CHROMiX

'Plot as vectors' and the perceptual rendering intent

Hello

Applying ‘plot as vectors’ to both a Lab and a ProPhoto tiff and choosing different cmyk profiles as ‘Destination’ I can see differences among all the rendering intents but Perceptual. When I select it, nothing changes: is there a reason for this behaviour? If this is the case, shouldn’t it be simply disabled?

I’d also like to know whether I have it right that when building these plots ColorThink is using the BtoA tables of the cmyk profiles.

Thank you very much.

Giordano

My configuration:
Colorthink Pro 3.0.1b5 for macintosh
Mac OSX.4.6
G5 Quad with GeForce 7800GT
Plenty of ram

Frankly, I wonder which threshold my questions have to pass to arouse Chromix’s interest. Let’s put it in a different way.

Is anybody else having problems with the Perceptual intent when plotting images as vectors?
Alternatively, has anybody been able to see the least difference between Perceptual and the other rendering intents?

Thank you very much.

Giordano

I have the same problem. I’ve become skeptical of the accuracy of all the intents as a result. This problem is definitely annoying. If someone could fix it or tell us what we are doing wrong, that would be great.

Hello,

Me too. Additionally, when you view the vectors for Relative Colorimetric intent, it shows pixels within the gamut of the output profile shifting dramaticallyas I understand it, R.C. rendering is supposed to clip the OOG colours only, leaving all else alone.

At 12:49 PM -0700 6/28/06, Giordano Galli wrote:

Hi, sorry for the delayed reply…

there should typically be a difference but it does depend on the tables in the profile. Also, when you say there is a difference among all the intents but perceptual how does that work. Wouldn’t perceptual look different than the others? I’m a bit confused with your description…

It uses both. The original image or Lab data is converted first to device space (RGB, CMYK, whatever) using the BtoA tags and the selected rendering intent. Then the device values are converted back to Lab for plotting. The difference between the source and destination Lab values is the vectors you see.

At 9:03 AM -0700 7/4/06, devlinr wrote:

I’ll need a more detailed description of what you are seeing or concerned with before I can give you a full answer. From our testing we have been able to confirm that the intents are working properly here…

One gotcha is working space or monitor profiles that are matrix-based as they have no intent tables within and can only do relative or absolute colorimetric conversions. I had considered disabling perceptual and saturation intents for these profiles but I decided to leave it as Photoshop presents it to the user (though I think Photoshop should disable unavailable intents)

Regards,

Steve


o Steve Upton CHROMiX www.chromix.com
o (hueman) 866.CHROMiX


Post generated from email list

At 7:41 PM -0700 7/4/06, Kaisa Breeden wrote:

Yes and no. I was surprised by this when I first wrote the graphing routines.

In fact, relative colorimetric does shift colors relative to the new profile’s white point. If the white point of the new profile is considerably different than the source profile you will see significant vectors for all colors. They are largest near white and smallest near black.

In order to see the smallest shift possible you should choose absolute colorimetric as the intent. Then you should only see in-gamut shifts when there is a potential error with the profile.

Though we rarely output images with the abs col intent, that is the only one that tries to reproduce all colors as close to a measurable match as possible. Not that rel col is a bad idea. Colors really should be mapped relative to the new paper white so they look natural on the page. Perceptual and Saturation intents also do this mapping but we expect to see shifts like this.

To see it clearly, plot a white color from the source to destination and flip between the intents…

I hope this helps. I’m glad to see there’s discussion of these topics on the forums as I’m sure others are surprised by these things as well.

Regards,

Steve


o Steve Upton CHROMiX www.chromix.com
o (hueman) 866.CHROMiX


Post generated from email list

Alright, here goes.

I open “U.S. Web Coated (SWOP) v2” in Profile Inspector and confirm that there are four different curves for the four different intents. Then I open an image in Grapher and plot it as vectors using the above profile. When I select Saturation, Relative, and Absolute, the vector plot changes. That is, I see a visual change occur each time I select one of these intents. When I select Perceptual, the vector plot does not change. That is, the plotted vectors do not move from the position of the previous plot: if I select Saturation and then select Perceptual, the Perceptual plot looks the same (remains the same) as the Saturation plot; if I select Relative and then select Perceptual, the Perceptual plot remains the same as the Relative plot; and so on.

I have a PC version of ColorThink.

At 1:16 PM -0700 7/6/06, devlinr wrote:

ah… so it’s like the perceptual choice isn’t “firing” huh?..

interesting… OK, we’ll look into it here…

Regards,

Steve


o Steve Upton CHROMiX www.chromix.com
o (hueman) 866.CHROMiX


Post generated from email list

Hi,

I just sent a messsage through the Feedback System, and this is what happens to me:

Graph in 3D:

  1. Load image
  2. Select “Plot as vectors”
  3. Select Destination Space (in my case, a printer space)
  4. Absolute Colorimetric is the default intent = renders change in vectors OK.
  5. Then select Perceptual intent = NO change in vectors (ie. “Absolute Colorimetric” render has “stuck”)
  6. Then select Relative Col intent = renders change in vectors OK.
  7. Then select Perceptual again = NO change in vectors (ie. This time, “Rel Colorimetric” render has “stuck”)
  8. Then select Saturation intent = Renders change in vectors OK.
  9. Then select Perceptual again = NO change in vectors (ie. This time, it’s gotten “stuck” on Saturation!)

Sure enough.
It looks like we have a bug in the way vectors are displayed with the perceptual rendering intent. We have confirmed this in house for both Mac and PC platforms.

We will track this down and get a fix for it in our next update.

Pat,

Any confirmation of the 50 other bugs I have reported? See my bug list in an earlier thread for details.

–John

Hi Steve

I’m perfectly fine with delays, especially when replies have to come from Steve Upton in person.
I was under the impression that there was a lack of interest on your side: I apologize for this.

I confirm the descriptions given by Devlin and Kaisa, the only difference being in the profiles I used for my tests (eg Euroscale Coated v2).

I also appreciate Patrick’s commitment to a fix. Meanwhile, does anybody know of a way to switch the rendering intents of a profile to fool ColorThink into plotting perceptual vectors when I choose, let’s say, saturation?

Thank you all for your interest and help!

Giordano

I don’t know if this is a help for anyone with the above issues, but I have discovered a way:

To trick ColorThink Pro into giving you a 3D Perceputal Rendering:

  1. Select Graph in 3D
  2. In Plot List, select your destination printers profile & give it a colour.
  3. In Plot List >Open your image.
  4. Select Vectors
  5. In Destination, select any profile BUT the one you really want.
  6. Select the Perceptual intent.
  7. Now, in Destination, select the profile you REALLY want.
  8. It renders with Pereptual OK.
    Keep a snapshot for comparison with any further intents you select, as once youre out of Perceptual, you cant ask it to go back!

Hi Kaisa

I don’t know if this is a help for anyone with the above issues,
but I have discovered a way

Thanks, but it doesn’t seem to work here. Using both the “Euroscale Coated v2” and the “U.S. Sheetfed Coated v2” profiles, the ‘perceptual’ plot I get is identical to that of the non-perceptual intent that was previously selected in the interface.

Could you please let me know which profiles you chose for your tests?

Thank you very much.

Giordano

The trick was/is to let CTPro chew on a “dummy” Destination profile with Perceptual Intent selected, then, leaving Perceptual selected, go and choose your preferred Destination profile. If you change nothing else but your Destination profile, it will render your preferred profile in Perceptual.

Methinks you have perhaps found another bug. I too get no joy using the above “trick” when I select US Sheetfed Coatedit rendered Perceptual OK, but when I then selected Rel.Col it got “stuck” on the Perceptual rendering!

With a ProPhoto image plotting vectors to an inkjet profile, the trick works. While this bug exists, CTPro is not terribly useful to me.

I even went back and tried working the trick in reverse (letting CTPro chew on a dummy Destination profile with Rel.Col. Intent selected, and then go and choose the US Sheetfed Coated Destination profile) but the resultant rendering was the same.

Hi Kaisa

Thanks for your answer.

when I select US Sheetfed Coatedit rendered Perceptual OK

So, if I understand you correctly, you choose some bogus output profile, select the Perceptual intent and then choose “US Sheetfed Coated”: the plot you get this way is different from both the RelCol and the AbsCol plots. Could you please let me know the name of the bogus profile? I’d really like to duplicate your steps since whatever Perceptual plot would serve my purpose of illustrating rendering intents.

While this bug exists, CTPro is not terribly useful to me

This is troublesome to me too: when using the software in colour classes I have to put forward that I wont be able to plot Perceptual conversions due to a bug in the software.

Thanks again.

Giordano

Yes, you choose the bogus output profile in order to have the Intent set on Perceptual before you select your preferred profile. (It’s like you’re “pre-selecting” Perceptual Intent). It will then give you a Perceptual plot of your preferred profile.

But once you have that precious Perceptual plot, grab a shot of it or save it as a QuickTime, because once you select another Intent, it won’t go back to plotting Perceptual. You’d have to go through the whole routine again.

For the bogus destination, I select just about anything at random (eg. Adobe RGB). It doesn’t seem to matter, just as long as I can get that Perceptual setting on.

Hi Kaisa

It will then give you a Perceptual plot of your preferred profile.

Unfortunately it gives me the plot of the colorimetric intent that was previously selected.

When I switch from the bogus to the real profile the plot changes due to the differences between their gamuts (so that it could seem that CTP is plotting the perceptual intent), but if I take a screen shot, then change the rendering intent to the aforementioned colorimetric intent and take a second screen shot, the two images are the same. Too bad…

Giordano

Funny. Doesn’t do that for me… You using the beta update? (I was, but I can’t access ColorThink Pro today because it’s telling me my update has expired and I’ve yet to hear wherefore art the darn update.)