- This topic has 19 replies, 2 voices, and was last updated 7 years, 4 months ago by
.
-
Posts
-
Well, thanks to your suggestions and some observations I seemingly managed to get it into some semblance of order, thanks to your help, Florian.
Two questions though:
1. I have these printouts that look more saturated than LUT profiles that pass verfication passes. They’re closer to Curves profiles although the latter would inevitably show large dEs. So who should I really trust – the display and the numerical results or the prinouts by a professional laboratory that does museum prints?
2. Back to the original question, could you explain what are the reasons for imperfect additivity, i.e. why can’t we measure just R, G, and B alone without measuring combintations separately for profiling? I’m interested in theory, not practical implications, since normal logic would suggest that the phosphors shouldn’t really interfere with their neighbors of different color. Is that essentially what’s happening in displays? Is it different between the technologies (CRT, LCD…)? It would be really interesting to know. If there’s a white paper somewhere I’d love to read it.So who should I really trust – the display and the numerical results or the prinouts by a professional laboratory that does museum prints?
Definitely the former. You don’t know how accurate the prints are colorimetrically until you verify them (e.g. with a specrometer), which would also require some test patches to be present on the printout and in my experience anything that’s not explicitly sold as hardcopy proof usually doesn’t provide a trustworthy basis for colorimetric evaluation (the aim for non-proof is usually pleasing rendition, not colorimetric accuracy). I would attribute the better visual match of the printout to the curves-based profile as pure coincidence.
As an example, a while ago I examined a Fuji print, and the errors towards the reference values obtained via the print profile (sRGB source material) was over 6 dE76 average and outliers of over 30 de76 (and if I remember correctly this was even ignoring the substrate white difference). The first impression visual match to the display wasn’t as bad as the numbers implied, although on closer inspection the errors were very apparent.
In a carefully set up, controlled soft-proof environment, it’s usually possible to obtain a good screen-to-print match (if both display + print are characterized with high enough accuracy) both visually as well as colorimetrically.Back to the original question, could you explain what are the reasons for imperfect additivity, i.e. why can’t we measure just R, G, and B alone without measuring combintations separately for profiling?
This would require the display behaving perfectly additively. Real displays don’t fall in that category (although they can be quite good), e.g. due to the display characteristics, internal processing, and in case of Plasma due to things like ABL, or the power supply not providing stable output in all scenarios.
I’m interested in theory, not practical implications, since normal logic would suggest that the phosphors shouldn’t really interfere with their neighbors of different color. Is that essentially what’s happening in displays?
In theory, the plasma cells should be completely independent from one another, but in reality I’m inclined to think that there is some crosstalk.
Is it different between the technologies (CRT, LCD…)?
Yes. The devices with best additive behavior have typically been CRTs.
If there’s a white paper somewhere I’d love to read it.
I don’t know of any specific white papers, but there’s whole books on the topics of additive color mixing and colorimetry.
