Pages: [1]   Go Down

Author Topic: ACR Highlight Recovery Mystery?  (Read 25177 times)

Redcrown

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 507
ACR Highlight Recovery Mystery?
« on: May 11, 2015, 01:06:52 pm »

I just discovered something about highlight recovery I never noticed before. The strength of highlight recovery depends on the how much of an image is "highlights" and how much is mid-tones and shadows.

I discovered this while processing a set of landscape shots of a mountain scene with a dramatic sky full of white clouds. I composed one shot as 25% sky and 75% mountain and one shot as 75% sky and 25% mountain. Identical (manual) exposures, and identical highlight recovery applied to both.

On the image with 75% clouds and 25% darker foreground the highlight recovery was not very strong. On the image with 25% clouds and 75% darker foreground, the highlight recovery on the clouds was much stronger.

For example, with highlight recovery at -100, a sample point on the image with 75% clouds moved from 230 to 224, only 6 points. On the image with 25% clouds the same sample point moved 14 points, from 230 to 216.

So I'm curious why this is?
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: ACR Highlight Recovery Mystery?
« Reply #1 on: May 11, 2015, 01:29:38 pm »

I just discovered something about highlight recovery I never noticed before. The strength of highlight recovery depends on the how much of an image is "highlights" and how much is mid-tones and shadows.

I discovered this while processing a set of landscape shots of a mountain scene with a dramatic sky full of white clouds. I composed one shot as 25% sky and 75% mountain and one shot as 75% sky and 25% mountain. Identical (manual) exposures, and identical highlight recovery applied to both.

On the image with 75% clouds and 25% darker foreground the highlight recovery was not very strong. On the image with 25% clouds and 75% darker foreground, the highlight recovery on the clouds was much stronger.

For example, with highlight recovery at -100, a sample point on the image with 75% clouds moved from 230 to 224, only 6 points. On the image with 25% clouds the same sample point moved 14 points, from 230 to 216.

So I'm curious why this is?

start with exposure parameters and the actual raw data in that area ( www.rawdigger.com ) = you might as well have different exposure and different clipping (or whatever) ... but if you are using process 2012 then Adobe implemented a lot of non linear, adaptive things related to exposure processing vs process 2010 ...

PS: if M mode for both with identical exposure then compare just the raw data and blame process 2012
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: ACR Highlight Recovery Mystery?
« Reply #2 on: May 11, 2015, 06:29:56 pm »

I'ld suggest you carefully examine in the same manner you've described the differences of White vs Highlight vs Exposure.

Shadow Recovery really is a huge improvement over the old Fill slider when it comes to retaining shadow clarity while providing a much more gradual tapering off into midrange that would accentuate Clarity induced dark halos around cloud edges using PV2003-2010.

IOW set your Exposure to almost clip the whitest white in the cloud but slightly back off and try to adjust either Highlight or White Recovery to add back in cloud definition.

It helps to have ProPhotoRGB set as output profile to show you the effects on highlight clipping in the histogram. Without a robust output reference what you see on the preview may not be reflected in the histogram on what is considered as clipping.
« Last Edit: May 11, 2015, 06:31:42 pm by Tim Lookingbill »
Logged

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6229
    • http:www.schewephoto.com
Re: ACR Highlight Recovery Mystery?
« Reply #3 on: May 11, 2015, 10:15:15 pm »

So I'm curious why this is?

It's the result of image contextually based adjustments...a bit more info in this article "Magic or Local Laplacian Filters?"

And if you want to dive a bit deeper, check out "Local Laplacian Filters: Edge-aware Image Processing with a Laplacian Pyramid"
Logged

Redcrown

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 507
Re: ACR Highlight Recovery Mystery?
« Reply #4 on: May 11, 2015, 11:58:52 pm »

Thanks folks. Now I know I can blame it on Process 2012 and Laplacian Pyramids. Thankfully it's not Lipshitz constants or Wronskian determinants. Those two are what ended my math major in college and sent me into the new field of computer science (circa 1968). But they have served me well ever since to impress people at parties.

But it seems like something that needs more blame than praise. I set ACR adjustments for one image in a series and copy those settings to the rest of the series, expecting equal results. With highlight recovery I don't get what I expect. Probably shadow recovery too, haven't tested that yet.

Imagine if the same thing happened with white balance. Set one of a similar series, clone to the others, get different results because some have more green than others.
Logged

Jimbo57

  • Full Member
  • ***
  • Offline Offline
  • Posts: 180
Re: ACR Highlight Recovery Mystery?
« Reply #5 on: May 12, 2015, 06:17:40 am »


But it seems like something that needs more blame than praise. I set ACR adjustments for one image in a series and copy those settings to the rest of the series, expecting equal results. With highlight recovery I don't get what I expect. Probably shadow recovery too, haven't tested that yet.



It's called software democracy - produce software to suit the majority even if it disadvantages some minorities.

I have no doubt that the "image adaptive" algorithms used in Lightroom (and ACR) are of huge benefits to most photographers (myself included) but will throw up problems such as you suggest for particular users.
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: ACR Highlight Recovery Mystery?
« Reply #6 on: May 12, 2015, 10:00:42 am »

Imagine if the same thing happened with white balance. Set one of a similar series, clone to the others, get different results because some have more green than others.
but it is the case with WB when illumination is different - here with non linear, adaptive, content aware, mumbo-jumbo processing of various exposure related parameters (that you set fixed in UI) you have the same situation as trying to apply the same WB to shots with different illumination, no ?
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: ACR Highlight Recovery Mystery?
« Reply #7 on: May 12, 2015, 01:03:36 pm »

I set ACR adjustments for one image in a series and copy those settings to the rest of the series, expecting equal results. With highlight recovery I don't get what I expect. Probably shadow recovery too, haven't tested that yet.

Imagine if the same thing happened with white balance. Set one of a similar series, clone to the others, get different results because some have more green than others.

I hadn't considered that angle. Good catch, Redcrown. Wedding photographers who crank out thousands of images would have a problem maintaining consistency with that issue.

This explains the inconsistencies I've been getting in LR4's PV2012.

So as I understand this the Adobe engineers figured out a way to override the eye's adaptive nature viewing/editing images that have an overall disproportionate amount of lights vs darks in order to maintain overall perception of clarity. This explains why I don't get the overall local/global contrast induced graupel/cottage cheese texture in LR4's PV2012 over ACR6.7's PV2010 when attempting to bring out definition and clarity consistently throughout the tonal scale of the image.
« Last Edit: May 12, 2015, 01:05:17 pm by Tim Lookingbill »
Logged

Plateau Light

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 52
    • Robert Park Fine Art Photographer
Re: ACR Highlight Recovery Mystery?
« Reply #8 on: May 17, 2015, 01:46:13 pm »

It's the result of image contextually based adjustments...a bit more info in this article "Magic or Local Laplacian Filters?"

And if you want to dive a bit deeper, check out "Local Laplacian Filters: Edge-aware Image Processing with a Laplacian Pyramid"

So does this non linear "Smart" adjustment happening behind the scenes make ETTR increasingly difficult as the software is doing hidden adjustments that are not necessarily what you want.
In essence it might be doing a lot of recovery and crushing highlights when you really want to have less exposure with a more linear rendering?

I do a lot of art repro and have found that C1 Pro with the linear raw gives proper tonality with ETTR  whereas CR always looks interpretative and inconsistent.

Maybe ETTR and ACR 9 just don't mix that well.

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: ACR Highlight Recovery Mystery?
« Reply #9 on: May 17, 2015, 02:13:45 pm »

Maybe ETTR and ACR 9 just don't mix that well.
Works well for me. But I take exposure + development into consideration.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".
Pages: [1]   Go Up