Epson 3800 - problem printing certain yellows

[hugowolf]

Thanks, I'll try it. I stuck with Relative Colorimetric a few years ago because it's supposed to preserve more the image's original colors than Perceptual does (according to the description in the Photoshop print dialogue). But I also just noticed that Perceptual is described to work better with out of gamut colors.

Honestly, I used to compare those two rendering intents quite often and never saw a substantive difference so I just stuck with Relative Colorimetric. I just reprinted an image for my exhibition with Perceptual (the first print was Relative Colorimetric) but the out of gamut colors looked the same to me

But I'll try it again, taking a closer look at how they both render out of gamut colors.

Relative colorimetric will be more accurate and produce a larger gamut, providing there are few or no areas of out of gamut colors. If there are sufficiently many pixels out of gamut, then perceptual's in gamut compression will some times give more pleasing results. If you aren't checking for out of gamut colors, then perceptual is the safer choice.

Perceptual’s problem is that it compresses whether there are out of gamut colors or not.

Brian A

[Robert Ash]

Brian,

Thanks. Will keep that in mind. I'd forgotten that distinction, it's been years since I'd studied the topic and I never looked at that mapping as a problem necessarily but you're right, it could be one because theoretically the proportional reduction of all values along the color axes would shift every color.

In practice, as I've shared, I've rarely seen any visible difference between the two intents. There have been specific cases now and then but it happened so seldom I stopped comparing them some years ago and just stuck to Relative Colorimetric based on Adobe's summary description the Photoshop print dialog.

[Robert Ash]

Thanks. It will be great if you can do this. I'm sure there are a lot like me that would like to see what causes the problem and how/if it might be resolved. Could happen to any of us any time soon, and forewarned etc.....

Ok, I'll try to provide an example, I have a good one in mind, it's a capture of Firefall from a Yosemite workshop.

[RHPS]

A relatively old NEC monitor (wireframe) about 96% of Adobe RGB vs the AdobeRGB (1998) profile (Solid color).



Notice the difference between those saturated yellow/oranges?

Brian A

But the 3800 (wireframe) is not so bad in the yellow/orange, which is why I'm eager to find out how it goes so wrong.

[hugowolf]

But the 3800 (wireframe) is not so bad in the yellow/orange, which is why I'm eager to find out how it goes so wrong.

But what is this a comparison of? What exactly do the wireframe and solid represent?

Brian A

[RHPS]

Solid is AdobeRGB and wireframe is 3800 on Harman Gloss Baryta.

[hugowolf]

Solid is AdobeRGB and wireframe is 3800 on Harman Gloss Baryta.

So the images are processed in ProPhotoRGB, monitors have colors outside of AdobeRGB, and the problems occurs on Canson Platine Fiber Rag, Epson Exhibition Fiber and Epson Premium Luster. I am still not sure where you are going with this.

Brian A

[RHPS]

I'm simply trying to understand why these problematic yellows can be mapped successfully to the monitor color space but not to the printer color space. I'm suggesting that the printer gamut is similar to (if not better than) the monitor gamut in the yellow/orange region so that should not be a big factor. If it happened with just one paper/profile we might suggest a bad profile, but can we believe all the printer profiles are defective? So what's the common factor?

If I examine images in ProPhoto I don't see any real problems with yellows turning muddy orange. But then I don't know exactly what colors the OP is having trouble with. Maybe my 3800 will make a mess of these yellows. I would just like to know

Robert - can you clarify whether you have soft-proofed the images using the relevant printer profiles? If so, does the soft-proof show the problem?

[Farmer]

This is where we discover that the concept of "in gamut" isn't so straight forward.  Gamut viewed or measured under what circumstances and at what distance, etc?

The muddy looking result wouldn't look muddy at a longer distance because your eyes wouldn't perceive the individual variations.  Of course if you inspect any inkjet image closely enough, the colours change (until you get to individual dots).

A monitor creates colour in a different way to a printer - the printer only has yellow in this case (unlike, say, a 900 series Epson which also has orange) and so "dark yellow" or various oranges may be less "pure" than you're looking for.  Overall, it's still within gamut, but the result may not be ideal on close inspection (but fine at longer distances).

Just something to think about.  Sometimes the science has to give way to the art in order to really see what's happening :-)

[Robert Ash]

RH,

You can't see gamut problems from the JPEG files I posted, those are in sRGB color space for the web. Once they're exported in sRGB color space they're permanently clipped (compressed, actually) and putting them back into ProPhoto RGB cannot reverse the clipping.

It's like a Thanksgiving turkey If you cook the turkey in a large pot then clip the turkey to fit it into a smaller pot and eat the clipped portion, then returning the remaining turkey back to the larger pot will not make the turkey the same size again that it was before being clipped.

If someone here can recommend a good no-cost site where I can upload a file or two, I'll downsize a flattened PSD or TIFF that exhibits the problem and upload it for you guys.

[Robert Ash]

Phil,

Very thoughtful post, but from what I've noticed:

• The problem isn't dependent on viewing distance.
  
• The problem doesn't seem to happen with other colors (at least not from what I've noticed).

Regarding the 3800's yellow ink, that's thought-provoking because theoretically the printer should be able to use only yellow ink in the problem areas....... I essentially forced it to do so in one case but that was a couple of years ago and I forgot to record all the conniptions I needed to do to make that happen......

[RHPS]

RH,

You can't see gamut problems from the JPEG files I posted, those are in sRGB color space for the web. Once they're exported in sRGB color space they're permanently clipped (compressed, actually) and putting them back into ProPhoto RGB cannot reverse the clipping.

I realise that. I was actually referring to other Prophoto files that I had examined.

If someone here can recommend a good no-cost site where I can upload a file or two, I'll downsize a flattened PSD or TIFF that exhibits the problem and upload it for you guys.

You could try wikisend.com It's free and seems to work OK.

[dgberg]

At least in your JPEG I see an out of gamut issue.
In PS CS6 using the paper profile you have chosen a good portion of the cliff wall is out of gamut. (Not the falls.)
Remove the gamut warning and your yellowish cliff wall turns into a muddy orange as you describe.
In ColorThink some of the yellow,gold is out of gamut for the 3800 with your Epson PLPP paper profile as well.

[Farmer]

Robert - if you printed a large enough page of that colour and stood far enough away, it would look OK.

Printing just with yellow means you have yellow and paper-white as your mixers and anything that can't be created with that alone is going to need input from the other colours or the blacks.

Eric Chan and I actually delved into this some years ago and I was in a position to send him samples from a 900 series using orange and we can see the differences compared to his 3800.  The colours were "in gamut" but still not as pure as desired but the addition of orange and the updated LUT on the 900 series improved it.

I think there are just some "hot spots", for want of a better term, that can show up even when in gamut (or moved into gamut with a rendering intent).  Slight adjustment of the colours may help, otherwise I suspect there's not a lot you can do to resolve it, unfortunately