Starting from Raw, when you know your final image needs to be in sRGB, does it make more sense to
1) open it and perform PP in a much larger color space like ProPhotoRGB, converting to sRGB at the very end; or
2) open the Raw file directly in sRGB and stay in it throughout?
A few years ago I used to use Melissa D65 as my primary working color space, fine tuning images in that large space and only converting to sRGB at the end of the worlflow if needed. Often in these cases the sRGB version required additional fine tuning but at least the original with all my adjustments would be in Melissa D65 for archival purposes and I wouldn't have to revisit it in the future if/when monitors/media improved. Or so went the theory.
It worked well, except when I realized that I was spending a lot of time re-fine tuning most of my keepers because the vast majority of them needed to be turned into sRGB after all. Statistics to the rescue: 90+% of my keepers need to be in sRGB because someone wants a copy via email or because someone wants to make a wallmart print today - only 1 or so a month get the special fine-art treatment, eventually being printed large to perfection.
And I started thinking that if red flowers clipped when going straight to sRGB, they probably still will when ending up in sRGB after ProPhoto: except that in the latter case you'd only realize that you are clipping them at the end of your session, adding additional PP time to get them the way you want them. So I now do it the other way around: sRGB throughout for most keepers and only start in aRGB or a larger space with the very few images that I print large. CNX2, which I use on 100% of my captures (90% of the time ending it there, without needing a trip to CS5), makes it easy to make this change after the fact leaving all other adjustments intact. This thread gave me the impulse to revisit this decision.
Thanks to
Tim's Blue Ball,
gollywop (on DPR)'s sunset and
a flower Raw files, all of which have clipped histograms in sRGB that are not clipped in ProPhoto in the areas of interest, I used ACR 6.7 and CS5 to investigate the differences to be expected when (1) opening a Raw file in ProPhoto, applying adjustments and converting to sRGB at the end (perceptual, no black point compensation) vs opening the same file directly into sRGB and (2) sticking with it throughout.
If opened with the neutral camera profile and no adjustments are applied between opening and converting, the images resulting from the two workflows are virtually indistinguishable. Here Tim's blue ball is shown with workflow 2, 1 and just ProPhoto :-) left to right on my calibrated/profiled Dell U2410 monitor run by Win7, which covers 95% of aRGB.
http://imgur.com/dRbwZKIVery slight differences between the two workflows became apparent when adjustments were introduced. Something as simple as changing the ACR 6.7 camera profile from Camera Neutral to Camera Landscape caused some slight but visible differences to appear in the two resulting images, neither necessarily better than the other. Switching back to Camera Neutral and pushing a more aggressive adjustment (CEP3 Tonal Contrast at default settings) resulted in this comparison:
http://imgur.com/QgMIufnThere are tiny localized differences (sRGB only to the left, ProPhoto converted to sRGB to the right). I know where they are, so I can spot them easily. But neither image is clearly more accurate or preferable to the other in my view.
So unless someone has a good argument for otherwise, I think I am going to stick with my current approach: sRGB as my day-to-day working color space, and aRGB/MelissaD65 for the few occasions when I feel it is necessary. As opposed to the other way around.
Cheers,
Jack
PS For those who are wondering, ProPhoto/aRGB from start to finish give clearly better colors on my U2410 monitor than the sRGB/ProPhoto+sRGB workflows discussed above. This is especially evident in gollywop's sunset image.