Colors getting crushed - photoshop advice needed

The back story is I shot in RAW, I used a custom camera profile created with a ColorChecker Passport and Adobe DNG profile editor. In Lightroom the image looks fantastic in ProPhotoRGB.

I use Zenfolio and MPIX, and they both recommend sRGB, so I export as such. I've had some similar problems with oranges before, but this is the first time I can recall having the problem with reds. I'd like to learn more about fixing this.

Here is a straight export from LR as sRGB:

What I suggest is staying in ACR and using your color channels to "tame" the individual colors. You can switch between aRGB, ProPhoto RGB and sRGB to see what is in/out of gamut in each color space. In this case, bringing down the saturation and maybe the luminance in the Red channel should help. You may want to start by lowering the Exposure slider a bit so that your individual colors will keep some balance. Then, if need be, you can up the Brightness a bit which can boost the darker and midtones but keep the highlights in check.

This is common when shooting bright colors -- the camera captures more than will "fit" in the sRGB color space. When you shoot, use the RGB histogram, and to avoid this, you can set the camera to sRGB (and the Picture Style to Neutral or Faithful) and gauge your exposure to keep the colors in range. But in post processing you have to manually tweak the colors into the sRGB color sace to keep them from clipping.

Second set of "softproofing" screenshots, you are softproofing for CMYK.
Not what you want.

BluewookieJim wrote in post #10036714
Even after that my sRGB photo still looks the straight LR export.

No. I see a clear difference: The "straight export" shows more saturated reds, yet less detail there and more clipping.
AdobeRGB image shows less clipping and a bit more saturation. And the clipping that's still visible might be caused by my screen and not be in the file (color managed browser, didn't open in PS)

Thanks, I follow you somewhat, but need to ask for some clarification. I'm much more proficient in Lightroom than Photoshop, so bear with me.

So lets say I'm editing my photo in LR. I would then use the HSL sliders to tame those problem colors. Now, to see what is still out of gamut I would have to open in photoshop (as a smart object) and then do the soft proofing? Am I following that correctly?

I've experienced this before, but mostly with orange colors. I do use the RGB histogram on my 5DM2 (don't have that option on the 20D), but admit I wasn't really paying attention to it when I shot these.

René Damkot wrote in post #10037160
Second set of "softproofing" screenshots, you are softproofing for CMYK.
Not what you want.

Thanks. I must've fat-fingered that when I threw that sample together. I was intending in doing it Adobe RGB..

BluewookieJim wrote in post #10037186
Thanks, I follow you somewhat, but need to ask for some clarification. I'm much more proficient in Lightroom than Photoshop, so bear with me.

So lets say I'm editing my photo in LR. I would then use the HSL sliders to tame those problem colors. Now, to see what is still out of gamut I would have to open in photoshop (as a smart object) and then do the soft proofing? Am I following that correctly?

I've experienced this before, but mostly with orange colors. I do use the RGB histogram on my 5DM2 (don't have that option on the 20D), but admit I wasn't really paying attention to it when I shot these.

Unfortunately, Lightroom doesn't have "color space soft-proofing", so the LR histogram is "stuck" at a wide gamut color space (for now).

So, in these situations, if it's important to keep a color from clipping in sRGB you need to either "wing it" in Lightroom (tamp down the channel using the histogram to just get an "educated guess") or use another tool such as ACR or DPP to get a more precise guess. Since you have ACR, you can open the image through Bridge in ACR, switch to the sRGB color space, and bring the colors into line, then use those adjustments in LR before bringing the image into Photoshop proper, or you can do as you suggest and bring the image into PS as a Smart Object then open it in ACR to do the channel adjustments.

Of course you can do like you show with PS layers, but in ACR for one thing you are working with the Raw file and for another thing "color space soft proofing" is pretty quick, although you can also do this in Photoshop.

Thanks Tony, I'm going to try some of that out at home tonight. Usually when I do anything in Photoshop, the image is sent to Photoshop as a smart object from LR. I just never thought about using ACR from there for more adjustments to the RAW file. Maybe next week when I get my CS5 upgrade it'll be even easier...

Tony, when you are talking about "color space soft proofing" in ACR, do you using the hyperlink below the image to change the workflow option to your desired color space? If so, I think I see what you are saying. It was relatively easy, start in LR, open in photoshop as a smart object, edit the smart object in ACR, hit ok, use the soft proofing in PS, etc...

It seemed pretty painless. I think I was able to get a better result, and quicker too. (LINK 3MB)

Yes, that's what I mean, and great that you could see good results!

It's funny that this issue has seemed to come up lately -- it's rather obscure except for those of us who photograph things with bright colors -- flowers, fabrics and such, and in fact it was relatively recently that I really saw what the "soft proofing" could do to help fine-tune your colors. And, it was a real pleasure to see that available in DPP, not just ACR.

But I wouldn't rule out Lightroom for the adjutstments. It is just being aware that the LR histogram shows only a wide gamut (no soft proof) and knowing about how much you need to back a color off in LR to get it within the sRGB color space. In fact, I do this in LR myself, but it is instructive and visually cool to be able to use DPP and ACR with the "soft proof" capability.

Thanks again. Part of my issue is I'm still getting used to the 5dM2. I've been shooting with the 20D (tiny lcd and all) using the regular histogram for the last few years, and I know almost instinctively how far I can push my exposures on that camera. On the 5dM2, I've only had it for a few months now, and I've only got about 2,200 exposures on it. I'll get more comfortable with it, but I've been pushing the exposures a bit too far more frequently than I'd like so far.

Hey, you're having a good time with your new gear!

Typically, the color space problems aren't the issue, just getting a great exposure, and in most scenes the highlight blinkies and the tone histogram are your friends. But when bright colors are involved, that RGB histogram is your best buddy.

Now, if you are interested in pushing things a bit you could experiment setting your camera in the aRGB color space and, using the RGB histogram, see how far you can push a color (beyond sRGB) and then pulling it back a bit in your Raw processor. The RGB histogram has more "room" in aRGB than it does in sRGB (although the actual camera space is even larger), so you can use a version of "Expose To The Right" to push a color right to the edge, but realize that in the sRGB color space and in your monitor, the color(s) will appear clipped (intentionally) and it is up to you to be aware of this and compensate for this. Your other colors, though, will be given that bit of ETTR boost, which could be very good for your image.

A couple of points. One is that the saturation slider affects the most saturated colors, while the Vibrance slider affects the least saturated colors. In theory, you would want to decrease the saturation slider, not the vibrance.
If you have a choice of rendering intent in Lightroom (I don't have it), I would think that Perceptual might do better than Relative Calorimetric in this case.
Did you try using the recovery slider in Lightroom/ACR?

Oh, one more thing. Zenfolio will accept AdobeRGB. My understanding from past conversations with them is that they will convert the web file to sRGB for viewing, but will send the original AdobeRGB file to the printer for printing. That may have been before they started using MPIX as their lab, so I don't know if it is still true.

A conversion to a working (matrix based) color space can only be Relative Colormetric. The implementation of Perceptual R. I. requires the presence of one or preferably more LUTs, which by definition are device-dependent.