Pages: 1 [2]   Go Down

Author Topic: Converting ProPhoto RGB to Adobe RGB - which Rendering Intent to use?  (Read 7419 times)

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Converting ProPhoto RGB to Adobe RGB - which Rendering Intent to use?
« Reply #20 on: July 30, 2016, 02:34:16 pm »

It's not odd - that's the way it has to work. Introducing a defined intermediate gamut doesn't solve the problem that the source and destination gamuts are not simultaneously known to the gamut mapping. And without both, you can't know whether it's actually an expansion or compression. So you end up expanding the source gamut to the PRMG, and then compressing the PRMG to the destination gamut, resulting in a "saturation" type mapping. You mightn't notice if the source gamut mostly exceeds the destination (i.e. the case for a lot of the gamut for a display space to ordinary printing space), although there is excessive movement of colors that might result in inaccuracy, compared to the mapping situation where there is no PRMG involved.

Mostly, this is why I stay in RelCol and adjust using soft proofing or AbsCol for repro. The ICC V4 sRGB Perceptual reference profile hasn't been widely adopted to say the least. It was described as a Beta with the Beta to be removed in 6 months. That was 7 years ago and it's still a "Beta."

On further investigation the reference sRGB expands saturation in the cyans, yellows, and magentas but compresses saturation along hues that are closer to the primaries, red, green and blue consistent with subtractive  ink v additive color gamut characteristics.

Quote
If I want "pop", I'll choose a saturation intent, not a saturation intent that is masquerading as a perceptual one.

Yep, I much prefer controlling colors myself rather than have this 'enhanced" perceptual PRMG mapping.


Logged

smthopr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 612
    • Bruce Alan Greene Cinematography
Re: Converting ProPhoto RGB to Adobe RGB - which Rendering Intent to use?
« Reply #21 on: July 30, 2016, 02:50:53 pm »

I think this issue might be over thinking things a bit.  When we color correct images on our computers, we are, by definition, working in the color space of the display.  We are seeing the photoshop transform from PP (if that's what the image is stored as) into our display space and making adjustments.  I would assume, we would want "convert to profile" to use the same conversion math as the behind the scenes conversion that we see as we are editing the image.

I suppose the challenge is converting from the PP original data to a printer space that doesn't match the display space.  We are hoping that somehow it will match the display, but yet be somewhat better, taking advantage of colors that can be printed but not displayed. This might happen, but can't be truly previewed on the display.  Test prints will be necessary to see the result.

But as to converting from PP to aRGB?  We should want the conversion to match the behind the scenes conversion that we use to edit the image.  I can't see much of any advantage here to having a choice of rendering intents.

When editing images for sRGB computer displays, I think it would be best to set the working space to sRGB and adjust the image by eye (PP original file) if necessary.  Editing in a very wide space like PP and then doing an auto conversion at the end might lead to some surprises that can't be undone as the data has been deleted in the conversion.

My 2 cents :)
Logged
Bruce Alan Greene
www.brucealangreene.com

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Converting ProPhoto RGB to Adobe RGB - which Rendering Intent to use?
« Reply #22 on: July 30, 2016, 06:25:41 pm »

I suppose the challenge is converting from the PP original data to a printer space that doesn't match the display space.  We are hoping that somehow it will match the display, but yet be somewhat better, taking advantage of colors that can be printed but not displayed. This might happen, but can't be truly previewed on the display.  Test prints will be necessary to see the result.

Fully agree, except for the last sentence. Obviously, for a fully predictable preview/proofing the display gamut should be identical (or larger but under utilized compared) to the output medium's gamut. But because the output media's gamut differs for each medium, and partially exceeds that of current displays, it would require a display gamut that exceeds the best possible output medium.

The (compromise) solution, is to use a more numerical approach, i.e. a gradual clipping indicator/mask that represents the Out-of-Gamut data. I achieve that by (with a Photoshop action) creating 2 versions of a file, and mask according to their Layer differences.
If we take the original image, e..g in ProPhoto RGB, and a copy of the same image but converted to the output profile, we can calculate the Absolute differences, and use that as a mask for desaturation or hue change. That mask is then used to actually e.g. desaturate the original in PPRGB. This works quite well, because usually 99% of the image is not Out-of-Gamut (OOG), so why reduce the entire gamut (which is what Perceptual Rendering Intent does) to accommodate just a relatively small part of the image (unless to avoid gamut banding).

Quote
When editing images for sRGB computer displays, I think it would be best to set the working space to sRGB and adjust the image by eye (PP original file) if necessary.  Editing in a very wide space like PP and then doing an auto conversion at the end might lead to some surprises that can't be undone as the data has been deleted in the conversion.

Correct, but then we will be still guessing in the 'close to aRGB' display case, or future technology displays.

We need better tools for that, and I am (unfortunately, mildly) amazed that we're not provided with those. Not large enough a market would be the more plausible explanation, but by no means an excuse.

Cheers,
Bart

P.S. I think that the subscription earnings model of e.g. Adobe Photoshop/Lightroom, takes away that need to stay ccompetitive and innovative (which is one of the reasons that I think it's bad for quality/innovation).
« Last Edit: July 30, 2016, 06:46:12 pm by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==
Pages: 1 [2]   Go Up