There is no CS bug.
And it’s more like just a PS 7 omission rather than a bug (not reading some of the later EXIF additions).
Minolta uses the EXIF data rather than embedding a profile – you have to manually assign the Adobe RGB 1998 profile (if that is the correct profile for the image). That hasn’t changed.
But CS recognizes more EXIF tags than PS 7 did.
Thank you, however I ‘ve realised this and do manually assign the profile in PS. The colour is still way off I’m an experienced PS user and the only thing that’s different is the camera. any other ideas? This is what I’ve already checked:
1. Shooting in sRGB with the Minolta ie the same setting as the Nikon, then
a) converting to working space and printing
b) leaving it with the embedded sRGB tag and printing
2. Shooting in Adobe 1998 and printing.
In both cases, processing in Photoshop. All produce a pronounced colour shift.
3. Processing the photos in Minolta DiMage viewer: colour fidelty is perfect.
4. Putting the finished files onto CD and getting prints made at a local camera shop: colour perfect.
In summary: the screen photo looks good, the print preview from the printer driver (in Acrobat) doesn’t, nor does the actual print. With the Nikon, all were a good match.
So is theresomething about Photoshop, in the file to print process, that does not seem to be compatible with the A1? Or am I still missing something?
If you have the same image in PS 7 and CS, and assign the same profile – then the color will be the same.
You need to break this down into parts – is it the shooting and assigning that’s wrong, or the printing?
If the image on screen looks good, then it’s the printing that’s wrong – and that has nothing to do with the camera. (and why are you going through Acrobat to print?)
See <
http://www.computer-darkroom.com/> for how to print correctly.
The Canon print preview uses Acrobat.
I convert all images to the working profile (Adobe RGB 1998). I agree that regardless of what camera I’m using, provided I’m using the source space as the document space (working space as Adobe RGB 1998) and the correct printer profile then it should match. But it does with photos from the Nikon and doesn’t with photos from the Minolta. (Colour management on the printer is turned off so I’m not doubling up there.)
That means that you are assigning the wrong profile to the images from the Minolta.
After exhaustive tests I can say that Paul Lima is quite correct in his assessment of the A1 colour problems in PS-CS. Whether it’s the fault of Minolta or PS I don’t know, but PS does not recognise the A1 setting AdobeRGB that is not embedded, and if you manually assign the profile the colour does differ slightly from the working space profile.
When I shoot using the setting Minolta call Adobe RGB(1998), PS tells me that it is sRGB and does not recognise it as the Adobe profile. unwittingly, I was then converting from sRGB to working space, hence the colour shifts. If I instead assign the Adobe profile, theoretically all should be well. However there is still a very slight colour difference that is noticeable that I detected when, out of curiosity, I asked PS to then convert to the working space, though again, theoretically, if the Minolta profile was correct, there should not have been a difference.
Finally, if I shoot in the camera setting AdobeRGB(embedded) then PS does recognise it and all is well.
Also, being a Mac user I was downloading into iPhoto because of it’s superior thumbnail function, and I had not realised that this programme does not recognise embedded profiles, so I was merrily thinking that the shots I thought were embedded, actually weren’t.
So yes, it was the wrongly assigned profile, but it was not straightforward to sort out.
I can’t think of any digital one shot cameras (Canon, Nikon, Fuji etc) that actually embed a profile. They use EXIF data to describe what you asked for (and actually didn’t get) in the RAW to color conversion. DCF (DESIGN RULE FOR CAMERA FILE SYSTEMS 1.0 was introduced by a group of Japanese camera manufacturers which "embed" this EXIF data. Version 1.0 only specifies "sRGB" or "None" so if you ask for Adobe RGB 1998 this system places no real info for Photoshop to see. It’s really the DCF guys who didn’t think this out very well.
DCF 2.0 (pretty new spec) does allow Adobe RGB 1998 to be embedded in the EXIF data but not too many newer cameras yet support it.
IF you ask your camera to give you "Adobe RGB 1998" (notice the quotes) you need to assign this in Photoshop (use Assign Profile command). IF your working space was set to Adobe RGB 1998, the image without the tag would still look fine since the data IS in Adobe RGB 1998 and Photoshop is assuming all untagged images are in Adobe RGB 1998 but you still want to use the Assign Profile command to tag (embed) an actual ICC profile into the file.
Oh, you all know there isn’t a camera (or scanner or output devicee) in the world that actually produces or output’s sRGB or Adobe RGB 1998. These are synthetic color spaces (like LAB). The camera manufacturers would like you to believe this is what you’re getting. Might be close. You want Adobe RGB 1998 (or better)? Shot RAW, convert using Adobe Camera RAW.
Thanks Andrew your post confirms what it took me two months to work out. However, the Minolta A1 does seem to "embed" a profile when you shoot in the Adobe RGB(1998) embedded mode, because PS then opens the files into that working space without giving the warning.
However, I take your point about shooting in RAW I’ll just have to save up for bigger memory cards!
Be interesting to know if it’s actually embedding an ICC profile or it’s just one of the cameras following DCF 2.0 and using the correct EXIF data. If you’re on a Mac, you can extract a profile using the supplied AppleScripts (Extract Profile). I would suspect if a profile isn’t present, it will pop an error and if the profile really is in there, you’ll see a profile of Adobe RGB 1998 show up on your desktop. Not a big deal but an interesting test.
Photoshop CS does recognize the Adobe RGB tag in the EXIF data.
Photoshop 7 was written before that tag was added.
