Jump to content

Highlight Recovery


SrMi

Recommended Posts

Advertisement (gone after registration)

When you say "good highlight recovery", it often means that your raw converter applied positive brightness correction, and moving "exposure" slider in the raw converter to the left, you force it to display what was initially hidden. (Iliah Borg)

"Highlight recovery" isn't a sensor characteristic. It's something you hope to squeeze out of the image when you have missed your exposure. If you expose properly (eg. Exposure To The Right) your highlights are properly exposed and "highlight recover" is moot. Then, how much you can "recover" from the shadows is ultimately what matters. (Bill Claff)

Adobe considers highlights recovery when you clip at least one channel (link). I am looking at Adobe's definition.

I have observed the highlight recovery of M10-P, M10-R, and M11 using the same lens and subject (in live-view with multi-field metering). RawDigger was used to determine the clipping, and Lightroom was used to assess highlight recovery. M10-R and M11 have similar highlight recovery. M10-P degrades quicker from fully recoverable with color loss to useless.

ISO 200

M10-P
no clipping: 1/750
clipping starts at: 1/500
fully recoverable until: 1/350 (color changes)
Useless: 1/250

M10-R
no clipping: 1/1000
clipping starts at: 1/750
fully recoverable until: 1/350 (color changes)
Useless: 1/180

M11
no clipping: 1/1000
clipping starts at: 1/750
fully recoverable until: 1/350 (color changes)
Useless: 1/180

Edited by SrMi
  • Like 6
  • Thanks 4
Link to post
Share on other sites

This is what I wrote in my report:

The sensor has very impressive capabilities of highlight recovery. I photographed a greyscale and overexposed up to 5 stops at ISO 64, 200, 800, 3.200, 6.400 with 18, 36, 60 MP. The result: At every ISO and every resolution you can recover 3 stops easily, but not 4 stops.

So I see a little more capability of highlight recovery. But I didn´t look after the color chanels.

  • Thanks 2
Link to post
Share on other sites

26 minutes ago, elmars said:

This is what I wrote in my report:

The sensor has very impressive capabilities of highlight recovery. I photographed a greyscale and overexposed up to 5 stops at ISO 64, 200, 800, 3.200, 6.400 with 18, 36, 60 MP. The result: At every ISO and every resolution you can recover 3 stops easily, but not 4 stops.

So I see a little more capability of highlight recovery. But I didn´t look after the color chanels.

What is the point above which you can recover 3 stops?

Link to post
Share on other sites

Just in case someone thinks something weird happens at 1/750 or 1/500th - all the time, anywhere....

I'd just point out that ISO and shutter speed are pretty meaningless without knowing the scene brightness and aperture.

Indoors, outdoors overcast, outdoors sunny, nightime? f/1.4, f/16?

  • Like 2
Link to post
Share on other sites

Adan is correct. The highlights can start clipping at any shutter speed. The shutter speed numbers in my test are only significant as you observe the number of stops before the highlights are no longer recoverable. 
Also, the numbers of stops depend on the scene and light. The number of stops before the highlights become non-recoverable varies. What seems to remain constant is the relation between the cameras. M10-P's images seem to give up a bit earlier. M10-R's and M11's images seem to have a similar range of highlight recovery. 
The highlight recovery range is from the point when the first channel clips to the point where I can still reconstruct data in the post.

Once you have overexposed a channel, your image is 'compromised'. Note that Lightroom and Photoshop do not show clipping properly, but they are pretty good in reconstructing some missing data.

Link to post
Share on other sites

6 minutes ago, elmars said:

Sorry, my English is too bad to understand what You mean.

Or my English is too bad to explain clearly. Sorry about that.

When you say 3 stops of highlight recovery, how do you measure?

I used the point when clipping starts to the point when I lost details in the images.

If no channel is clipped than all highlights data is recoverable, regardless of the camera used. 

Link to post
Share on other sites

Thank You!

I took pictures of a greyscale and overexposed them up to five stops over the right exposure (or what I regarded a right exposure). In LR I tries for each photo if I could get a proper greyscale. This was possible up to an overexposure from 3 f-stops. At 4 f-stops the white side of the greyscale could not be restored completely.

  • Like 1
Link to post
Share on other sites

@SrMi Your test of the M10-P agrees exactly with my test of the M10, as you would probably expect. You have maybe 1/2 stop of recovery above the clipping point. Additionally the highlight blinkies come on perhaps 1/2 stop before clipping starts. All of this assuming a white object. Introduce different tones and things will change.

 

As a comparison with another common camera. In the same test I did with the M10 I tried the Sony A7iii. I found highlight blinkies to start almost 1 stop before the clipping point and was able to recover 1 stop above the clipping point. So slightly better than the M10 but not as big a difference as I expected.

 

All in all its best never to clip if possible, except tiny specular highlights.

  • Like 1
Link to post
Share on other sites

On 1/27/2022 at 3:07 AM, Adam Bonn said:

did you use that trick in LR where you set process 2010 (maybe v2 these days) zero out all the sliders, then go back to the current process to take away the default ACR tone curve?

I used Rawdigger to determine when the clipping starts, and default LrC  settings to recover details. I would assume that the reconstruction of missing data is independent of tone curves applied. I should have tested it with linear profile as well.

  • Thanks 1
Link to post
Share on other sites

  • 2 months later...

SrMi - Thank you for your test.

I'm wondering why the clipping starts at 1/500 for M10-P and at 1/750 for M10-r and M11? What is this indicating exactly? does that mean the clipping starts half stop faster in M10-R than M10?

Also did you compare the shadows of these cameras too? Some people think Leica shifted the DR towards highlight for m10-R (not sure about m11) so the highlights are more recoverable but shadows are more recoverable (or less noisy) in M10-P (or M10). What do you think based on your comparisons?

If that's the case maybe a half stop under exposing with m10 (to keep the highlights from clipping) and then adding a half stop to the shadows in post process should give a pretty much close result to M10-R picture (in theory) Any thought? 

Link to post
Share on other sites

17 hours ago, SOHODE said:

If that's the case maybe a half stop under exposing with m10 (to keep the highlights from clipping)

It's not as simple as 'just deduct 1/2 a stop' - it would be about determining exactly where the highlights clip for that particular scene.

I might be getting 5 here from adding 2 + 2... (in which case apologies) but if you're trying to decide between the M10 and M10R, I'd think more about things like do I want 40mp and am I ok with the perception of increased noise from the larger files far more than I would about getting a potential 1/3 (or whatever) of a stop more HL recovery from 10R, and speaking personally I'd find some DNG samples online to see which native colour palette I liked more

At the end of the day, no matter the camera - if you clip data it's gone... whether the camera has 8 stops, 12 or 14.. whether you clip data will depend on your and your camera's ability to meter the scene

  • Like 3
Link to post
Share on other sites

On 4/21/2022 at 4:43 PM, SOHODE said:

SrMi - Thank you for your test.

I'm wondering why the clipping starts at 1/500 for M10-P and at 1/750 for M10-r and M11? What is this indicating exactly? does that mean the clipping starts half stop faster in M10-R than M10?

Also did you compare the shadows of these cameras too? Some people think Leica shifted the DR towards highlight for m10-R (not sure about m11) so the highlights are more recoverable but shadows are more recoverable (or less noisy) in M10-P (or M10). What do you think based on your comparisons?

If that's the case maybe a half stop under exposing with m10 (to keep the highlights from clipping) and then adding a half stop to the shadows in post process should give a pretty much close result to M10-R picture (in theory) Any thought? 

I assume that different metering is the cause for the difference in shutter speeds where clipping starts.
My observations are that shadows are comparable except with heavy lifting of shadows, where M10-R does a bit better.
I do not support a permanent EC on any camera. If the scene lighting is challenging, I will check the live histogram. The best is to get a lot of experience in metering with a camera. That way, you'll be able to be prepared to adjust exposure even without using live view.

  • Like 2
Link to post
Share on other sites

On 4/22/2022 at 1:43 AM, SOHODE said:

SrMi - Thank you for your test.

I'm wondering why the clipping starts at 1/500 for M10-P and at 1/750 for M10-r and M11? What is this indicating exactly? does that mean the clipping starts half stop faster in M10-R than M10?

Also did you compare the shadows of these cameras too? Some people think Leica shifted the DR towards highlight for m10-R (not sure about m11) so the highlights are more recoverable but shadows are more recoverable (or less noisy) in M10-P (or M10). What do you think based on your comparisons?

If that's the case maybe a half stop under exposing with m10 (to keep the highlights from clipping) and then adding a half stop to the shadows in post process should give a pretty much close result to M10-R picture (in theory) Any thought? 

I am wondering what you could mean by “shifting the DR”. DR is a measurement between the noise floor and 100% fill of the pixels. There is no way to shift those values.  Highlight recovery is something different. It only means the ability to extrapolate data from the non-clipped colour channels and has nothing to do with DR. If you photograph a one primary colour subject highlight recovery will be zero. 

  • Like 1
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...