear why & when to use it.

Yes, ISO50 is problematic with full automatic exposure (no exposure compensation). Here is a topic for another thread ;-).

But, neither topic is the issue of this thread. The problem is the phenomenon where only at ISO50, there is a large discrepancy between the histogram/clipping of live view vs. image review. The difference seems to be caused by overexposed saved JPEG (either as a separate file or embedded JPEG).

P.S.: Just noticed the same issue to Q2, but not surprised as both cameras have the same ISO50 characteristics (exposure, DR).

[SrMi]

5 minutes ago, Dr. G said:

Thank you for clearing this up.  I thought the base ISO on the SL2 was 50.  Knowing that it's 100 makes much more sense considering what's happening at ISO 50.

Base ISO is 50 not 100 (see PhotonsToPhotos). The problem with ISO 50 is incorrect automatic exposure which may require an underexposure of 2/3 EV.

 

[BernardC]

1 hour ago, SrMi said:

I am reluctant to comment on ETTR as it may hijack the thread

No problem, I was just curious.

I think your issue has to do with how the SL2 processes "extended ISO." My gut feeling is that the extended setting is the same as ISO 100, with a different curve applied, and that the histogram/blinkies don't take that curve into account. I have no inside knowledge, so it's just a theory.

[SrMi]

Just now, BernardC said:

No problem, I was just curious.

I think your issue has to do with how the SL2 processes "extended ISO." My gut feeling is that the extended setting is the same as ISO 100, with a different curve applied, and that the histogram/blinkies don't take that curve into account. I have no inside knowledge, so it's just a theory.

Your ETTR question makes sense, and I would gladly discuss my view & application of it.

ISO50 is native ISO with highest measured DR. It is not extended ISO (there are issues with exposure, though). 

Both live-view and image-review should have the same histogram/clipping as they are both based on the generated JPEG.

[Photoworks]

10 minutes ago, SrMi said:

Yes, there is lots of talk on this forum about ETTR, but apparently, some are still unclear why & when to use it.

Yes, ISO50 is problematic with full automatic exposure (no exposure compensation). Here is a topic for another thread ;-).

But, neither topic is the issue of this thread. The problem is the phenomenon where only at ISO50, there is a large discrepancy between the histogram/clipping of live view vs. image review. The difference seems to be caused by overexposed saved JPEG (either as a separate file or embedded JPEG).

P.S.: Just noticed the same issue to Q2, but not surprised as both cameras have the same ISO50 characteristics (exposure, DR).

 ETTR, but apparently, some are still unclear why & when to use it.

over exposing make sense for canon and sony cameras that has noice in shadows and a higher recovery capability than the Leica sensors.

Leica sensor can recover 2-3 stops in the shadows and it is very clean.

[BernardC]

2 minutes ago, SrMi said:

ISO50 is native ISO with highest measured DR.

That is somewhat unclear. Leica calls it 'extended" in some place, but I haven't seen a full explanation of what that means. Photons to Photos says that they use DxO to measure dynamic range, but DxO doesn't really say how they measure it.

DR is a point of contention because it's very hard to measure. You could, in theory, build a chart with 15+ stops of DR (nobody sells one), but then lens flare would give you the illusion of more DR. You can take several exposures, but then you rely on the assumption that they will be processed the same way, which isn't true of any commercial camera. Noise is also an issue, because it's the lower bound of dynamic range (in the shadows), and most cameras apply noise suppression even in RAW images. Is a deep shadow really part of your "range" if it's been wiped clean by your camera's processor? Surely such an image will ace a DR test, but it won't hold any details in the shadows.

Enough interruption, back to discussing the SL2's behaviour at ISO 50.

[SrMi]

1 minute ago, BernardC said:

That is somewhat unclear. Leica calls it 'extended" in some place, but I haven't seen a full explanation of what that means. Photons to Photos says that they use DxO to measure dynamic range, but DxO doesn't really say how they measure it.

DR is a point of contention because it's very hard to measure. You could, in theory, build a chart with 15+ stops of DR (nobody sells one), but then lens flare would give you the illusion of more DR. You can take several exposures, but then you rely on the assumption that they will be processed the same way, which isn't true of any commercial camera. Noise is also an issue, because it's the lower bound of dynamic range (in the shadows), and most cameras apply noise suppression even in RAW images. Is a deep shadow really part of your "range" if it's been wiped clean by your camera's processor? Surely such an image will ace a DR test, but it won't hold any details in the shadows.

Enough interruption, back to discussing the SL2's behaviour at ISO 50.

PhotonsToPhotos has also data from DXO, but much is computed by Bill himself. I have provided him Q2 data that is used in PhotonsToPhotos graphs. His measurements always show/confirm extended ISOs. Based on past work, his data can be trusted.

[BernardC]

12 minutes ago, SrMi said:

I have provided him Q2 data that is used in PhotonsToPhotos graphs.

What did he ask you to provide? Was it a number of random DNG files taken at various ISO settings, or were there specific things that he wanted in the images? I trust that he did not ask you to separate your Q2's lens from its sensor.

Again, sorry if I am intruding, I am just curious. This topic has been discussed often in cinematography forums, mostly because computed numbers are often at odds with real-world results. For instance, a camera with more DR in tests might be more likely to blow-out, compared to a camera with less DR, even when using the same lens. Post-processing steps are very different for cine work (Resolve, After Effects, etc.), so what's happens in one field may not apply in the other.

[Advertisement]

Hallo,

Look here: Leica L-Mount

[SrMi]

27 minutes ago, BernardC said:

What did he ask you to provide? Was it a number of random DNG files taken at various ISO settings, or were there specific things that he wanted in the images? I trust that he did not ask you to separate your Q2's lens from its sensor.

Again, sorry if I am intruding, I am just curious. This topic has been discussed often in cinematography forums, mostly because computed numbers are often at odds with real-world results. For instance, a camera with more DR in tests might be more likely to blow-out, compared to a camera with less DR, even when using the same lens. Post-processing steps are very different for cine work (Resolve, After Effects, etc.), so what's happens in one field may not apply in the other.

Cannot find the original email so I am working from memory. Bill provided me with test patterns to display on the monitor and with instructions how to take a series of exposures.