Out of Gamut unfixed, compresses all hues

[skid00skid00]

skid00skid00,

I've read about the desaturate monitor colors setting.  I should try that!

RE: calibration, I used my i1 spectro on a GMB scanner calibration chart (to get the LAB values), and thought I had calibration licked.  The test shots of the chart looked fantastic.

But the flowers (pansies) I shot in the late afternoon were nuclear blue, instead of purple...  even though the purples on the test chart were spot-on.

Seems there's something going on WRT as-shot WB in ACR.  My latest tests a few hours ago have me getting better conversions -always- using flash WB in ACR, and then tweaking WB manually.  In many cases, my trouble colors look very good with the calibration tab set to all zeros!

[PeterLange]

Hello Peter,

Shouldn't you at least try to define what you mean by better? (I don't mean scientifically, but in this particular case).

Hello Oscar,

Just let me repeat: Anything “better than RelCol” would be fine (IF possible at all): more details, less channel clipping, no hue shifts – of course without deterioration the overall perceived saturation & vibrancy.  All in all, characteristics which are commonly attributed to a Perceptual conversion in a positive sense; somehow realized for a pRGB > sRGB conversion.

Oscar, - as if you would have to ask...

The preview of the image is a clunky clipped file. If you convert this to sRGB with standard relcol, then you simply get what you see (assuming of course that your monitor behaves somewhere in the sRGB ballpark).

Is there a chance that you could make the ‘Color Clipping’ algorithm available as a Script or Action. I’d be happy to give it a try on a broader basis.

And if you are going to choose some contorted way of converting to sRGB using multiple profiles, might I suggest trying a decent Epson profile as an intermediate space? This will allow you to convert Perceptually to a smaller space, including hue constancy, and allows for the greatest correlation between what you see and what you print, no?

Well, a burned child dreads the fire - referring to direct conversions from ProPhoto RGB to Epson profiles; its Lut can easily behave like an “invisible elephant in the ICC print-rendering room” (wording by Edmund Ronald; see above link).

For the time being, I prefer first to convert from ProPhoto RGB to a couple of smaller matrix spaces, then Softproof to the printer profile, and finally take as the starting point for print (or print-related finishing) whatever looks best. At the risk that this doesn’t comply to ‘fine art’, but in the sense of speed & pragmatics I’m willing to miss some corners of the printer space.


Peter

--

[32BT]

Is there a chance that you could make the ‘Color Clipping’ algorithm available as a Script or Action. I’d be happy to give it a try on a broader basis.

Well, I have been busy transitioning to XCode & intel. I am still in the middle of it. Even so, here is a link to the Mac OS X CMM that is supposed to do this:

Color Clipping CMM

For those of you willing to give it a try, by downloading the file via the link you agree that:
I remain the sole owner of the copyrights,
You are allowed to download, copy, and install this software for non-commercial, testing purposes only,
The software is not fit for any particular purpose, and thus:
I am not responsible in any way, shape, or form for any consequences resulting from the installation and use of this software.

Having said that, the cmm should be able to convert from a normal matrix-based profile or Lab to another normal matrix-based profile.

Installation:
Copy the decompressed file to
/Library/ColorSync/CMMs/

Restart Photoshop and open an RGB image.
Select Convert to Profile
Select "Excalibur" from the engine menu.
Switch intent to see the following conversions:

Relative Colorimetric intent:
As normal relcol but preserves hues in the destination space.

Perceptual intent:
As RelCol, but partly emulates brightness of the source image in destination space.

Saturation intent:
Converts source space primaries to destination space while preserving hue, then uses these primaries for the source space. This fits the source space in destination space and allows for a more accurate "desaturate monitor colors" option.

clicking the "preview" checkbox will switch between normal Photoshop preview and the selected CMM conversion. Also note that Photoshop uses a table based emulation for the preview, so there may be artifacts in preview mode which aren't part of the image after calculation.

[PeterLange]

Well, I have been busy transitioning to XCode & intel. I am still in the middle of it. Even so, here is a link to the Mac OS X CMM that is supposed to do this:

Color Clipping CMM

Oscar,

Many thanks! And yes, I’m impressed.

For the time being, I regret that I can’t follow-up due to my windowz box.

Maybe an experienced Mac user would like to fill in.

Peter

P.S.:  In case that Dennis is still around here. I’ve forwarded his question about ‘how to create a saturation mask’ to the Adobe forum, and got interesting advice so far.
--

[Dennis]

P.S.:  In case that Dennis is still around here. I’ve forwarded his question about ‘how to create a saturation mask’ to the Adobe forum, and got interesting advice so far.

Oh, thanks a lot!

[Ray]

I've steered clear of this thread because I've wasted a lot of ink and paper in the past attempting to examine the differences between Perceptual and Rel Col with out-of-gamut images.

The issue seems to be complicated by irregularities in printer/paper/ink profiles and the fact that Photoshop's out-of-gamut warning might not be reliable or accurate.

I've had a good printer for some time, the Epson 7600, which I use with good profiles (created by Bill Atkinson) and I generally use Epson Premium Lustre paper which seems to have a wide gamut (in conjunction with the ultrachrome inks).

My monitor is of course calibrated, but I find when I'm adjusting an image for printing, it's prudent to check how the rendering intents in proof setup affect the image, before I print. It's so easy to get into the habit that 'perceptual' or 'rel col' is always right or the best. With some images I've found that 'Absolute Col' produces better results, and a few images even "saturation' intent.

Generally, I find that 'perceptual' produces better shadow detail. I tend to do all editing of an image with 'proof colors' ticked. I can't be bothered with numerous layers which can astronomically add to the file size. Keep it simple is my motto   .

[Eric Myrvaagnes]

I can't be bothered with numerous layers which can astronomically add to the file size. Keep it simple is my motto   .
[a href=\"index.php?act=findpost&pid=66635\"][{POST_SNAPBACK}][/a]

Ray,

I'm gratified to learn that I'm not the only one here who doesn't feel that an image isn't "Art" unless it has at least 17 layers!

Eric

[skid00skid00]

Well, tonight I processed some hugely saturated purple, light purple, blue, deep maroon, and yellow pansies (errrr, that's an image...).

In soft-proofing, I found that my i1 printer profile actually produces different results between RC, Perc, Sat, and Abs. Col.  And for this image, Abs. Col. produced the best-looking print on Prem. Glossy Photo Paper (on an Epson R1800).  Yes, some of the subtle shading within a single flower petal was lost, but the overall effect was one of much brighter, more saturated flowers-which is closer to the real thing.

What a surprise for me...Abs. Col!  =:-O

FYI...