There's been some discussion on the POTN forum about colour space, profiles, colour management etc and whether or not to use sRGB or Adobe RGB (I'll refer to it here as aRGB). The forum there is predominantly Canon-based, and I notice that the majority of DSLR users here use Canon also. I don't know enough about other brands like Nikon, Pentax, Sony etc but those users may find this thread useful too.[1]
The advice over there is that if you have to ask the question "What colourspace should I use?", then you should set everything (camera and software) to sRGB. Yes, if you shoot RAW, and you should be, you can change it later.
The reasons for this advice seem to be, in no particular order:
- The Canon manuals suggest this unless you are using printers that are capable of utilising the larger gamut that aRGB provides.
- If you set everything to the one colour space, you don't have to think about it and make it part of your workflow. Less chance of doing it wrong.
- sRGB is more widely used by printers and software. Unless you know that your lab/printer uses aRGB, they'll likely convert to sRGB anyway.
- Most of us will be processing for web display and standard printers only.
Time to point out that the above discussions are for normal "terrestrial" photography. Many of us here are using modified cameras to catch light outside the visible spectrum. I'm yet to find out whether or not this affects any of the recommendations above. I'll report back when I do.
I did a quick search here on IIS, noticed that some are using aRGB for their workflow, and wondered if there is a reason for it beyond the recommendations above.
So, what colour space do you use? And why?
[1] Disclaimer - I am not an expert on colour management. Just thought I'd share what I've taken away from some reading up on it recently.
Troy - if you're preparing your material for commercial print (magazine, book, etc) then you'll also want to think in CMYK space. Separation and going to print from RGB to CMYK is a bit of a minefield at times.
Image 1 below shows you the relative gamuts and what to expect in print (CMKY) if you keep thinking in RBG and then want to print something commercially.
Image 2: RGB (red, green & blue - additive colour)
Troy - if you're preparing your material for commercial print (magazine, book, etc) then you'll also want to think in CMYK space. Separation and going to print from RGB to CMYK is a bit of a minefield at times.
Image 1 below shows you the relative gamuts and what to expect in print (CMKY) if you keep thinking in RBG and then want to print something commercially.
Image 2: RGB (red, green & blue - additive colour)
Thanks for pointing that out. Might do a bit of digging about that too.
Quote:
Originally Posted by Octane
I process in ProPhoto RGB and convert to either Adobe RGB or sRGB as required.
So your camera is set to capture in aRGB? And when you convert back from ProPhotoRGB to the others, don't you get clipped colours because you're going from a much wider gamut to narrower ones?
I had a quick look at the FITS file output from a Nebulosity stack, saved it as 16 bit TIF, and opened it in PS setting working colour space to ProPhoto RGB. After just a little bit of stretching there was a noticeable difference. Think I'll be adopting Humayun's workflow now that I (think I) understand the benefits. The ProPhoto RGB colour space seems to be the closest fit to the colour space that my camera's sensor captures based on some of the reading I've recently done.
Check out this article from Luminous Landscape. See the rollover image showing the ProPhoto RGB compared to 20D sensor and the better coverage in that front corner of the reds? I instantly saw that come out in my image test. If I get a chance I'll post the comparison. Much richer colours, not as washed out.
Putting this colour management together with some other processing tips and tricks I'm learning at the moment and seeing vast improvement. Looking forward now to new data, fully calibrated and colour managed!
Thanks for the heads up on that. You're the only one that recommended it in a couple of places I asked, and I reckon it's the "right" solution, for my gear anyway.
Quick questions. I know you use IRIS for much of your processing. Assume it saves as 16 bit TIF for you to process in PS? Then you just have PS set to use ProPhoto RGB as default workspace/profile and convert everything as you open it? Then when comes time to output and you convert to sRGB (assumed for web) or whatever, how does the converted image compare? No clipping? Does it compress the data down into the sRGB gamut acceptably?
When it opens, I assign ProPhoto RGB or Adobe RGB (depends on what I'm working on, really). Before I save for web (I don't use the Save for Web and Devices tool, except to save my 640x427 images for IceInSpace attachments), I convert to sRGB. So, yes, pretty much as you've described.
I have found that there is very minimal clipping. I reckon the majority of the damage occurs when changing the mode of the image from 16-bit to 8-bit.
Apart from saving the original in ProPhoto RGB, I save all my images that are to be printed, in two formats: Adobe RGB and sRGB. Adobe RGB goes to print, sRGB goes to web.
Neither, it's a colour profile in CS4 Photoshop. Canon cameras only have 2 colour spaces available - sRGB and Adobe RGB. But if you shoot RAW you can assign to anything at RAW conversion stage.
If you have CS4 (not sure about earlier versions), you go to Edit->Colour Settings, Working Spaces and set it to ProPhoto RGB. I have also set the Color Management Policies in the same window to RGB: COnvert to WOrking RGB, and to ask when opening etc to give me the option.
Well, my advice is to shoot in Adobe RGB. It has a far wider colour gamut. That said, no monitor other than very expensive Eizo's can handle the wider colour gamuts anyways. If you have money to burn, then go for it:
An eye one calibrator tool will do nicely as well ;-)
You can always convert from Adobe RGB to sRGB as the final step when going from TIF to JPEG (it's what I do). Going with sRGB is throwing away colours - much like shooting in 8 bit instead of 16 bit.
I find it's better to set the camera to Adobe RGB, that way when you take it into DPP/C1 PRO, etc, it's already set. Saves you forgetting ;-) I export as Adobe RGB to 16 bit TIF files, and from there work on the image in Photoshop. My final act is to save as JPEG, convert to sRGB for web display.
I really wouldn't recommend playing with CMYK - it's a delicate operation and easy to screw up. A lot of printers will accept Adobe RGB and do the conversion to CMYK themselves.
Agree with you Dave. The camera is set to Adobe RGB. Was just clarifying that it didn't really matter (so long as you didn't forget to set it in post!). I have been caught out before.
Yep, the print mob I use requests Adobe RGB files.
Dave, have a read of that LL link I provided above. No argument about Adobe RGB having a wider gamut than sRGB, but check out the comparison with ProPhoto RGB.
I've long since read that link Troy (you're a few years too late lol!). ProPhoto has an even wider colour gamut than Adobe RGB, and currently, no monitor in existence can fully accommodate the entire colour range it provides. It's overkill imho. Adobe RGB is perfectly fine.
It's not overkill if it gets results, and Humayun has more results than you and I combined. I value his experience and advice. I have also done some quick side-by-side comparisons myself of the 2 workspaces and have seen some benefits.
I'm thinking of it like the reason why you edit an uncompressed 16 bit per channel TIF file instead of 8 bits. Trying to keep things as lossless as possible right up until the final point of output, then you convert to 8 bit sRGB or whatever that your output (monitor, printer) can handle.
There is colour/data captured by your sensor beyond the Adobe RGB gamut. We (wannabe) astrophotographers need to value every ounce of data we can get to fall on our sensor - without oversaturating. We're stretching the absolute crap out of the histogram, so subtle little improvements in that initial data make a huge difference to the results.
That's why we go to so much trouble taking lights, darks, flats, biases, calibrating, aligning, stacking - trying to get the best signal to noise ratio possible. Why wouldn't you try to do the same with colour?
), I just find it's easier to set and forget (mostly). Now, what was I saying?
