Pages: [1]   Go Down

Author Topic: Why does the PS Camera Raw filter differ from ACR itself?  (Read 4406 times)

Redcrown

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 312
Why does the PS Camera Raw filter differ from ACR itself?
« on: February 24, 2016, 01:13:04 PM »

Convert a raw mage with no adjustments. Then do it again with tone adjustments, maybe just a +0.5 Exposure. Do it all in ProPhoto.

Stack the two conversions as layers in Photoshop. Then run the Camera Raw filter on the "no adjustments" layer, and apply the same ACR adjustment. Compare the two layers. Why are they different? Same adjustments, same colorspace, but one directly from raw and one not.

I think I know the answer, but looking for confirmation. Is it because the colorspaces are not really the same? Prophoto in Photoshop is not the same as the infamous "Melissa" colorspace (and its sRGB tone curve) inside ACR. And one image making two passes through Melissa comes out different.

Raw to Melissa to Prophoto, back to Melissa back to Prophoto Not= Raw to Melissa to Prophoto?
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11996
    • http://digitaldog.net/
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #1 on: February 24, 2016, 01:49:38 PM »

Convert a raw mage with no adjustments. Then do it again with tone adjustments, maybe just a +0.5 Exposure. Do it all in ProPhoto.
Stack the two conversions as layers in Photoshop. Then run the Camera Raw filter on the "no adjustments" layer, and apply the same ACR adjustment. Compare the two layers. Why are they different?
They are absolutely identical on this end (I ran a Difference and the results were black). What version of ACR? Earlier versions of ACR when run in Filter mode currently always reports RGB values using ProPhoto RGB, regardless of the source document space, so could you have that older version and perhaps not be using ProPhoto on both?
Logged
Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1996
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #2 on: February 24, 2016, 02:08:48 PM »

infamous "Melissa" colorspace (and its sRGB tone curve) inside ACR.

there is no "Melissa RGB" in ACR and never was AFAIK

LR : uses MelissaRGB (ProPhoto primaries w/ sRGB TRC/curve) for its color readouts and histogram, unless you switch to softproofing (at least as of LR v4.x)

ACR : uses the actual selected output color space for its color readouts (at least since ACR v7.x) normalized to 8 bits precision.

so that abomination solely belongs to LR
Logged

Redcrown

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 312
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #3 on: February 25, 2016, 02:44:09 AM »

I made a short video to demo what I'm talking about:
https://www.youtube.com/watch?v=YIX2rDkF5F4

Any more ideas? Is my testing procedure flawed?

Andrew, when you saw a black "difference" screen, did you check to verify there was truly no difference? I assume you tested on a Mac?
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11996
    • http://digitaldog.net/
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #4 on: February 25, 2016, 10:17:48 AM »

Andrew, when you saw a black "difference" screen, did you check to verify there was truly no difference? I assume you tested on a Mac?
Saw black, saw no visible difference toggling layers on or off, I'm on a Mac. I used a  plus 5 and plus 10 exposure for my tests.
Maybe if you upload the raw you tested, we can try that. Also, I'm not (never) use Bridge to access ACR FWIW.
Logged
Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/

Redcrown

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 312
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #5 on: February 25, 2016, 11:43:49 AM »

Thanks, Andrew. Raw file from the video is here:
https://www.dropbox.com/s/iyb0hgidlu05whd/italy240v000.dng?dl=0

Probably takes more than a plus 0.10 exposure to visibly see a difference, but it still should be measurable with stddev or Threshold analysis.
With only a +0.10 exposure, this image shows me a stddev of 0.55 and a difference in Threshold through level 3.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11996
    • http://digitaldog.net/
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #6 on: February 25, 2016, 12:03:21 PM »

Probably takes more than a plus 0.10 exposure to visibly see a difference, but it still should be measurable with stddev or Threshold analysis.
With only a +0.10 exposure, this image shows me a stddev of 0.55 and a difference in Threshold through level 3.
I tried the test based on your first post: Then do it again with tone adjustments, maybe just a +0.5 Exposure. Do it all in ProPhoto.
Logged
Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11996
    • http://digitaldog.net/
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #7 on: February 25, 2016, 12:09:22 PM »

With your raw, I see a very, tiny difference only in the sky with a minus 30 exposure used. Probably a difference in the gamma encoding; raw linear, ProPhoto run as a filter 1.8.
Logged
Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/

Pictus

  • Full Member
  • ***
  • Offline Offline
  • Posts: 210
    • Retouching
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #8 on: February 25, 2016, 07:15:11 PM »

Here it is Windows 10 + ACR 9.4.0.548 + Photoshop CC 2015.1.2
With your RAW I see exactly the same problem even with a custom gamma 1 color space.
The only thing I know is that ACR used as a filter(SHIFT+CTRL+A) is not 16 bits, but 15+1 bits.
 :o
Logged

FranciscoDisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1084
    • Frank Disilvestro
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #9 on: February 26, 2016, 04:32:03 AM »

I don't expect them to be the same:

- When you open a DNG/RAW in ACR there are a few things that happen under the hood which are not necessarily linear, such as automatic invocation of highlight recovery, especially with process 2012. This difference is more evident if you have an image with overexposed or even clipped areas.

- When you use the PS Camera Raw filter, you just open in ACR the rendered Photoshop image (not the original RAW), which is bounded between 0-255. Another way to see it is that the RAW file might have highlights and shadows beyond the tentative rendering, which can be used for highlight/shadow recovery and the tiff/rendered image don't.

Try this: Save your image in Photoshop as a 16 bits Tiff, Prophoto RGB and repeat the experiment using this file
I mean: open the tiff in ACR with no adjustments to PS, then open again the tiff in ACR, adjust exposure and then Photoshop. Now back to the previous file and use the Camera Raw filter and adjust exposure: You will find that both files are identical, because the input to ACR was essentially the same, a rendered image not a RAW.

If you want the capability of doing adjustments after you have opened the image in PS use Smart objects (of course, ACR will not show any adjustment made in PS)

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2124
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #10 on: February 26, 2016, 11:50:32 AM »

I've always understood once a Raw image is brought into Photoshop it's in a gamma encoded environment which basically means you're working on a high quality jpeg pixel image. It's the reason why when I create a cross process look saved preset on a jpeg which involves a lot of HSL, Split Tone and WB adjustments, it doesn't give the same look applying the preset to a Raw image and vice versa.

Ask yourself what does ACR do differently editing a jpeg vs a Raw image. Both present a normal preview but from where and what transform construct is it defined?
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1996
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #11 on: February 26, 2016, 11:59:54 AM »

Ask yourself what does ACR do differently editing a jpeg vs a Raw image.

in both cases the data will end up as RGB coordinates in PP/Gamma 1 working space... the different part is naturally before that... you can further output the data as PP/Gamma 1 if you want...
Logged

Redcrown

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 312
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #12 on: February 26, 2016, 12:22:48 PM »

Thanks to all,

There have been a number of forum discussions (LuLa and others) about a part of the logic inside ACR that has been called "under the hood"or "secret sauce", and even "Hijinks" in a recent POTN thread. Logic that is hidden and outside of user control.

This exercise seems to demonstrate that logic. I think FranciscoDisilvestro's explanation is the most likely. It's the under-the-hood aspects of ACR that lead to differences in tonal adjustments applied during raw conversions vs. identical adjustments applied via the Camera Raw filter in Photoshop. Apparently the under-the-hood stuff does not happen when the Camera Raw filter runs on a Photoshop layer.

I got into this testing due to a thread in a different forum where the claim was made that the Exposure adjustment in ACR/LR was "image adaptive" and "dependent on image content", with those quotes attributed to Martin Evening and Eric Chan. I ran a bunch of tests that did not support that claim. I found that the Exposure slider would move identical pixel values the exact same amount regardless of where they fell in an image and regardless of the overall image content.

But I ran all my tests using the Camera Raw filter on Photoshop layers, because it was easier to construct tests that way. But if you make the effort to find two identical pixels in significantly different raw images, and run identical Exposure adjustments on those images, the two identical pixels will not have the same adjusted value. Apparently because the secret sauce is applied to raw.
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2124
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #13 on: February 26, 2016, 12:33:14 PM »

in both cases the data will end up as RGB coordinates in PP/Gamma 1 working space... the different part is naturally before that... you can further output the data as PP/Gamma 1 if you want...

Not quite sure what you're getting at and how it relates to this topic. The slider and other edit tool behavior going by how it acts on the preview editing a jpeg in 16bit ACR gamma 1 is like night and day doing it on a Raw version of the image. I've tried it.

So when invoking the ACR filter within Photoshop I'ld suspect the same thing is going on with the Raw image that has now become a pixel image much like working on a jpeg in ACR.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11996
    • http://digitaldog.net/
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #14 on: February 26, 2016, 01:07:22 PM »

in both cases the data will end up as RGB coordinates in PP/Gamma 1 working space... the different part is naturally before that... you can further output the data as PP/Gamma 1 if you want...
Exactly! Raw or rendered image in any gamma encoding, if you run that data through ACR, it's processing is still ProPhoto primaries, 1.0 'gamma'. But the data before processing isn't necessarily the same encoding.
Logged
Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1996
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #15 on: February 26, 2016, 02:21:19 PM »

Not quite sure what you're getting at and how it relates to this topic. The slider and other edit tool behavior going by how it acts on the preview editing a jpeg in 16bit ACR gamma 1 is like night and day doing it on a Raw version of the image. I've tried it.

I am just stating that before you can move a slider in ACR's UI all the data will get to PP/G1 working space, that's it...
Logged

FranciscoDisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1084
    • Frank Disilvestro
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #16 on: February 26, 2016, 03:55:57 PM »

I am just stating that before you can move a slider in ACR's UI all the data will get to PP/G1 working space, that's it...

But it is not the same data

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11996
    • http://digitaldog.net/
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #17 on: February 26, 2016, 03:58:28 PM »

But it is not the same data
Right. And so is AlterEgo and hence, that's why there are slight differences seen by the OP I believe. 
To clarify what I believe is the three of us agreeing on: the data is different (linear encoded raw vs. gamma corrected non-raw) and, the processing is always linear encoded.
Logged
Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1996
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #18 on: February 26, 2016, 04:21:31 PM »

But it is not the same data
no argument with that... I am just trying to repeat the obvious that ACR/LR will go the common workflow inside ASAP for all different inputs (may be except things like supported Sigma cameras where it was noted that code-wise it is quite different in more places).
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1643
    • http://www.guillermoluijk.com
Re: Why does the PS Camera Raw filter differ from ACR itself?
« Reply #19 on: February 26, 2016, 04:43:23 PM »

When exposure is changed, specially when it is reduced, some decisions are to be made about how to deal with saturated/near saturared RGB values since no matter how much you reduce exposure, blown highlights need to remain blown (we don't want gray colour there).

This is best seen looking at how exposure reduction works in terms of a curve, look near saturation (the strange shadow non-linearity, probably gamma related, is another story to study):



The two ways you are producing the final image are clearly not taking the same way, but the deviation is not really important. You cannot even say one is more correct than the other, since both of them had to define a highlight strategy because it was needed. You can simply subjectively prefer one to the other.

As a perfectionist you'd have liked to see exactly the same result into an Adobe software?. Fine, but tools are developed subject to different restrictions and I guess it is not easy to share/match every single step of the processing pipe, even inside Adobe.

Regards

Enviado desde mi GT-I9195 mediante Tapatalk
« Last Edit: February 26, 2016, 05:01:31 PM by Guillermo Luijk »
Logged
Pages: [1]   Go Up