This content has been marked as final. Show 19 replies
> Recent conversations have led me to believe that there is little point starting with a ProPhoto colour space if the end medium uses a much smaller gamut ...
You don't sound that confused -- you simply need to decide on what your target color space (CS) gamut should be. You're correct in noting that color gamut conversions are somewhat destrustive, but there nothing wrong with minimizing them, and restricting them for going from large to small gamuts only.
A large CS gamut (PpRGB) is capable of all the color from your camera (depending on your camera), but like your friend implies most probably sould be editted in highbit depth (16bits). This CS is useful for archiving your images, or for aggressive tonal adjustments. (... although your images are 'really' archived as raw data, and ACR "undestructively" modified tonal values before exporting to Photoshop).
I find little utility in exporting to small CS gamuts (eg, sRGB), unless the image is intended for a small gamut (monitors in general - eg, web browser presentations, and/or printers that can only relate to sRGB).
In between are intermediate CS gamuts (eg, AdobeRGB), which are quite capable of small tonal adjustments with 8bit depth, for which the target gamuts are usually better printer technologies.
my CA$0.02 :)
As Jeff points out in a recent post, printers get better all the time. There are colors that CAN be printed that are outside of even AdobeRGB.
Jeff Schewe, "ACR INTERPOLATION UPSIZING" #30, 15 Sep 2008 10:29 pm
Thanks for your replies. The concensus so far does seem to be to work in ProPhoto and convert to sRGB at the end. Indeed, I have a shortcut action in Photoshop to convert to 8-bit sRGB.
Latterly, I've been working on internet/lcd images in sRGB all the way through, to avoid conversion. The results have been fine, and I always avoid clipping at the raw conversion stage regardless of the colour space I choose.
My main concern was that ProPhoto is an unnecessarily large colour space to work in, and that I would actually waste some of the 16 bit accuracy on colours never used in the finished article, as well as introduce errors at the conversion stage.
I'm intrigued by the suggestion that I can "contain" out of target gamut colours before final conversion, rather than throw them away at the raw stage. I don't really understand this, as the finished item can't contain the colours anyway. Is this something to do with Perceptual conversion?
I guess I still don't really understand that (as long as I don't clip the raw conversion) working in a huge colour space and converting is any better than sticking to the smaller target colour space, as the colours are more accurately represented in a smaller colour space.
As I don't deal with the web, I don't convert to a smaller color space or image bit depth at all. (Just to clarify.)
You know the blue channel doesn't provide a lot of density in a wide range of colors. The appearance of density is what prevents posterization in highly satruated colored objects like flowers. Pulling back blue gives you yellow which is a very low density color.
What you really have to be concerned about is the red and green channels clipping and especially the green in most colors.
> but for the final conversion those colors must be desaturated slightly.
That's why there's a Gamut Warning. ;)
>This is no problem editing in ProPhoto, but for the final conversion those colors must be desaturated slightly.
You sure? Quite honestly, for sRGB images the odds are you really don't need to bother. You can edit in ProPhoto RGB and softproof in sRGB and see what the conversion will look like after conversion. If you see a problem area, sure, you can do something about it. But the odds are you really won't need to. Saturation clipping is useful to know about in the event you do need to address it but you'll only know if it's relevant if you can see something in the image when softproofed.
As for Gamut Warning...that's pretty old tech in terms of dealing with out of gamut color. It doesn't tell how MUCH is out of gamut or what the gamut clipping will look like...only that it contains SOME color that is out of gamut. Only softproofing tells what the resulting conversion will end up looking like and whether or not it's something to be concerned with.
>Only softproofing tells what
Absolutely. That's why I've been very surprised to read lately that you're printing out of Lightroom. :/ The lack of soft proofing and the library/catalog paradigm have kept me away from Lighroom. I want to see what I'm going to print, and I don't want a scheme where the Finder and the application don't see files and folders in the exact same locations.
Gamut Warning is a useful tool in that it shows you where the out of gamut areas are, so you can focus on them quicker.
>That's why I've been very surprised to read lately that you're printing out of Lightroom.
Lightroom, by itself, is fine for quick prints as is from raws. For final prints, they round trip through Photoshop for final image edits at the pixel level, softproofing and then saved back into Lightroom for printing.
That way I can keep the images in at their "native resolution" and then have the ability to do whatever size prints I need to do and have auto-output sharpening. If I print on different media, I pop the images open from Lightroom into Photoshop, change the softproofing and do new media based tweaks and hit save again and the new layers are then available in Lightroom.
Actually pretty slick integration (still not optimal yet) between working both in Lightroom _AND_ Photoshop to take advantage of the best of both (and mitigate the weaknesses of both). And, printing out of Lightroom really does offer a much, MUCH better printing workflowwhich is why I use both Lightroom and Photoshop. Heck I got both so why not use both?
>And Jeff, I disagree that I wouldn't have to do anything about it.
Well, simply looking at the histograms would tell me anything about the aesthetics of the image and whether or not the saturation clipping (and that's what that is) would make a difference in the image. Soft proofing would...
I suspect there is an adverb missing in Jeff's last paragraph:
>simply looking at the histograms would tell me anything about the aesthetics of the image and whether or not the saturation clipping (and that's what that is) would make a difference in the image. Soft proofing would...
simply looking at the histograms would not tell me anything about the aesthetics of the image and whether or not the saturation clipping (and that's what that is) would make a difference in the image. Soft proofing would...
Well, I feel quite stupid. Despite lots of replies, I'm still not aware if my original question has been answered, maybe because some of the answers go over my head - especially those referring to saturation clipping.
The only clipping I'm aware of is that which occurs at the high and low ends of the histogram, which I usually try to remove during the developing process. I occasionally leave small amounts of clipping, but this is because there is no noticeable benefit to detail, and extra recovery or shadow fill spoils other aspects of the image. I am aware that high/low clipping changes with the target colour space, but don't fully understand why, and I just adjust the developing sliders acordingly.
I re-read Bruce/Jeff's book again last night, and picked up on a reference to an online article written by Bruce on the Creative Pro website, which I skipped on first reading. Here he discusses two approaches to colour spaces input-centric and output-centric and explains the philosophy behind each approach.
Here's the articles, if anyone's interested:
> In some cases it might not even be worth the trouble, which I guess was Jeff's point a few posts back.
Yes...that was my point. I'll also add that trying to determine the best method of converting from PP RGB to sRGB really can't be dictated by the histogram. The histogram is simply a graph of the various levels in a file. It can't tell you what it LOOKS like after the results of a conversion.
> I'll also add that trying to determine the best method of converting from PP RGB to sRGB really can't be dictated by the histogram. The histogram is simply a graph of the various levels in a file. It can't tell you what it LOOKS like after the results of a conversion.
if having that information available suggests a way to approach the problem.
Anything that I can do to optimize the manual part of this process is a good thing.
The auto gamut mapping from a larger gamut to a smaller gamut is a very interesting but challenging problem. You can find algorithms via a Google search, but please note this is an advanced topic and definitely not as simple as histogram scaling or gamut volume scaling.
> The auto gamut mapping from a larger gamut to a smaller gamut is a very interesting but challenging problem.
As I am learning.
> You can find algorithms via a Google search, but please note this is an
advanced topic and definitely not as simple as histogram scaling or gamut volume
I'm an engineer. Even if I can't figure out some way of automating any part of
this, the exercise in attempting to do so will definitely provide some insight
into the issues involved here. If anything comes of it, I'll post a note on the
Thanks for the help.
Freeagent, can you please post the raw file with the blue clipping? (YouSendIt is fine.) I can then experiment and get back to you.
>In cool hindsight, I can see what happened: the 4.4 profiles fixed the problem that I had using the 2.4 profiles. But that's no excuse, and I claim none. So there it is. Shoot.
This is one of the coolest things about digital photography...in the old film days, once you shot and processed the film it was forever locked into a given state (unless it faded or was otherwise destroyed). It really could _NEVER_ get better over time...
That's simply not the case with digital. As long as you keep your raw files (either original or DNG), you can go back and use improved new software to improve over what you could do with previous processors.
Glad you figured out what was going on and that you now have an approach that moving forward, will give you better results than what you had before. Course, the downside is that for archive RGB where you've spent a lot of time post processing the raw files into PSD or TIFFs, the question is, can you live with the old processing or do you have to re-process the images. This also points out the new found power we have with parametric edits such as Lightroom and Camera Raw...you can go a lot further now while remaining in raw form without having to commit to a processed image. Course, that has it's own price as well...ain't nothing ever free!
You are a classy gentleman and a very cool cat. :) Thank you for reporting back.