Anybody else unhappy with Red River's R3000 profiles?

Started Mar 23, 2013 | Discussions
Hugowolf
Forum ProPosts: 11,209
Like?
Re: Brian, you win!
In reply to Petruska, Mar 25, 2013

Petruska wrote:

Brian, you win!

Hmm, you obviously haven't seen my bracket.

Brian A

Reply   Reply with quote   Complain
jtoolman
Senior MemberPosts: 4,041
Like?
Re: Brian, you win!
In reply to Petruska, Mar 25, 2013

Petruska wrote:

Hugowolf wrote:

Have your tried using a perceptual rendering intent instead of relative colorimetric? ICC profiles allow for six lookup tables (LUTs), two for each intent. There should be an input to connection space table, and a connection space to output device table for each intent supported, and even an inexpensive profile builder should support tables for at least perceptual and relative colorimetric.

Brian A

Yes, Perceptual intent produces the correct purple block. There is still something wrong with a lot of the R3000 RR profiles as that purple block soft-proofs blue with Relative and soft-proofs as purple with Perceptual. Other RR profiles, Ex: Paper Canvas, soft-proofs purple for both Relative and Perceptual. All my custom Red River paper profiles made with the Xrite I1PRO or Xrite Pulse system soft-proof as purple with both Relative and Perceptual.

Most users use Relative intent, I don't, I use Perceptual 99.9% of the time as I print mostly portraits and find the out of gamut remapped lighter colors of Perceptual more pleasing. This is why I have never noticed anything wrong with RR supplied profiles when I first got my R3000. My custom profiles eliminate all issues. Relative intent gives more saturated colors, especially with BP Comp. turned off. So I can see where some users may be upset with not being able to use it correctly with some prints. The Epson profiles also soft-proof the purple correctly with Relative or Perceptual. RR has a problem.

Bob P.

Bob, does it matter if one use a specific rendering intent used for printing the test patches during the production of the profile?

Will the user then be "Locked" to the specific Rendering Intent used during the production of that profile when one prints using that same profile?

Reply   Reply with quote   Complain
Petruska
Veteran MemberPosts: 6,481
Like?
Re: Brian, you win!
In reply to jtoolman, Mar 25, 2013

jtoolman wrote:

Petruska wrote:

Hugowolf wrote:

Have your tried using a perceptual rendering intent instead of relative colorimetric? ICC profiles allow for six lookup tables (LUTs), two for each intent. There should be an input to connection space table, and a connection space to output device table for each intent supported, and even an inexpensive profile builder should support tables for at least perceptual and relative colorimetric.

Brian A

Yes, Perceptual intent produces the correct purple block. There is still something wrong with a lot of the R3000 RR profiles as that purple block soft-proofs blue with Relative and soft-proofs as purple with Perceptual. Other RR profiles, Ex: Paper Canvas, soft-proofs purple for both Relative and Perceptual. All my custom Red River paper profiles made with the Xrite I1PRO or Xrite Pulse system soft-proof as purple with both Relative and Perceptual.

Most users use Relative intent, I don't, I use Perceptual 99.9% of the time as I print mostly portraits and find the out of gamut remapped lighter colors of Perceptual more pleasing. This is why I have never noticed anything wrong with RR supplied profiles when I first got my R3000. My custom profiles eliminate all issues. Relative intent gives more saturated colors, especially with BP Comp. turned off. So I can see where some users may be upset with not being able to use it correctly with some prints. The Epson profiles also soft-proof the purple correctly with Relative or Perceptual. RR has a problem.

Bob P.

Bob, does it matter if one use a specific rendering intent used for printing the test patches during the production of the profile?

Will the user then be "Locked" to the specific Rendering Intent used during the production of that profile when one prints using that same profile?

Not sure of the answers, I would need to dig into that.

Now we are printing the test patches with no color management, thus that is without an ICC profile and intents. When we make custom ICC profiles we have the 4 comonly used intents available and they do what they are supposed to do with the colors.

Obviously this is a subject that we could dig too deeply into.  If one looks at all the R3000 RR profiles you will see some that work just fine soft-proofing that purple block with either the Relative or Perceptual, where other RR profiles have a problem with just the Relative intent.

I looked at the RR APL profiles for the R3000, R2880, and 3880 and all 3 soft-proof with blue for purple with Relative intent, and purple with Perceptual.  Looking at the Canon PRO9000 RR APL profile you get purple with both Relative and Perceptual.  I say it's a broken RR profile for a lot of RR papers and Epson printers when using Relative intent.

My custom profiles are fine with both intents on any paper/ink/printer.

Bob P.

Reply   Reply with quote   Complain
jtoolman
Senior MemberPosts: 4,041
Like?
Re: Brian, you win!
In reply to Petruska, Mar 25, 2013

Petruska wrote:

jtoolman wrote:

Petruska wrote:

Hugowolf wrote:

Have your tried using a perceptual rendering intent instead of relative colorimetric? ICC profiles allow for six lookup tables (LUTs), two for each intent. There should be an input to connection space table, and a connection space to output device table for each intent supported, and even an inexpensive profile builder should support tables for at least perceptual and relative colorimetric.

Brian A

Yes, Perceptual intent produces the correct purple block. There is still something wrong with a lot of the R3000 RR profiles as that purple block soft-proofs blue with Relative and soft-proofs as purple with Perceptual. Other RR profiles, Ex: Paper Canvas, soft-proofs purple for both Relative and Perceptual. All my custom Red River paper profiles made with the Xrite I1PRO or Xrite Pulse system soft-proof as purple with both Relative and Perceptual.

Most users use Relative intent, I don't, I use Perceptual 99.9% of the time as I print mostly portraits and find the out of gamut remapped lighter colors of Perceptual more pleasing. This is why I have never noticed anything wrong with RR supplied profiles when I first got my R3000. My custom profiles eliminate all issues. Relative intent gives more saturated colors, especially with BP Comp. turned off. So I can see where some users may be upset with not being able to use it correctly with some prints. The Epson profiles also soft-proof the purple correctly with Relative or Perceptual. RR has a problem.

Bob P.

Bob, does it matter if one use a specific rendering intent used for printing the test patches during the production of the profile?

Will the user then be "Locked" to the specific Rendering Intent used during the production of that profile when one prints using that same profile?

Not sure of the answers, I would need to dig into that.

Now we are printing the test patches with no color management, thus that is without an ICC profile and intents. When we make custom ICC profiles we have the 4 comonly used intents available and they do what they are supposed to do with the colors.

Obviously this is a subject that we could dig too deeply into. If one looks at all the R3000 RR profiles you will see some that work just fine soft-proofing that purple block with either the Relative or Perceptual, where other RR profiles have a problem with just the Relative intent.

I looked at the RR APL profiles for the R3000, R2880, and 3880 and all 3 soft-proof with blue for purple with Relative intent, and purple with Perceptual. Looking at the Canon PRO9000 RR APL profile you get purple with both Relative and Perceptual. I say it's a broken RR profile for a lot of RR papers and Epson printers when using Relative intent.

My custom profiles are fine with both intents on any paper/ink/printer.

Bob P.

If you want to really see something shocking, do your softproofing with either intent using preview, but click on a profile and then use you up or down arrows to toggle up and down the list of profiles.

As you know, I have several hundred of them loaded. The rendition of that image is all over the place as you run through all the profiles. There is nothing even resembling any consistency.

I really need to make a video demonstrating this just so you can really see just how much the rendition of practically all aspects of an image changes with each profile. I am still speechless.

Joe

Reply   Reply with quote   Complain
Hugowolf
Forum ProPosts: 11,209
Like?
Re: Brian, you win!
In reply to jtoolman, Mar 26, 2013

jtoolman wrote:

Petruska wrote:

Hugowolf wrote:

Have your tried using a perceptual rendering intent instead of relative colorimetric? ICC profiles allow for six lookup tables (LUTs), two for each intent. There should be an input to connection space table, and a connection space to output device table for each intent supported, and even an inexpensive profile builder should support tables for at least perceptual and relative colorimetric.

Brian A

Yes, Perceptual intent produces the correct purple block. There is still something wrong with a lot of the R3000 RR profiles as that purple block soft-proofs blue with Relative and soft-proofs as purple with Perceptual. Other RR profiles, Ex: Paper Canvas, soft-proofs purple for both Relative and Perceptual. All my custom Red River paper profiles made with the Xrite I1PRO or Xrite Pulse system soft-proof as purple with both Relative and Perceptual.

Most users use Relative intent, I don't, I use Perceptual 99.9% of the time as I print mostly portraits and find the out of gamut remapped lighter colors of Perceptual more pleasing. This is why I have never noticed anything wrong with RR supplied profiles when I first got my R3000. My custom profiles eliminate all issues. Relative intent gives more saturated colors, especially with BP Comp. turned off. So I can see where some users may be upset with not being able to use it correctly with some prints. The Epson profiles also soft-proof the purple correctly with Relative or Perceptual. RR has a problem.

Bob P.

Bob, does it matter if one use a specific rendering intent used for printing the test patches during the production of the profile?

Will the user then be "Locked" to the specific Rendering Intent used during the production of that profile when one prints using that same profile?

It shouldn't matter. The patches are produced without color management by the driver and without color management from the software. The media setting, which dictates things like how much ink is put down does matter, but without CM, there is no redering intent used. Without color management, you can't use a rendering intent.

It is up to the profiling software to produce LUTs for each of the three common intents from those patches, either iteratively, as with ColorMunki, or using a much larger number of patches in a single pass with the more expensive alternatives. All three intents, perceptual, colorimetric, and saturation, can be calculated from the unmanaged output. (The difference between relative and absolute colorimetric is only in the position of the white point, so the same table pair is sufficient for both.)

How colorimetric and saturation tables are calculated is relatively simple with little variance between methods, Shifting the gamut within the space for perceptual is much more of a back art, and will vary greatly between implementations.

And it should be noted that there are some obvious differences here between version 2 and version 4 International Color Consortium profiles. You could try reading the specs at icc.org, but like any commitee driven specs, they are as thick as mud. It is probably better to start with someone who has to deal with the fallout of those specs, such as Argyll, if you want a better understanding:
http://argyllcms.com/doc/iccgamutmapping.html

Brian A

Brian A

Reply   Reply with quote   Complain
just Tony
Contributing MemberPosts: 530
Like?
Re: Brian, you win! Perceptual is a definite improvement
In reply to Petruska, Mar 26, 2013

I had been lulled by the often written advice to stick with Relative unless the out of gamut content filled a relatively large percentage of the image area. In this image something minimal like 5% of the area was out of gamut according to what I saw in the CS6 preview window. I was worried that Perceptual would shift too many of the in-gamut colors.

I now see that was an unfounded fear. The skin tones remained good, and several sections benefited greatly. The sky in the Monument Valley scene has better cloud definition, the black of the CD player is more open, and the strawberries look more inviting. Much better overall.

Reply   Reply with quote   Complain
Hugowolf
Forum ProPosts: 11,209
Like?
Re: Brian, you win! Perceptual is a definite improvement
In reply to just Tony, Mar 26, 2013

just Tony wrote:

I had been lulled by the often written advice to stick with Relative unless the out of gamut content filled a relatively large percentage of the image area. In this image something minimal like 5% of the area was out of gamut according to what I saw in the CS6 preview window. I was worried that Perceptual would shift too many of the in-gamut colors.

I now see that was an unfounded fear. The skin tones remained good, and several sections benefited greatly. The sky in the Monument Valley scene has better cloud definition, the black of the CD player is more open, and the strawberries look more inviting. Much better overall.

With a half way decent profile, you should not have these problems with relative colorimetric. Indeed, one of the main reasons for having several types of color rendering intents (putting aside the saturation and absolute colorimetric) is to give you a choice of what works best for a particular image given its gamut, and the output gamut of your printer/inks on your chosen paper.

It definately sounds like a RR profile problem. To quote RR "Rendering Intent: We recommend Relative Colorimetric for most prints. You can also try Perceptual if your results are not satisfactory."

The Outback test image is designed to show flaws in workflow and profiles. And it certainly seems to have done that this time.

Brian A

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads