Gamut help
Pindy
Registered Users Posts: 1,089 Major grins
Can anybody point me to a really good site or article (or thread) on why to use Adobe RGB or ProPhoto? I use sRGB and a gamma of 2.2 which suits me well on photosharing sites, but there is much positive opinion about the wider gamut spaces and I'm looking for an education.
Thanks,
DP
Thanks,
DP
0
Comments
The basic workflow is to convert from scanner or camera device, input RGB to a synthetic RGB working space space that is at least slightly larger than your input, so as to preserve gamut.
But first, you should settle your imaging philosophy. Some use wider gamut spaces for 'insurance', even if their output is limited to say sRGB, at least their master file has the potential for better output on a different device in the future. Those more concerned abou the here and now or one off use may prefer to work in smaller gamut spaces.
If you work in a small (sRGB) to medium gamut space (Adobe RGB 1998 [which I avoid] or ECI-RGB [my preference over using A98] ), then you can work with an 8 bpc or 16 bpc workflow. With sRGB, the image gamut is close to the viewing gamut, so not many surprises (make sure you softproof to the final device). With A98 or ECI, perceptual rendering softproofing is more critical for highly saturated detail, as this may be outside of the monitor gamut.
If you work in wide gamut spaces like ProPhoto, you are going to be in a 16 bpc workflow. It will be critical to perceptually render softproof to output space while editing to see what is taking place (as even more of the image may be out of the monitors gamut making edits a visual guessing game).
You will be given much theory, but at the end of the day only you can do the tests with your images and workflow and make the decision on which one wins and which one matches your imaging philosophy.
Beware, such topics can become a 'religion' for some people.
If I find any links to websites/articles I will post them.
Please inspect the raw camera image conversions below. This image is not so typical. It has detail is highly saturated yellow/orange areas - this type of saturated hue really needs a wider gamut otherwise clipping often occurs and detail is lost. The image is in sRGB for web display. Originally the image on the left was sRGB and the one on the right was ProPhoto RGB. Each image was perceptually rendered from RGB to the same output profile (not all profiles offer perceptual transforms, which is a critical thing for converting a large gamut into a smaller one). Which image at the small size do you prefer, the left one or the right? Not much detail can be seen at the small size.
[to be continued, as I can only seem to be able to attach one photo per post]...
Hope this helps,
Stephen Marsh.
http://members.ozemail.com.au/~binaryfx/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
http://www.outbackphoto.com/workshop/NEF_conversion/neffile1.zip (3.5mb)
http://www.outbackphoto.com/workshop/NEF_conversion/nefconversion.html
sRGB on the left, ProPhoto on the right (both perceptually compressed into the same output device space before being converted to sRGB for web display).
Now that you can see the detail, which do you prefer? Has your preference changed? Does knowing which image is sRGB and which is ProPhoto skew/bias your preference against your expectations? This is just monitor inspection, what about output to inkjet or press?
Note: Working in ProPhoto RGB and converting to a smaller gamut space using Relative Colorimetric rendering intent will clip the out of gamut detail. So if you work in a wider gamut RGB working/editing space and your images are destined for sRGB - you may not see any benefit (as below) without extra work to manually build in the detail lost by not having access to a perceptual rendering intent (ProPhoto RGB > A98 RGB > sRGB, larger space to small may introduce clipping, as these three profiles only have relcol intent and do not offer perceptual intent for colour transforms).
This is just a single image, a real world example, but not presented to make or break either side (smaller vs. wide gamut RGB editing spaces). Other images can be found to support the case for each workflow. Your clients and your opinion matter the most, as does your workflow and other variables.
Good luck!
Stephen Marsh.
http://members.ozemail.com.au/~binaryfx/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
http://www.adobe.com/digitalimag/pdfs/phscs2ip_colspace.pdf
The gamma of a working space is totally separate from the gamma in other areas of imaging (the display). 2.2 is said to be more perceptually uniform, discussed in the article above. For press work (ink on paper), there's a slight advantage to using a 1.8 gamma in a working space, hence the reason we see this in some spaces, notably ColorMatch RGB.
ColorMatch uses 1.8 because there is less quatization on the way
to CMYK according to Karl Lang, the man who designed this space for the Radius Pressview. The eye is closer to 2.2 (luminance response) but presses have dot gain. Using a source space that is a little lighter reduces the quantization when you correct for press gain. (few people know that
Xerox PARC and Apple used 1.8 as a source space because of the
natural dot gain of toner based laser printers.)
Author "Color Management for Photographers"
http://www.digitaldog.net/
http://dgrin.com/showthread.php?t=27460
ALL conversions from ProPhoto RGB to a smaller space will use the Colorimetric intent, that's the only table (Colorimetric) in a simple matrix profile. There's always going to be clipping, no big deal if you need to post an image to the web or use a working space of limited gamut due to that output need.
There are three areas where you need to look with respect to color gamut:
1. The gamut of the capture device. In reality, digital cameras and scanners don't have a gamut (they have what is called a color mixing function). None the less there is a gamut to film (which limits what the scanner produces). There's a real world of color gamut we see and it's pretty huge. Anytime you see one of those horseshoe plots of color (CIE chromaticity diagram), you're seeing the gamut of human vision. A digital camera can capture a huge range of color. All you need to do is take a RAW file of a very colorful scene in Adobe Camera RAW and examine it's Histogram for color clipping. I have tons of images that fall far outside Adobe RGB gamut. IF (big if) I want to contain the colors I shot, I need to use a much larger gamut like ProPhoto RGB (in 16-bit).
When you set your camera to sRGB or Adobe RGB (1998) and throw away the RAW file, you have no control over this. However, there's no question that selecting Adobe RGB will provide a larger gamut of captured colors compared to sRGB assuming the scene has a larger gamut than sRGB. That's pretty likely. Is capturing those important to you?
2. The gamut of the output device. You've decided that you do want to contain colors in the scene your capture device has collected. Now there's the gamut of the output device. Anyone that tells you "no output device exceeds sRGB" is smoking some very bad stuff or simply doesn't know what they are talking about. Again, by viewing a gamut plot (in 3D preferably), you can find all kinds of output devices that have colors that it can reproduce outside of sRGB. Are those color important to you? As I mentioned, the new K3 inkset from Epson produces colors that fall outside of Adobe RGB (1998)!
3. There's the gamut of your display. 98% of users have a gamut that's no larger than sRGB. There are new displays on the market that exceed Adobe RGB (1998) but at a huge cost (today). So there ARE colors you can capture and output you can't see on your sRGB display. The question becomes, do you throw away those colors you captured AND CAN reproduce away because you can't see them all? I think not but maybe some do.
This is all about flexibility and future output needs. IF you know that the scene gamut/camera can't exceed sRGB AND you know your printer can't produce anything larger than sRGB (get a new printer ), then sure, stick with sRGB. But if you look at just Epson and the evolution of increasing gamuts in ink technology, why paint yourself into a corner???
Those silly labs that tell you "just send is sRGB and all is well" are either lazy, don't understand color management or simply don't want YOU to decide how to handle YOUR files! There are NO printers on the planet that produce sRGB!!! The only sRGB device is a CRT display in a very fixed and defined environment. In fact sRGB is a totally synthetic color space designed using pure math (as are all the other RGB working spaces). These labs just don't want to profile their devices so you can soft proof the image, they don't want to worry about embedded profiles and want their automatic systems to simply assume all files are sRGB. Then they do a conversion to the printer color space on the fly. It makes them very productive. However, you lose a lot of control over how you render an image for an output device.
Author "Color Management for Photographers"
http://www.digitaldog.net/
For flexibility, you just need to save your originals or some lossless equivilent (like a .dng). The future may make it possible for you to get better output results from your pictures. When that time comes, why not simply re-bake the pictures. The technology is improving by leaps and bounds. People who took the trouble of preserving everything in earlier versions of ACR would probably start from scratch and use the new and improved ACR 4.x tools anyway. In my opinion, going out of your way to achieve flexibility is just buying insurance that you will probably never need or use. If the need comes in the future to redo a picture for a wider output gamut, just do it again. The tools are getting better and faster all the time. Hopefully, your knowledge is increasing as well.
Also, using a wide gamut is not as cost free as, say, editing in 16-bit. For one thing, when you edit with a wide gamut, you will not be able to see what you are doing. If you don't have the right output devices, you might not ever be able to see what your image looks like. Furthermore, even though Photoshop allows 16-bit editing in its calculations, the controls still work on a 0-255 scale. That means a single point move in a wide gamut space can be a pretty big move. If you are doing highly critical work, this can lead to some issues. The narrower spaces make it easier to make very subtle shifts.
Duffy
Let's just say—for the sake of argument, and that I had obtained a wide-gamut monitor and was outputting to some new device that could happily approximate ProPhoto RGB spaces—what steps do I have to take (on a Mac, mind you) to have everybody talking to each other is ProPhoto RGB (or any profile for that matter) from soup to nuts?
In other words:
1. I'm using Aperture. It doesn't seem to have a default space that you can adjust, just for export. Would it pay attention to my display's profile, if I set it for ProPhoto? How can it be said that I am "working in ProPhoto RGB within Aperture"?
2. The Display profile. Would a Huey (or something better) sort out the display with respect to ProPhoto? Do you tell a calibrator's software which space you want to use? I assume it makes actual monitor display calibrations on top of this preference.
3. I print (currently) to Smugmug's EZPrints service who have an ICC file to soft proof. Since I can softproof in Aperture, is that acceptable? Since the EZPrints service receives my photos as 8-bit sRGB through Smug, I'm not sure exactly what to expect here.
Thanks for tolerating a neophyte—I'm keen to understand though. Reading this back it's not the most lucid thing I've ever written. I hope it comes across.
This is CR2, converted in Aperture and exported sRGB to Smug. Nothing apart from some mild sharpening.
To my eyes, on 2 different monitors (Dell 24" LCD and MacBook Pro), the reds are intense beyond imagination and are burning a hole in my retinas. I feel like there is detail lost in the intricacies of the flower, but the histogram shows no sign of over or underexposure. There is nearly no info at the top; it's a big mountain in the low mid-tones and tapers to barely anything by the time you get to the upper mid-tones. There is no saturation.
I have noticed that de-saturating it reveals some detail, which got me thinking about this in the first place. This is de-saturated (actually the luminance in the red was lowered):
Perhaps my own inadequate knowledge is at work here.
Any thoughts? To be honest, I've had trouble with red flowers ever since I started caring about this stuff 9 monts ago.
The weakest link here is the display, although that's changing (slowly).
Author "Color Management for Photographers"
http://www.digitaldog.net/
next stop... monitor calibration.