DavidBrainard
sRGB
edit8/22/10 -- sRGB standard. This is a longer comment about the edit I made (and that someone else undid) to the sRGB matrix specified on the wikipedia sRGB page.
It is true that there is a definition (from 1996) of the sRGB standard at http://www.w3.org/Graphics/Color/sRGB.html. But the standard published by the iec (IEC 61966-2-1:1999) differs in various small ways from this earlier document. My edit of the transformation matrix brought the wikipedia page into line with the standard.
Note also. Before I edited page, it already refered to the iec standard as the official definition. Also, even with the undo of my edit of earlier today, what is on the wikipedia page did not match what is at http://www.w3.org/Graphics/Color/sRGB.html. The threshold constants for the linear portion versus power function portion of the gamma correction at http://www.w3.org/Graphics/Color/sRGB.html also differ from the final standard. These, however, matched the standard in what was there when I found the page this morning. That is, prior to my edit the wikipedia page matched neither the official standard nor the earlier http://www.w3.org/Graphics/Color/sRGB.html.
Unless there is some compelling reason why the iec document should not be considered the archival source for sRGB, I think the wikipedia page should match it. I will shortly undo the undo of my edit.
A quick review suggests that the numbers in the draft iec standard (http://www.colour.org/tc8-05/Docs/colorspace/61966-2-1.pdf) match those in the final version.
I'm new to editing wikipedia, but not new to color vision. If there is a better place for me to write these comments and have them connected with the edit trail, please let me know. I couldn't fit very much into the small comment box available when one edits.
- Hi David. The best place for this discussion is probably talk:sRGB (where everyone interested in the subject will see it), but anyway, you’re probably fine in supporting the IEC document as the most authoritative standard; I hadn’t actually looked at that document. Because these matrices differ only in the 4th decimal place, in practice it probably makes no difference whatever which one is used, but I’m curious about the differences. It would seem that doing the math at high precision (that is, the truly official part of the standard is the chromaticity coordinates of primaries and white point, and the XYZ -> RGB matrix is only a derived quantity) and then rounding to the nearest ten-thousandth would yield a single obvious approximation at that precision. Maybe I’ll run the numbers myself sometime and see what I get.
- While I’m at it, welcome to Wikipedia. Hope you like the place and decide to stay: we can definitely use your help. Cheers, jacobolus (t) 00:05, 23 August 2010 (UTC)
- Thanks! [[Last night my 15 year old son explained this all to me at dinner!] I have indeed repeated much of the above in what I think is a more useful form on the sRGB discussion page. I added my speculation there about why the matrices differ, which is similar (i.e. rounding differences) to your suggestion above. I also made a small change to the sRGB page to clarify the source of the numbers provided there. DavidBrainard (talk) 02:37, 24 August 2010 (UTC)