best color space for printing

This is another color management questions, so I expect to get 200 responses.



I have a canon pro9000 printer. Good printer. It can ''recognise'' adobe rgb color gamut.



I export tiff files to qimage for printing. Should I maintain the prophoto rgb color space or should I assign argb? Or does it make a difference?



Andy

''
best color space for printing
I've never used Qimage but one thing in your post caught my attention: I hope by 'assign argb' you actually meant 'convert to argb'.
best color space for printing
Assign or convert, the meaning is clear. When I export a tiff file, I have a choice of 3 color spaces. srgb, argb, or prophoto rgb.



Question is which one gives the best results in printig. Mike Mchaney, who developed Qimage did an article that said that in some instances very large color spaces like prophoto rgb can produce worse results than ARGB. (more banding and posterization). It was a technical article, so I can't explain his reasoning.



Andy

I researched this quite a bit when I first begin using external labs for my prints. I've found that the majority of labs calibrate their printers to the srgb color space. This definitely makes things easier for me because I shoot all my images in the srgb color space. I don't worry with any conversion for development.



However, if you have a 'nice' printer, which you do and care to take the time to produce your own prints - sometimes the larger color spaces can be taken advantage of.



Personally, the less work I have to do, the better for me. I know I'll have consistent colors.

Again, I believe it's one of those things that only the individual can determine, although asking for theoretical advice is a good idea.

For testing, I might recommend a prepared image - one came with my Spyder pro, and I found others on the web a few years back. Bill Atkinson comes to mind.

The journey to printed output is one of always traveling from larger gamuts to smaller ones. The world we photograph has an incredible (infinite?) gamut, our cameras can only capture a part of that, our monitors (most monitors work in the sRGB space) and printers display or print a subset of that. The trick is preserving as much image information as possible as you make your way towards printed output.



Here's a graphical image of various gamuts overlaid on each other that helps illustrate the point: http://the37.com/manray/gamuts.jpg



While your printer ''recognizes'' Adobe RGB, it is very likely that it would convert that working space to the (smaller) gamut that is defined by one of it's built in printer profiles.



John has the best advice, do whatever works for you that produces results that meet your standards.

Well, there's a lot of misinformation available out there...and a lot of the choices regarding what's ''BEST'' depend on the user, their knowledge and experience and the level of quality they want.



It could be said if you want quick and easy without a lot of fuss, shoot in sRGB and output sRGB files. That's what the photo labs want you to do because 1) they end up dealing with a lot of people who don't have a clue what to do and 2) want to lower their customers' expectations and 3) try to avoid dealing with color management.



If you want to go that route, fine and dandy. But there will be trade offs.



Pretty much ALL current DSLRs in raw (as well as most point-N-shoot cameras that shoow raw) can capture a range of colors that even Adobe RGB can not contain.



Saturated yellows, oranges and reds and even browns will get clipped if you capture in or process to Adobe RGB. It's very easy to prove...process a shot of a yellow flower with deep saturated colors into Adobe RGB and Pro Photo RGB (which WON'T clip ANY colors captured by today's cameras). Open the two images side by side and do a slight Saturation bump of say + 12 to each...then look into the deep saturated yellows. See blobs in the Adobe RGB image? That's because deep saturated yellows get clipped and results in uneven saturation increases. There won't be any blobs in the Pro Phot RGB image.



What does that mean?



Well, in order to contain all the colors your camera can capture, you'll need to work in a color space large enough to prevent clipping.



Will that produce better prints?



Possibly.



Todays inkjet printers (recent Epson, HP and Canon pigment printers) can reproduce colors that fall outside of even Adobe RGB (let alone sRGB). So, if you work in Adobe RGB, there are colors that you can capture _AND_ print that the color space clips out. Not so with Pro Photo RGB. Which is why Lightroom is using the Pro Photo RGB chromaticies but in a linear gamma (as apposed to PP RGB's 1.8 gamma).



If what I wrote above gives you a headache, shoot in sRGB and output sRGB files. If you want the best your camera and printer can print, consider using ProPhoto RGB in 16 bit.



And, even if you use and outside source for printing, even Costco has custom printing profiles you can download for sotres around the county...so the next time your pro lab tells you to send them sRGB files, you might ask them why Costco can figgure out how to use profiles...



:~)

Jeff:



VERY nicely put!

Jeff,



Costco doesn't actually provide profiles, they provide a link to drycreek's profiles (which is fine, I'm just saying...).



But help me out here. I read that these Noritsu and Frontier printers that Costco uses (my local one uses a Noritsu 3111) can't produce a gamut much bigger than sRGB using the paper they use (Fuji Crystal Archive). That's why I've never bothered to seek using wider gamuts in my processing. However, if that's not true, I may look into it. Is it true or not?

These drycreek websites seems to indicate that these printer/paper combos can't provide much outside of sRGB, and even not all of sRGB:



http://www.drycreekphoto.com/tools/printer_gamuts/vrml/noritsu/nori-3101-CAGls.h tm

http://www.drycreekphoto.com/tools/printer_gamuts/vrml/fuji/fuji-370-CAGls.htm

I just checked the Costco profile for a store close to me (Costco #381Glossy paper profile, October 5, 2006 - Noritsu model 2901, Fuji Crystal Archive Paper.) and there is an excursion of reproducible color that falls OUTSIDE of sRGB in the Costco profile-particularly deep saturated yellows going both to the green side and the red side of yellow.



The Costco profile DOES fit entirely within the Adobe RGB color space...but you still have the issue of Adobe RGB clipping deep yellows...which can limit your edits.



So, if you shoot in sRGB, you loose colors you _COULD_ print - yellows, and guess what color is the primary color of Caucasian skin?



So much for settling for sRGB huh?



BTW, I use ColorThink from Chromix to view 3D gamut maps for evaluating profiles. http://www.chromix.com

Yeah, the 3101 on gloss has a corner outside on the yellow side too (outside the aRGB space too). I'm sort of amazed at how much of the sRGB space isn't even close to being inside of the printer space, however.



%26gt; So, if you shoot in sRGB, you loose colors you _COULD_ print - yellows, and guess what color is the primary color of Caucasian skin?



So, you're saying it's more interesting to be able to see all the colors you can print than it is to be able to print all the colors you can see. I guess soft proofing is all about seeing *only* the colors you can print, correct?



Any thoughts on the Huey Pro?

Interesting.



According to this:

http://www.drycreekphoto.com/tools/printer_gamuts/gamutmodel.php?Submit=Build+Mo del%26amp;CS=1%26amp;CR=120%26amp;TC=73



Even my 20D can capture colors in the blue direction that prophoto can't represent. It can capture every color the printer can print which isn't surprising.

The article that got me thinking about this is

http://www.steves-digicams.com/techcorner/tc_index.html



It was an article by Mike Mchaney who is the creator of Qimage. He seems to say that since most printers only accept an 8 bit channel and that prophoto rgb is ''overkill'' and in fact can cause some degradation to the final product. Pro photo probably has some value on a printer that accepts a 16 bit channel. I hope I haven't misinterpreted his complicated and technical article.



I agree with Jeff that printing in srgb will show inferior results and should be avoided if you can use the larger gamut argb.



I can't say whether using pro photo rgb will show better printer results though.

''He seems to say that since most printers only accept an 8 bit channel and that prophoto rgb is ''overkill'' and in fact can cause some degradation to the final product. Pro photo probably has some value on a printer that accepts a 16 bit channel.''



Yeah, ok. . .there are some problems with your understanding of what Mike was saying (and his name is ''Chaney'').



First off, you seem to be confusing bit depth with color space, while there is a degree of ''relationship'' between bit depth and color space, they are not directly related nor in the least bit interchangeable.



Let's deal with bit depth...it relates to how many discreet shades (or levels) you can have in an image. 8 Bits/channel (24 bit color made up of red, green %26amp; blue, 8 bit/channel x 3 channels) can have 256 discreet shades of gray. A 16 bit/channel Photoshop file is not actually a full 16 bit, but is actually 15 bits plus one and has 32,769 shades of gray. (it's 15+1 for some math reasons you don't want me to get into-so 16 bit ain't 16 bit). The problem is that pretty much any device like a camera or scanner is only capable of 12 or 14 bit output. DSLRs are putting out 12 bit (4096 shades-although the new 1D MII from Canon is claiming 14 bit). Some medium format camera backs claim 16 bit, but it's really only about 15 REAL bits plus some noise.



So, higher bit depth is useful when you need to stretch out (expand) the spacing of the levels such as when increasing exposure or contrast. With an 8 bit image, you may get to the point where there aren't enough discreet shades to maintain a gradation and you end up with banding.



Cameras all start-even if you shoot jpg-with more than 8 bits. The high bit depth is needed for image processing but here to fore, the main OS's (both Mac %26amp; Win) had print pipelines that precluded using anything beyond 8 bits/channel so the print drivers didn't use more bits. The question is, how do you get to 8 bits?



If you start with a 16 bit image in Photoshop and let Photoshop manage color, Photoshop will do a color transform from the working space (in 16 bit if you are working in 16 bit) to the final printer profile in 20 bit/channel precision (if you use Adobe ACE CMM, ColorSync is limited to 16 bit precision) and drop the final image data sent to the print driver to 8 bits/channel. So, you _DO_ get the usefulness of 16 bit in the final transform from color space to print space.



Only one export plug-in (Canon) and a few rips can actually bypass the print driver and send actual high bit depth to the printer. The Canon is actually only a 10 bit/channel device, not 16 bit.



So, where bit depth comes in is sending the ''PERFECT'' 8 bits/channel to the printer and Photoshop (and Lightroom if you use Lightroom managed color) handles the color transforms. So, you DO get the bennies of high bit depth.



So, what about color space? Well, I've already covered a lot of it. Bigger CAN be better if what you do in the color space is useful-such as preserve colors the camera can capture and the printer can print. It should also be noted that the individual red, green %26amp; blue points in a color space can be tighter (closer together) or looser (further apart). Pro Photo RGB has it's color points wider apart while sRGB has its color points closer. This is where bit depth can play a role...if you take a widely spaced color space (Pro Photo or Adobe RGB) in an 8 bit/channel image and try to expand them, you might introduce banding and a larger color space may be more likely to band than a more tightly spaced color space. Even a 16 bit/channel image can be made to band if you expand the points in the file enough.



So, are larger color spaces useful? Yes..potentially

Are higher bit depth images useful? Yes..potentially

The ideal is a large color space in high bit depth.



As long as you have half a clue what you are doing.



And, if you REALLY want to get your head spinning, try figuring out 32bit/channel HDR...32 bit floating point can have shades whiter than white and blacker than black and currently, there are no ''color spaces'' for HDR-although MSFT is trying with their new HD Photo image format including a new color space called scRGB and is capable of up to 32 bit/channel images. See: http://en.wikipedia.org/wiki/HD_Photo

Jeff, thanks for your explanation. You are right that it is a bit head-spinning though. Andy

Jeff,



So when we're in Lightroom we are using ProPhoto with a linear gamma and then we export to Photoshop we are using ProPhoto with a 1.8 gamma. What kind of hit are we taking for that conversion?



Since you brought up the 1D III and bit depth, how much practical difference is there between the 14bit 1D III file and the 15bit (plus noise) digital back file?



Jim

''What kind of hit are we taking for that conversion? ''



For a relatively minor gamma adjustment very little.



Well, there are other issues at play with ther 1D MIII-seems I remember something about needing 14 bit because the live preview will cause heat build up and more noise so they upped the bit rate. I don't remember exactly...but the result is that going from 12 to 14 bit isn't a total net gain of 2 bits...

Jeff,



''I don't remember exactly...but the result is that going from 12 to 14 bit isn't a total net gain of 2 bits...''



So once we pass 12 bit it becomes increasingly harder to get the full benefit of those extra bits?



Jim

Unless well capacity increases, or a continuous cell reset approach of some sort is implemented, yes. Plus, either approach would require you to shoot at very low ISO (%26lt; 100) to get the benefit.

''So once we pass 12 bit it becomes increasingly harder to get the full benefit of those extra bits? ''



No...the Canon was using (loosing) some of it's bits for something...heat I think, I just can't remember. But, going beyond 14 bits is kinda tough with today's cameras, yes. Although there is a greyscale camera out there that can capture 96 bits of data and has a dynamic range of, well almost everything (actually over 22 stops if I remember correctly). But it's experimental and no current camera is in that ballpark-that we know about-the CIA or NSA may have soemthing though.



;~)