Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: Brad Proctor on December 15, 2009, 02:51:04 pm
-
This seems like it would be a pretty common question, but I couldn't seem to find an answer. After working on a RAW image in Lightroom I export to a 16-bit TIFF in ProPhotoRGB and then load it into Photoshop to do additional work. I don't print my own images and my printing company requires JPEG images. I usually just covert the image to 8-bit and then save a level 12 JPEG while still being in ProPhotoRGB. The printer doesn't care what color space I use as long as I have it attached to the file. My question is, because I have to convert to an 8bit JPEG, should I be converting the color space to something else like AdobeRGB before converting to 8bits?
-
The processing order is important here. It's best to make all you image editing adjustments in 16 bit ProPhoto to avoid risks of banding and posterization which could arise from reworking a relatively small number of luminosity levels in a huge colour space. Once you have done all your editing, flatten the image, convert it to 8-bit and save it as a high quality JPEG. You should be OK. If you make a colour space conversion from ProPhoto to ARGB(98), do that before you flatten and before you convert to 8-bit. You would need to run a few tests with your printer to see whether colour space conversion is either necessary or causing you to lose any hues that are visible in ProPhoto alone.
-
Hi,
Mark's suggestions seem reasonable to me. JPEG has eight bit resolution and it's a bit to short to hold ProPhoto RGB. It may work or it may not. If you convert to sRGB you may miss some colors which could have been printed, but quite a few possible issues could be avoided. So sRGB is the safest choice. You may also ask your printer for his color profile. If his color gamut fits inside sRGB there may be little benefit in sending files in a larger gamut.
If you are on the Mac the ColorSync Utility can show and compare color gamuts. On the PC I used to use Gamut Vision. ( http://www.gamutvision.com (http://www.gamutvision.com) ).
Best regards
Erik
The processing order is important here. It's best to make all you image editing adjustments in 16 bit ProPhoto to avoid risks of banding and posterization which could arise from reworking a relatively small number of luminosity levels in a huge colour space. Once you have done all your editing, flatten the image, convert it to 8-bit and save it as a high quality JPEG. You should be OK. If you make a colour space conversion from ProPhoto to ARGB(98), do that before you flatten and before you convert to 8-bit. You would need to run a few tests with your printer to see whether colour space conversion is either necessary or causing you to lose any hues that are visible in ProPhoto alone.
-
Hi,
Mark's suggestions seem reasonable to me. JPEG has eight bit resolution and it's a bit to short to hold ProPhoto RGB. It may work or it may not. If you convert to sRGB you may miss some colors which could have been printed, but quite a few possible issues could be avoided. So sRGB is the safest choice. You may also ask your printer for his color profile. If his color gamut fits inside sRGB there may be little benefit in sending files in a larger gamut.
If you are on the Mac the ColorSync Utility can show and compare color gamuts. On the PC I used to use Gamut Vision. ( http://www.gamutvision.com (http://www.gamutvision.com) ).
Best regards
Erik
Thanks Erik.
I think the key distinction which needs to be made here is what happens during image editing versus what happens afterward. There's a near-consensus I believe that converting to 8-bit and then doing image editing in a very wide colour space could be asking for trouble. I think it is much less risky converting to 8 bit and remaining in the wide space AFTER all the editing is completed, but whether the wide space is then necessary or worthwhile becomes the empirical question to test with the printer - and I agree with you - soft-proofing with the printer's profile could be indicative- but not fully determinative because most of our displays do not really "see" all the colours which a number of contemporary printers can reproduce.
-
You would need to run a few tests with your printer to see whether colour space conversion is either necessary or causing you to lose any hues that are visible in ProPhoto alone.
Mark, I don't understand this. I don't know if there is any printer capable of printing ProPhoto directly, but if there is, then that is the very exception. Thus the color space has to be converted, either by the user or by the printing service - but then why would the user/photographer leave that so, instead of doing the conversion himself and verifying the result immediately?
-
Mark, I don't understand this. I don't know if there is any printer capable of printing ProPhoto directly, but if there is, then that is the very exception. Thus the color space has to be converted, either by the user or by the printing service - but then why would the user/photographer leave that so, instead of doing the conversion himself and verifying the result immediately?
I have been routinely sending (thousands of) images in ProPhoto colour space to my Epson 3800 and before that my Epson 4800 and before that my Epson 4000. As the printers advance, the size of the colour gamut they can reproduce expands. The 3800 can print some colours exceeding the Adobe RGB98 space and all the more so for the newer Epson 3880 and Epson 7900/9900, so there is nothing exotic about this procedure. It's just a matter of capturing all the colours the printer is capable of reproducing. I don't know exactly what printer or printing process is in play for this gentleman because he sends the work out, why I suggested it just needs a bit of experimentation.
-
I think the key distinction which needs to be made here is what happens during image editing versus what happens afterward. There's a near-consensus I believe that converting to 8-bit and then doing image editing in a very wide colour space could be asking for trouble. I think it is much less risky converting to 8 bit and remaining in the wide space AFTER all the editing is completed, but whether the wide space is then necessary or worthwhile becomes the empirical question to test with the printer - and I agree with you - soft-proofing with the printer's profile could be indicative- but not fully determinative because most of our displays do not really "see" all the colours which a number of contemporary printers can reproduce.
I agree with Mark and Eric, but before converting to sRGB (if you choose that route), I would make certain that no significant colors are being compromised. sRGB is a reasonable fit to most printers using photographic paper, but must high end inkjet printers considerably exceed sRGB and even Adobe RGB. Why don't you ask your printer for a profile so you can convert directly to that profile with soft proofing and the proper rendering intent. That way only one profile conversion would be needed. Converting to 8 bit ProPhoto is not advisable because of possible banding.
If you have already applied the printer profile and have sized your print at the proper resolution, some printers will give you a discount as all they have to do is sent the file to the printer.
Bill
-
Converting to 8 bit ProPhoto is not advisable because of possible banding.
Bill
Bill, wouldn't you agree that most of this risk occurs DURING the image editing process, rather than afterward? I've tried to reproduce banding after completion of all editing by converting ProPhoto images to 8 bit before changing the colour space (rather than the other way around) and quite frankly I've very seldomly seen it.
-
I've tried to reproduce banding after completion of all editing by converting ProPhoto images to 8 bit before changing the colour space (rather than the other way around) and quite frankly I've very seldomly seen it.
But in the end there's another conversion from ProPhoto to output color space, so maybe it may make things worse then?
-
8-bit ProPhoto is not wise practice. You can get away with doing it sometimes without causing banding, but that doesn't mean it's a good idea to make your standard practice. If you need something wider than sRGB for your JPEG, I'd use Adobe RGB 1998, but nothing wider-gamut than that.
-
This is safe advice.
-
But in the end there's another conversion from ProPhoto to output color space, so maybe it may make things worse then?
Not that I've actually seen.
-
Thanks for everyone's advice. I put my printers profile into ColorThink and compared it.
Here are a couple of images showing the printer profile compared to AdobeRGB (in red). There is a little bit of color that the printer can do that is outside of AdobeRGB but it is not bad.
(http://proctorphotos.com/wp-content/uploads/2009/12/color1.jpg)
(http://proctorphotos.com/wp-content/uploads/2009/12/color2.jpg)
Here is the printer profile compared to sRGB(wireframe). Huge amount of color that the printer can do that is outside of sRGB.
(http://proctorphotos.com/wp-content/uploads/2009/12/color3.jpg)
(http://proctorphotos.com/wp-content/uploads/2009/12/color4.jpg)
-
Very nice demos Bradley, and shows that ARGB(98) should be safe enough for you. These results are not surprising - BTW what printer and paper are we talking about?
-
This seems like it would be a pretty common question, but I couldn't seem to find an answer. After working on a RAW image in Lightroom I export to a 16-bit TIFF in ProPhotoRGB and then load it into Photoshop to do additional work. I don't print my own images and my printing company requires JPEG images. I usually just covert the image to 8-bit and then save a level 12 JPEG while still being in ProPhotoRGB. The printer doesn't care what color space I use as long as I have it attached to the file. My question is, because I have to convert to an 8bit JPEG, should I be converting the color space to something else like AdobeRGB before converting to 8bits?
What type of prints ... who is your printing company? If you are using a company outputting to silver halide photo paper on printers like Durst and Noritsu, they will most likely push it to sRGB if you don't. These printers are designed around the assumption of an sRGB file coming in. Most likely you would be better off converting to sRGB yourself earlier in the process than having them do it right before printing.
-
Very nice demos Bradley, and shows that ARGB(98) should be safe enough for you. These results are not surprising - BTW what printer and paper are we talking about?
I use White House Custom Colour. They are photographic printers rather than inkjet. They print on Kodak Professional Supra Endura VC Digital Paper and use
Noritsu printers for prints 12x18 and smaller which is what I usually order and they use Durst Theta 76 printers for larger prints.
-
What type of prints ... who is your printing company? If you are using a company outputting to silver halide photo paper on printers like Durst and Noritsu, they will most likely push it to sRGB if you don't. These printers are designed around the assumption of an sRGB file coming in. Most likely you would be better off converting to sRGB yourself earlier in the process than having them do it right before printing.
WHCC is my printing company. And they are using Noritsu printers for their smaller prints which is what I typically order. Thanks for the tip. I might have to just start doing that.
-
Wayne's point is a good one insofar as it gives YOU control over the process - especially if you have their profile for the Noritsu and the paper. (I'm assuming here that this process uses ICC profiles - I'm not familiar with it.) Wayne?
-
Noritsu silver-halide printers are using Noritsu's own color space, which is somewhat smaller than sRGB. One should use sRGB, or ask the printing service for their own profile and use that. Once I contacted a Noritsu guy (a regional sales manager) and he wrote me, that they do not give out the profile, because that depends on the paper and chemistry at the printer. Some, like Costco, do provide the profiles. This gentleman wrote me, that that profile should be APPLIED to the print file, do not CONVERT the print file to the profile. I still don't understand it, but the location I used did not have profile anyway.
-
Wayne's point is a good one insofar as it gives YOU control over the process - especially if you have their profile for the Noritsu and the paper. (I'm assuming here that this process uses ICC profiles - I'm not familiar with it.) Wayne?
WHCC may be using profiles (it sounds like they are), but not in the conventional manner as we think of using profiles. The printers are designed around an sRGB model. It sounds like what they are doing is using profiles to adjust the files to compensate for the variations in the printers (a real problem with these type of printers).
WHCC operates on the premise you should adjust your monitor to match their output, and they will adjust their processes to maintain a consistent output across all devices. This is why you have to submit 5 files to be printed ... then you adjust your monitor to match the resulting prints. Sending them 8bit AdobeRGB or sRGB files should yield good results if you are willing to tweak your setup accordingly.
In my previous company we operated 180 locations with Noritsu processors. We controlled the entire process so our method was to tweak the files at capture for the output device. Works great until they send the file to the central lab for different product, where we had to tweak the files for that printer. We didn't use profiles to do this, but certainly could have done it that way.
-
Thanks for everyone's advice. I put my printers profile into ColorThink and compared it.
I'm pretty surprised at your results. Where did you get the printers profiles? I've never seen a comparison like this where silver halide could exceed sRGB to this degree. Is this the "soft proof" profile WHCC refers to?
this is more typical of a Noritsu profile (solid) against sRGB (wireframe).
[attachment=18696:Noritsu_Profile.jpg]
-
I'm pretty surprised at your results. Where did you get the printers profiles? I've never seen a comparison like this where silver halide could exceed sRGB to this degree. Is this the "soft proof" profile WHCC refers to?
this is more typical of a Noritsu profile (solid) against sRGB (wireframe).
[attachment=18696:Noritsu_Profile.jpg]
Noritsu treats everything as sRGB by default, so it cuts it's own gamut with sRGB. But maybe it's possible to turn off the sRGB-centric mode, so you can utilize the whole potential of the paper. I've profiled Koda Endura printed on KIS DKS 1750 minilab, where you could feed the machine with pure, untagged RGB, and the gamut exceeded sRGB in the same way as in the above shown example.
-
WHCC may be using profiles (it sounds like they are), but not in the conventional manner as we think of using profiles. The printers are designed around an sRGB model. It sounds like what they are doing is using profiles to adjust the files to compensate for the variations in the printers (a real problem with these type of printers).
WHCC operates on the premise you should adjust your monitor to match their output, and they will adjust their processes to maintain a consistent output across all devices. This is why you have to submit 5 files to be printed ... then you adjust your monitor to match the resulting prints. Sending them 8bit AdobeRGB or sRGB files should yield good results if you are willing to tweak your setup accordingly.
In my previous company we operated 180 locations with Noritsu processors. We controlled the entire process so our method was to tweak the files at capture for the output device. Works great until they send the file to the central lab for different product, where we had to tweak the files for that printer. We didn't use profiles to do this, but certainly could have done it that way.
Interesting account of how that system works. It all sounds like the basic premises of colour management stood on its head.
-
It's completely bass-ackwards, which is why I use other vendors that actually understand color management if I need something printed online.
-
The printers are designed around an sRGB model.
Considering there is no such thing as an sRGB printer... I think what you’re referring to is the silly half assed “color management” workflow for labs (not end users) where you funnel everything into sRGB and send to the lab, after which they do convert the data (they have to) to their printer’s native color space which isn’t sRGB to make it easier for them to crank out prints without a logical front end that handles multiple working spaces customers may wish to use or a true output profile customers could receive to convert the data and post edit as they desire.
The “send us sRGB” workflow was great in 1993, its pretty stupid, at least for those customers working with modern color management aware products today.
WHCC operates on the premise you should adjust your monitor to match their output...
Wow, like using Adobe Gamma. Kludge. Now thanks to WHCC, you’ve hosed your display calibration for all other software and output needs. So this is a pro lab or simply Walmart? Sounds like the later (which is great for people like my mom who don’t even have Adobe Elements and wouldn’t know an ICC profile from a pickle).
...and they will adjust their processes to maintain a consistent output across all devices.
IF all the devices really are calibrated and consistent, why would they be editing your files (do you want them to edit your files)? Point is, I bet dollars to donuts that if you sent out a series of color patches to all the devices, over the course of a week, the average deltaE would be fascinating to see. You suppose the average of sat 988 patches is below a deltaE 2000 of say 4? I wonder if WHCC would allow someone like me to test this.
This is why you have to submit 5 files to be printed ... then you adjust your monitor to match the resulting prints.
Ridiculous really. You adjust your monitor (apparently an LCD what has a single control over its physical behavior) how?
-
I'm pretty surprised at your results. Where did you get the printers profiles? I've never seen a comparison like this where silver halide could exceed sRGB to this degree. Is this the "soft proof" profile WHCC refers to?
this is more typical of a Noritsu profile (solid) against sRGB (wireframe).
Just for comparison, here is the profile for my local Costco that uses the Noritsu 34 Pro and Fuji Crystal Archive paper. The gamut of the printer is considerably smaller than that of sRGB (first image below) and exceeds sRGB in only a small region in the higher luminance yellows (top view below).
[attachment=18698:sRGB_Noritsu_side.png] [attachment=18699:sRGBvsNoritsuTop.png]
These 3D gamut plots are from GamutVision. My version of Colorthink does not work with 64 bit Windows 7 This gamut is relatively small compared to that obtained with current inkjet photo printers. The next consideration would be to examine the gamut of the images to be reproduced and compare this to the gamut of the printer. Portraits and other images that do not contain a lot of saturated colors would be fine for the Noritsu, but for shots of flowers with saturated colors, one might consider selecting a printer with a wider gamut.
Converting the ProPhoto images to sRGB and sending them to the printer would be relatively safe, but one would not have control over the rendering intent, which is always colorimetric with matrix spaces such as ProPhoto and sRGB. If one were to use the profile for the printer, perceptual rendering and soft proofing would be available.
-
Interesting account of how that system works. It all sounds like the basic premises of colour management stood on its head.
Yes, indeed. That arrangement is actually best for those who do not use color management but merely shoot sRGB JPEGs in their P&S cameras and then send them to the printer. For more sophisticated users, I think that the profile for the printer should be used.
-
I'm pretty surprised at your results. Where did you get the printers profiles? I've never seen a comparison like this where silver halide could exceed sRGB to this degree. Is this the "soft proof" profile WHCC refers to?
this is more typical of a Noritsu profile (solid) against sRGB (wireframe).
[attachment=18696:Noritsu_Profile.jpg]
Yes, I used the profile that they provide for soft proofing.
-
It's completely bass-ackwards, which is why I use other vendors that actually understand color management if I need something printed online.
I'd love to do my own inkjet printing but can't justify the cost with as few prints as I make (couldn't afford it if I wanted to).
Is there an online service you could recommend?
-
Is there an online service you could recommend?
Been awhile since I’ve had anything printed (I do my own) but check out www.pictopia.com
-
The gamma plots are very interesting. Does anyone have similar gamma plots for the Durst Theta 76 printers?
-
You know Bradley - as long as you have ColorThink, you can make this determination on an image-by-image basis by bringing your actual image file into the 3D grapher along with your profiles. The actual pixels of the image will be graphed right with the profile gamuts and you can see if your images need to use ProPhoto as a working space or if you can get away with AdobeRGB. Oh, and if you're using ColorThink Pro, don't forget to downsample the image before bringing it into the grapher. (It can take quite a while to graph all those pixels otherwise.)
-
Interesting account of how that system works. It all sounds like the basic premises of colour management stood on its head.
When this technology was developed, color management as we use it wasn't nearly as common place, especially in the photographic industry at that time. There was no effort to use color management systems.
The “send us sRGB” workflow was great in 1993, its pretty stupid, at least for those customers working with modern color management aware products today.
Wow, like using Adobe Gamma. Kludge. Now thanks to WHCC, you’ve hosed your display calibration for all other software and output needs. So this is a pro lab or simply Walmart? Sounds like the later (which is great for people like my mom who don’t even have Adobe Elements and wouldn’t know an ICC profile from a pickle).
Actually until recently almost all images submitted to labs like WHCC, mPix, Millers, etc. were in camera jpeg files. Even today, most of the photographers doing portrait and wedding work have no real clue about color management and don't have calibrated monitors. If you are using a calibrated monitor, you most likely will get a pretty decent match.
This is changing, with more photographers becoming aware of color management tools because of the ongoing education from people such as yourself as well as the wide adoption of LR>
I'd love to do my own inkjet printing but can't justify the cost with as few prints as I make (couldn't afford it if I wanted to).
Is there an online service you could recommend?
Actually, WHCC is one of the most progressive labs in this regards. Understand that the technology behind these printers is well over a decade old. They were designed when shooting raw was pretty much unheard of. When they were developed virtually all of the images printed on them were from scanned film, not from digital files. It was a process film, scan film, print scanned film model. If you wanted reprints you brought in the negative, they scanned this. It was virtually a closed loop system. In 1999, Noritsu sent a team of engineers to one of our digital studios in S. Cal. where we met with them and got them to design a simple hot folder front end to one of their processors, until then there wasn't a good way to even submit digital files from a camera and you couldn't buy the printer without the scanner front end.
At this point the good labs are beginning to make changes to incorporate color management tools into their workflows. I believe for the most part this is from third party developers, not from the manufacturers. WHCC is one of those labs.
Also, I believe you'll find if you have a well calibrated device, your resulting prints will be satisfactory. A friend here at work just received his 5 sample 8x10's from both mPix and WHCC, mostly portrait work, and we felt both labs prints were decent matches to his display and quite acceptable. Later today I'm going to have him download the soft proof profile and we'll take a look at that ... has me curious. WHCC assumes most of their customers don't have a calibrated display, thus the recommendation to match the display to the prints, but this doesn't mean the prints won't be a decent match to a calibrated display.