Which colorspace?

Rhuarc

I'm trying to figure out what is the most common colorspace that people use, and how they use it. I've read everwyhere that Adobe1998 is the better colorspace to work in, since it has a wider color gamut, but then I've heard others say it is only useful to work in that color space if the end device you are going to be using supports that wider color gamut. My problem is that some of my shots I am printing on my Canon 6000D, which I have read supports the wider color gamut, and others Iupload to smugmug which doesn't support it. How does everyone reconcile these issues?

Art Scott

sRGB all the way from shot to upload/print

Try them out..and see the results.....I personally asked all the "PROFESSIONAL" Print houses I could.....as I was very new here and was speculative of the answers here...however I was told the same by all....Your screen displays in sRGB.....Printers Print from sRGB...in the end it is all sRGB unless it needs to be in CMYK for 4 color printing ...unless something has changed in the last couple of years....so I personally shoot and work in sRGB and that keeps it simple for me.

Andy

sRGB all the way from shot to upload/print

http://www.smugmug.com/help/srgb-versus-adobe-rgb-1998

Rhuarc

Hmm, I was under the impression that my IP6000D could print using the Adobe1998 colorspace, but it seems I was wrong. I did some further checking and it seems that only the ip8500D with the addition of the red and green tanks can print in the wider colorspace. So I guess I've been screwing up my prints even from my own printer!!

Oh well, live and learn right? So I guess I'll switch to an sRGB only mode, at least for now. One question though. If I am shooting in RAW I can change the colorspace in ACR and not lose any colors right? What I mean is that I could set my camera to shoot in sRGB, then when I was doing the RAW conversion I could change it to Adobe1998 in ACR and still benefit from the wider gamut. Or is that wider gamut gone since the camera was set to shoot in sRGB?

Thanks for your help everyone!

LuckyBob

Combination of the two (please explain)

Rhuarc wrote:

One question though. If I am shooting in RAW I can change the colorspace in ACR and not lose any colors right? What I mean is that I could set my camera to shoot in sRGB, then when I was doing the RAW conversion I could change it to Adobe1998 in ACR and still benefit from the wider gamut. Or is that wider gamut gone since the camera was set to shoot in sRGB?

Thanks for your help everyone!

If you're shooting in RAW there's no colorspace associated with the RAW image itself, just the preview JPEGs embedded in it. When you process a RAW file, you can "assign" a color space at that point to the resulting file; the upshot to this is if you find that the image is limited by sRGB you can re-process it as AdobeRGB or ProPhoto RGB.

In answer to the original question: I just about always use AdobeRGB unless I have any gradation issues in which case I'll switch to 16 bit (or sometimes 8 bit sRGB since despite the extra 281 trillion colors 16 bit gives, there still seems to be banding in some of my skies). There's been a handful of times I've had super-saturated images that AdobeRGB couldn't reign in and I grabbed ProPhoto RGB - 16 bit since "steps" between 8 bit ProPhoto RGB colors are pretty large jumps - but it's only been on a few sunset images with the sun in the image.

System

adobe rgb all the way through-
I print my own-
when I upload to smugmug, I convert the profile to srgb and save that to web-

Shay Stephens

sRGB all the way from shot to upload/print

I find the lure of adobergb to be mostly conceptual with not many real world benefits over srgb. For most photographers, it's an srgb world. There are of course exceptions, and they know who they are and why they do it. If you don't know any better, use srgb and be happy with a gamut wider than any printer can handle.

LuckyBob

Combination of the two (please explain)

Shay Stephens wrote:

[...]If you don't know any better, use srgb and be happy with a gamut wider than any printer can handle.

Really?! Last time I checked, my Epson 2200 can exceed even AdobeRGB's gamut in places...
http://www.drycreekphoto.com/tools/printer_gamuts/vrml/epson/epson-2200-EpPremLus.htm

All in good humor, Shay

Shay Stephens

sRGB all the way from shot to upload/print

LuckyBob wrote:

Really?! Last time I checked, my Epson 2200 can exceed even AdobeRGB's gamut in places...
http://www.drycreekphoto.com/tools/printer_gamuts/vrml/epson/epson-2200-EpPremLus.htm

All in good humor, Shay

Isn't that like the claim a horse is faster off the starting line than a racecar?

LuckyBob

Combination of the two (please explain)

Shay Stephens wrote:

Isn't that like the claim a horse is faster off the starting line than a racecar?

Still a lot bigger than sRGB, which was my real point.

It goes farther outside the aRGB space as L approaches 100, to boot.

Rhuarc

Well, thanks for the help everyone! I guess at this point I'll be sticking to the sRGB space. I'm having trouble finding a gamut chart for my 6000D, but even if it can print outside sRGB significantly, I don't want to have to process my files twice, once for smugmug, and once to print. I know I could always just go back to the RAW file and change it to Adobe 1998 for the shots I want to print later, but that would mean re postprocessing anything I want to print at home. At this point it isn't worth it to me I don't think.

Unless someone has an easy way to do this, that is!

Mike Lane

FWIW:

I shoot raw, open my images in 16 bit prophoto rgb, edit them w/ Lab most times, convert to 8 bit sRGB at the end of the process.

Rhuarc

Mike Lane wrote:

FWIW:

I shoot raw, open my images in 16 bit prophoto rgb, edit them w/ Lab most times, convert to 8 bit sRGB at the end of the process.

Doesn't this conversion to sRGB change what the image looks like? If there are colors that fit into the profoto space, but not into sRGB, then as soon as you do the conversion the shot would look different. Then wouldn't you have to reprocess to makre sure it looks ok in sRGB?

LuckyBob

Combination of the two (please explain)

Rhuarc wrote:

Doesn't this conversion to sRGB change what the image looks like? If there are colors that fit into the profoto space, but not into sRGB, then as soon as you do the conversion the shot would look different. Then wouldn't you have to reprocess to makre sure it looks ok in sRGB?

This is true, but the process still has two benefits: (a) you can control this conversion process with Photoshop, and (2) there is basically no loss in color accuracy while editing the photo. If you've got a sky with a small amount out-of-sRGB-gamut colors, you can still up the contrast or saturation while editing without having everything go out of gamut and possibly clip. When you need to convert to sRGB for web display, you could specify a "Perceptual" conversion, so while the colors will change a bit, the visible difference between colors will remain. Not to mention, you can preview the change to determine if the shifts imposed are acceptable.

Mike Lane

What LuckyBob said

The raw output is 12 bits per channel, sRGB and aRGB are both 8 bit per channel color spaces. ProPhotoRGB is a 16 bit per channel color space so while I'm editing the image, I get to use all the bits. Then I can convert it how I see fit at the end of the process rather then just letting bits get stripped right off the bat.

LuckyBob

Combination of the two (please explain)

Mike Lane wrote:

The raw output is 12 bits per channel, sRGB and aRGB are both 8 bit per channel color spaces. ProPhotoRGB is a 16 bit per channel color space so while I'm editing the image, I get to use all the bits. Then I can convert it how I see fit at the end of the process rather then just letting bits get stripped right off the bat.

, but not quite . All three colorspaces can be either 8 bit or 16 bit, depending on what you choose when processing the RAW file. sRGB can represent the smallest "range" of colors, while aRGB represents somewhat wider "range" of colors (slightly redder, a lot greener, and somewhat bluer), and ProPhoto RGB can hit the biggest extremes of color, some supposedly outside the approximate visible spectrum.

The 8/16 bit decision just determines how granular the steps between two adjacent colors are. With ProPhoto RGB 8 bit in particular, the "steps" between two adjacent colors are more like "jumps", and very visible banding can occur in skin/sky tones. 16 bit fixes this, but results in larger files and some limitations in Photoshop; I tend to stick with whatever the lowest common denominator that will encompass the necessary range.

Visual comparison from Luminous Landscape:



Edit:

I did some informal testing with a particular image I've had issues with in the past, especially since PhotoPro is supposed to dive really far into blue.

Click the image to download a 750KB PNG of the Photoshop screenshot I took.

The test consists of the same image, processed the exact same way in Photoshop, at 100% crops (yeah, I know there was dust on the sensor ), but with sRGB 8bit, aRGB 8bit, and PhotoPro RGB 8 and 16 bit output from Capture One. The things I note right away:

• sRGB has basically no banding (compare it to the PhotoPro 8bit), but the colors represented aren't anywhere near as vibrant as PhotoPro or AdobeRGB since the photo's right up against the limits of sRGB.
• aRGB is the darkest, and has a slight magenta cast. This is *not* an indication that aRGB is doing badly. This does indicate that for this particular color range, aRGB 8 bit only has "entries" in its color table that lean slightly towards magenta. They're super-saturated though (unlike sRGB), and there's very little banding (unlike PhotoPro).
• PhotoPro 8 bit has some horrible banding issues, but the colors are just about right. PhotoPro covers so much area that it's hard to miss the "correct" colors in its color table (see spectrum image above), but the steps between colors are pretty abrupt, hence the banding.
• PhotoPro 16 bit is just right for this image. The "banding" is basically up against the noise from the camera and the limits of what my (and your) monitor can actually display. Theoretically there's much more gradation in the image, but monitors can't display it since they're only (at best) around 12 bit, with most ~8 bit and some cheap LCDs 6 bit.

It's not a perfect test, but I wanted to confirm that my personal assumptions/knowledge about the colorspaces held true in real life . I hadn't tested the theory before, so I guess now was as good a time as any!

Mike Lane

LuckyBob wrote:

, but not quite ...

I should have said, I bring it into photoshop as a 16 bit prophoto so I can keep as much data as possible. Your tests confirm that is what is happening.

So thanks

semi-related question... You said that theoretically there's much more gradiation than the monitor can show since the monitor is only 12 bit. But unless you took that image with a 1D mkIII or a medium format digital, you've only got 12 bits in that image anyhow. So you're actually seeing the info that the image actually has. True? Or am I missing something?

LuckyBob

Combination of the two (please explain)

Mike Lane wrote:

I should have said, I bring it into photoshop as a 16 bit prophoto so I can keep as much data as possible. Your tests confirm that is what is happening.

So thanks

semi-related question... You said that theoretically there's much more gradiation than the monitor can show since the monitor is only 12 bit. But unless you took that image with a 1D mkIII or a medium format digital, you've only got 12 bits in that image anyhow. So you're actually seeing the info that the image actually has. True? Or am I missing something?

From what I've seen there's only a handfull of 12bit LCD monitors out there. I dunno what CRTs generally are, although they may be truely analog. Even with the RAW file being only 12 bit, ACR / C1 / Etc still do some luminance smoothing for noise, so I'm sure the blending of adjacent pixels (or whatever the process is, stupid IP laws ) creates more colors than the origional RAW, possibly exceeding 12 bits. Granted, at this point it's really splitting hairs, but it seems like sound logic in my head.

Then again I might be crazy

Shay Stephens

sRGB all the way from shot to upload/print

LuckyBob wrote:

Still a lot bigger than sRGB, which was my real point.

It goes farther outside the aRGB space as L approaches 100, to boot.

We are of course still talking about the printers printable gamut and not the theorhetical space adobesrgb provides right

Regardless though, if a photographer knows they need adobergb, then of course they should use it. For those who don't know why or how they should use it, srgb is going to give them less grief to use.

Rhuarc

What is a good way to tell if my printer can even print from the aRGB space? I can't find a gamut chart for it, so I thought maybe if tehre was a test image I could print. Then maybe I could see if it printed the exact same way on my printer set to both aRGB and sRGB then I would know that my printer can't pint further than sRGB. If the aRGB print has brighter colors on the test image than sRGB then I know it can handle it. Any ideas?

Shay Stephens

sRGB all the way from shot to upload/print

Rhuarc wrote:

What is a good way to tell if my printer can even print from the aRGB space? I can't find a gamut chart for it, so I thought maybe if tehre was a test image I could print. Then maybe I could see if it printed the exact same way on my printer set to both aRGB and sRGB then I would know that my printer can't pint further than sRGB. If the aRGB print has brighter colors on the test image than sRGB then I know it can handle it. Any ideas?

I don't believe there is a printer in existense that can print the entire srgb color space let alone anything broader. If your printer does not have color space charts than you can pretty much figure it is like all the other printer out there. Print and be happy and don't worry about chasing color spaces

Rhuarc

Shay Stephens wrote:

I don't believe there is a printer in existense that can print the entire srgb color space let alone anything broader. If your printer does not have color space charts than you can pretty much figure it is like all the other printer out there. Print and be happy and don't worry about chasing color spaces

I know the 8500 by Canon will print the aRGB, at least they say it will in their blurb about it. But I'm starting to agree with you. I'll probably just start processing only in sRGB, and then if I ever need to have a larger gamut for future printers I can alwaysgo back and reprocess the RAW.

Shay Stephens

sRGB all the way from shot to upload/print

Rhuarc wrote:

I know the 8500 by Canon will print the aRGB, at least they say it will in their blurb about it

The whole thing, or just a part of it? That is the problem with most of those claims.

Rhuarc

I don't know, it just said the 8500 takes advantage of the wider color gamut available from aRGB. You may be right, they could say that if it printed one color that wasn't available in sRGB, but was from aRGB.

slapshot

Check out www.drycreekphoto.com. It has comparison of various printers' color spaces against sRGB, Adobe and others, as well as information on profiling your own printer or other devices.

I print at home on an Epson R1800. While it can not print the entire sRGB color space, it does print some colors beyond sRGB that are in AdobeRBG. So, for anything I print at home, I use AdobeRBG. Anything sent out, uploaded, or given to someone else, I use sRGB.

Also, I think unless you are doing ad work/product photography, the differences are meaningless from a practical perspective. I doubt there are many people out there who could look at a print and tell if it was captured/processed/printed in sRGB or aRGB or any other space.

LuckyBob

Combination of the two (please explain)

Shay Stephens wrote:

We are of course still talking about the printers printable gamut and not the theorhetical space adobesrgb provides right

Regardless though, if a photographer knows they need adobergb, then of course they should use it. For those who don't know why or how they should use it, srgb is going to give them less grief to use.

Yes, and

Rhuarc wrote:

What is a good way to tell if my printer can even print from the aRGB space? I can't find a gamut chart for it, so I thought maybe if tehre was a test image I could print. Then maybe I could see if it printed the exact same way on my printer set to both aRGB and sRGB then I would know that my printer can't pint further than sRGB.

Illford provides color profiles for the IP6000D here for their papers (and printer profiles are specific to paper). It's not that you're "setting" your printer to sRGB or aRGB; the printer has its own distinct range of color capabilities all apart from everything else. You're looking for an editing environment (sRGB/aRGB/ProPhoto/etc) which can encompass the range available on the printer/paper combo. Check out the image that Shay posted a few pages back - describes what the printer/paper combo is capable of versus s/aRGB.

Shay Stephens wrote:

I don't believe there is a printer in existense that can print the entire srgb color space let alone anything broader.[...]

I really doubt there is a printer that can entirely encompass the sRGB space (let alone aRGB), but *lots* of printers can exceed sRGB in certain colors. Take a look at the Dry Creek Photo 3D models I posted earlier (and that Slapshot posted after me ). Granted I can't qualitatively compare it to Rhuarc's IP6000D, but the old Canon i950 can pretty radically exceed sRGB's gamut in the green - even poking out past aRGB a bit with Canon Glossy paper. Epson PGPP paper on the i950 puts it out even further past aRGB's green abilities. My Epson 2200 exceeds aRGB's gamut in the yellows, greens, and lighter sky blues on most papers.

Another interesting thing to have a look at on their site is the camera gamuts - some cameras have insanely large gamuts, so processing into aRGB/ProPhoto RGB is almost "necessary" to not limit what the cameras themselves can do. I don't know what their testing methodology was, but it's still interesting. Kind of makes me wonder if all those sRGB JPEG shooters out there are shooting themselves in the foot

Edit: I created a custom plot with sRGB as the solid reference space, a Canon 1Ds MkII as the red wireframe, and ProPhoto RGB as the grey wireframe. Gives you something to think about...

Top down view. Little grey dot near the center of sRGB is L=100

Rhuarc wrote:

I know the 8500 by Canon will print the aRGB, at least they say it will in their blurb about it. But I'm starting to agree with you.[...]

Shay Stephens wrote:

The whole thing, or just a part of it? That is the problem with most of those claims.

I agree with Shay here. It's only going to exceed the gamut of sRGB and possibly aRGB in a handful of areas, and only on certain papers. If you image has none of those colors, there's no point in worrying. For printing, I generally stick with aRGB (for the increased gamut, since my 2200 exceeds sRGB in a lot of areas) unless the image has a bunch of colors in the areas that poke outside aRGB's space; then it's ProPhoto.

Overall, I agree that sRGB alleviates a BUNCH of headaches - it took me a few years to really get the hang of color spaces, even while I was teaching classes involving Photoshop's color reproduction. If it's all new, stick with it for a while to get your bearings until you have one image that just doesn't work quite right in sRGB and it drives you mad, then drives you into one of the other spaces

slapshot

LuckyBob wrote:

Take a look at the Dry Creek Photo 3D models I posted earlier (and that Slapshot posted after me ).

Sorry there, Lucky Bob, I must not have read through the whole thread. I hate to be redundant. Anyhow, must be a case of great minds think alike.

rutt

Eventually we have to output our images, either on paper or on some computer screen. If to paper, then it's a CMYK world and the print will also be viewed under lighting which we might not really control. If to a monitor, then most likely it will be via a browser and, well, sRGB is the only safe option.

In the process, we do get to view the images ourselves on our nice monitors in Photoshop or similar which can handle wider gamuts well. Fine.

What is the gamut of photographic paper? Of the paper used in art books? National Geographic? When was the last time you looked at a National Geographic cover and thought that the shot didn't work because the gamut wasn't wide enough?

What's the gamut of oil paint on canvas? What do you think about Monet's gamut?

Let's try it another way. What's the gamut of the visual reality we live in every day? Can you capture a photograph which exudes heat and burns your eyes the way the sun does? Can you capture deep the deep shadows inside a window at the same time? And how do you think even Irving Pen's roses hold up to the real thing if tested with a spectrometer?

What's my point? The difference between the gamuts of these different colorspaces is trivial compared to the difference between reality and even the gamut of the widest capture we have available. The art comes in using what's available to represent reality in a convincing way, even even though we don't have nearly the same gamut, element of time, ability to focus attention and shift that focus.

Different colorspaces can be useful tools on the way to an end, but they cannot be that end.