Matching printer output

[Greg Miller]

In theory, it makes sense to use relative colorimetric when the saturated colors are unimportant (a red light on an instrument panel in the background) and perceptual when they're important (you're shooting the Uniqlo catalog). In reality I find I have to compare. For most of my work it makes very little difference; when it matters, I'm often surprised by what's better.

Rendering intents only affect colors outside the gamut of of the profile. So given a photo where all colors are contained within the gamut of the profiles, printing using Relative Colorimetric or Perceptual should yield an identical print.

Tyler's comment is accurate for images wheres there are out of gamut colors. Out of gamut colors will cause all colors (both in-gamut and out-of-gamut) to shift if using Perceptual, while using Relative Colorimetric will only change colors that our out of gamut (while leaving in-gamut colors untouched).

[Tyler Boley]

the profile conversion does not know if there are out of gamut colors. It simply sees (for example) something like ProPhoto source, and a smaller destination space like HP ink on RagPhoto... Perceptual will reduce all colors down to the smaller space in some proportionally subjectively visually pleasing manner, based on the difference between source and output space, irrelevant of image. Even if all the image colors are already in gamut. Colormetric matches source to destination colors (in a perfect world), simply clipping anything that doesn't fit the smaller space.. or in the reverse situation, not expanding color from a smaller space to a larger...
there is a slightly more complex matching process called cross rendering, used for press proofing, that may provide better matches, but at the expense of taking the full advantage of each output system. It's even easily available printing out of photoshop. Were it me I'd leave some wiggle room in the expectations of matching, and want the best most pleasing result from each system, given that each "feels" the same, if that makes any sense.... but I certainly had clients requiring a more particular result.

[andy]

the profile conversion does not know if there are out of gamut colors. It simply sees (for example) something like ProPhoto source, and a smaller destination space like HP ink on RagPhoto... Perceptual will reduce all colors down to the smaller space in some proportionally subjectively visually pleasing manner, based on the difference between source and output space, irrelevant of image. Even if all the image colors are already in gamut. Colormetric matches source to destination colors (in a perfect world), simply clipping anything that doesn't fit the smaller space.. or in the reverse situation, not expanding color from a smaller space to a larger...
there is a slightly more complex matching process called cross rendering, used for press proofing, that may provide better matches, but at the expense of taking the full advantage of each output system. It's even easily available printing out of photoshop. Were it me I'd leave some wiggle room in the expectations of matching, and want the best most pleasing result from each system, given that each "feels" the same, if that makes any sense.... but I certainly had clients requiring a more particular result.

I had a feeling that's how perceptual worked, but I've had trouble finding source info to back it up--thanks for the unwitting confirmation.

[Greg Miller]

Perceptual will reduce all colors down to the smaller space in some proportionally subjectively visually pleasing manner, based on the difference between source and output space, irrelevant of image. Even if all the image colors are already in gamut.

I've never seen this happen. I just did some testing converting images from ProPhoto RGB to sRGB where all colors in-gamut for sRGB. I'm not seeing any real world color changes or re-mapping.

[Tyler Boley]

there are no icc standards for mapping in perceptual, it's totally at the descretion of the profile making software designer. So, some map in gamut colors differently than others with perceptual. some may try to keep them as close as possible. Datacolor even said they try to make in gamut color map more like a colormetric intent, with color at and approaching the gamut boundary mapping more like saturation intent. There are standards for colormetric converions that software designers adhere to. I'm only stating the general theories, they may apply differently with different profiles from different manufacturers. I found in gamut colors remapped very differently depending on the profiler in StudioPrint, Monaco, DataColor, and Gretag. Monaco's was preferred by many with in gamut colors retaining a very pleasing level of saturation, and they carried that preference over to the perceptual mapping in Xrite i1 Profiler as well. But the nature of perceptual remapping is dependent on source and destination space, not image content.
There was talk of a "smart cmm" that could map differrently depending on image colors years ago, but it never came to pass and is problematic in concept. Pat from Chromix could state all this more eloquently than I should he chime in.