Dull colours on PIXMA Pro9000

[Gwillers]

I was wondering if anyone else was suffering with similar colour problems with the Canon PIXMA Pro9000 or similar printer. Or maybe, I'm doing something wrong?!:oops:

I have a Canon 30D camera and shoot in RAW format, assigning Adobe 1998 as the colour profile. I edit and print my photos in CS2 (Mac OS X) with Adobe 1998 as my work space. I use Canon inks and Canon photo pro paper.

I have had some success printing photos that are bright and have blue/red "themes". However, when I print rich, high contrast green photos (e.g. child portrait with tree background) the colours become washed out and have a green cast (the shadows grey and significant detail is lost e.g. my lovely round bokeh disappears into a muddy off-green blur!).

The problem seems to be related to the Print profile, when I view the print on screen (soft proof) with the Canon 9000 PR1 profile, it visibly changes - from a good photo to an unusuable one! and the hard copy print reflects the soft proof very well!

Any thoughts?

[René Damkot]

Have a read in the link from my sig. There are a few links on printing, softproofing, and profiles with Canon Printers. You also might wanna read the bits about color space conversion and rendering intent.
Some colors just aren't printable, so they will be replaced by ones that are printable. You have to decide how the colors are changed by choosing the right rendering intent...

[Gwillers]

Thanks for the reply.

I must admit, I have been using your link for the last few days, and without it I would not have got this far (I have Fraser on order, but fear it will be a long and tortous process to fully understand it - although I appreciate it is necessary) - so again thank you.

I have followed the various steps and can print photos that match the soft proof (both via PS "Print with Preview" and the Canon Easy-Photo Pro plug-in). Where I really need help is, from your experience, whether "soft proofing" in the manufacturers print profiles should change the view of the photo so significantly that it becomes "horrible" to look at (low contrast, green cast and grey whites). I have managed to edit the photo back to the original view but the layers are extreme.

Appreciate your time on this one, David

[CyberDyneSystems]

Unless you have a bad copy, it's not the printer,. Canon printers deliver rich vibrant colors..
Though I can't help you with this specifically, I can tell you that I had similar issues until I got my head wrapped around color management enough to get it working.
It improved the output from my Canon printer dramatically.

Colorspace/color management, and the right paper made all the difference.

[René Damkot]

Well, I just gave it a quick try, and loaded my display profile (Calibrated) and Canons PR1 for the Pixma 4000 into Colorsync utility.
My display displays roughly sRGB: Way less greens then AdobeRGB (Adobe goes off the image on the left). It is clearly exceeding the printers gamut in green. So I could imagine that there would be problems if the wrong rendering intent were used.
If I were to guess, I'd think perceptual would give best results...

If you could find a profile for the specific paper you're using, I'd think that would improve things quite a bit....

Also, do you check any boxes in the soft proof window?
Post a few screenshots...

[Gwillers]

The theory would definitely make sense, as I seem to get a much better match with photos that have a strong red and blue bias. The bit that suprises me is that skin tones change significantly - I would have thought that would have been within the range.

I am using Perceptual (it does give the better result - greater change then Relative Colourimetric but more pleasing output).

I am seeing the change when I go to VIEW/PROOF SETUP/CUSTOM and,

1. Device to simulate: Canon Pro9000 PR1 (canon's profile included with the printer for their max gloss paper)
2. I am NOT checking: Preserve RGB numbers
3. Rendering intent: Perceptual
4. I AM checking: Black point Compensation
5. I am NOT checking either: Simulate Paper colour or Simulate Black Ink

I then click the preview button and it all goes wrong! I get exactly the same results if I convert the profile

I have checked the bottom left hand corner of the photo in CS2 and it is Adobe RGB (1998) (16bpc).

Hey, again thanks for your time

[René Damkot]

The bit that suprises me is that skin tones change significantly - I would have thought that would have been within the range.

So would I.

The skin tones (if in gamut; check with gamut warning) should not change one bit when you use Relative Colorimetric. I wouldn't expect them to change much when using Perceptual.

Sounds like you do everything correctly.
Some images are just very difficult...

Something you could try, is duplicate the document, convert to PR1 profile using Perceptual, go back to the original document, duplicate agian, convert to PR1 using Relative colorimetric, and sandwich the two documents converted to PR1.
Use a mask to get the skintones from the one, and the rest from the other...

Didn't have a 'green' image at hand, but found another 'difficult' one...
The image below is a screenshot of an AdobeRGB image (taken inside of an Architects of air (http://www.architects-of-air.com/main.html) 'thingy'), with below that a proof, using Perceptual. If I proof the same image using Relative Colorimetric, the skintones stay virtually identical to the original, but the top left corner changes dramatically...
Impossible to get it 'good' with one setting.

[René Damkot]

third image is using relative colorimetric, last image is the two rendering intents sandwiched and masked...

[Sathi]

Hmmm seems like I have the opposite problem as you with my 9000pro. I shoot raw and open it in CS2 as Adobe 1998 like you do. If I print from photoshop in this colour space (which I read the 9000pro should be able to take some advantage of) I get very over saturated colours. Especially in red and green. If I convert to srgb and jpeg, and then use that easy photo print jammy of canon's then the print is almost a 100% match with what I have on the screen. I havn't done much testing so I do not know if the problem is resolved because I converted to srgb, converted to jpg, or using a different program to print....or a combination of those factors. It takes me almost no time to do the conversion so I don't really care, but I am concerned that I might be missing out on some of the argb gamut that the 9000pro is supposedly capable of tapping into.

[René Damkot]

Read the link in my sig: the links about printing and soft proofing...

[Gwillers]

I see what you mean. But the variations in all your samples are to me (untrained eye) subtle and certainly within the realms of my acceptability. I have attached 2 versions of the same photo (one with Adobe 1998 profile _MG_5042.jpg and the other after converting to the Canon Print profile _MG_5042 (1).jpg) - action performed in CS2 using Perceptual rendering.

I exported both the files as JPEG with a sRGB profile - must admit, this is where my head explodes, as they each look the same as they did in Adobe 1998 and Canon Print profile (but I wont worry about that right now) via CS2 - my Fraser book turned up today (hurray), so I'll worry about this bit later.

Hopefully you can see that the _MG_5042 (1).jpg is really dulled and detail is lost. Am I going mad?!

Yours truly grateful.

David

[Sathi]

On my monitor the top photo is darker and losing detail to shadows. But there is extra detail in the hair where the brighter one is lost to whiteness.

[René Damkot]

I see what you mean. But the variations in all your samples are to me (untrained eye) subtle and certainly within the realms of my acceptability.

Noticed the background in the upper left? ;)

I exported both the files as JPEG with a sRGB profile - must admit, this is where my head explodes, as they each look the same as they did in Adobe 1998 and Canon Print profile (but I wont worry about that right now) via CS2 - my Fraser book turned up today (hurray), so I'll worry about this bit later.

Nothing to worry about. The colors that are in the image once it's converted to the PR1 profile, will fit in sRGB quite well. So they won't change (much, if at all) when converting to sRGB when using relative colorimetric. The colors of the AdobeRGB image, probabely won't fit into sRGB, but will get changed more subtle.
Doing a screenshot is the easier way however...
If you use perceptual to convert from PR1 to sRGB you'll get your original colors back (providing no clipping occurred).

Hopefully you can see that the _MG_5042 (1).jpg is really dulled and detail is lost. Am I going mad?!

Yep, I see it. Hard to miss...
You're not going mad, but you ere seeing the differences between ink on paper vs. light from a screen. I'd expect the blacks to be (quite a lot) lighter. That's what ink on paper does... It happens as well in my example, but it's a lot more visible in yours, since there is a lot more dark background.

I am surprised about the vast change of the hair though...

Gave it a download, and tried some other printer and paper profiles on it.
Seems to me that the Canon profile is a 'generic' one, and not targeted at one specific printer/paper. Seems rather strange that my Pixma 4000 uses the same profile as your Pixma 9000. (Or doesn't it?)

I tried a profile for epson glossy paper and my Pixma 4000 (smaller gamut then your printer I'd think certainly smaller then PR1, according to color sync utility), which produced a way better proof. Must be a choice Canon made: "Since the darks get lighter, well also make sure the lights get lighter".
Post some screenshots in a sec...

[René Damkot]

Nice thread.
Makes my brain work....
Took a bit longer to upload the images, since, when taking a screenshot, I noticed my mac was using the wrong display profile ???
Have to look into that one...
Anyhow:
Canon profile vs. Ilford (http://www.ilford.com/html/us_english/prod_html/galerie/IJProfiles/default.asp) Glossy:
http://www.moonglade.net/rene/POTN/color/printprofiles.jpg

Using this profile for soft proofing:
Original (sRGB) image:
http://www.moonglade.net/rene/POTN/color/attachment.jpg
And two conversions:
http://www.moonglade.net/rene/POTN/color/rel_col_vs_perc.jpg
For comparison: The image converted with perceptual rendering intent to PR1:
http://www.moonglade.net/rene/POTN/color/attachment-1.jpg


Very different interpretation than the PR1 profile.
I'd say that's more because of choices made, or accuracy of the profile then gamut limitations...
Color management is about compromising, but I'd be interested what choices are made where and why.

Back after a reboot ;)

[René Damkot]

Well. Figured that one out... At least half of it.
I had been giving BasICColor Display 4 a try. While it did a better job at calibrating the monitor, it's profile wasn't used when making a screenshot or in Safari. Photoshop did see it, and so did the system prefs...
Very, very weird. I'll have to wait what their tech support has to say about that.
Oh well, back to Monacos' software...

Redid the screenshots in posts #7 and 8, so they look like they were supposed to...

Back to the regular programming & off to sleep for me :D

[Gwillers]

Hey Rene, massive thanks for all your time - this is really interesting (I'm learning at a rate of knots).

Bottom line - I'm not doing anything wrong and therefore I need to learn to work within the changes (one of your links in your SIG provides a workflow for this - again, thanks).

I am very interested to hear what tech support say (the Canon 9000 PR1 profile is not generic, it is designed for their high gloss "Pro" paper).

Cheers

[Gwillers]

Sorry, just read my note - to be clear I am going to contact Canon tech support to inquire about "the significant shift" in dark green colours with their 9000 high gloss Pro paper profile (strangely one gets less of a shift and better results with their cheap "everyday" gloss paper). I'll post their feedback - assuming they ever reply!

[René Damkot]

Well, done a bit of (re-) reading on soft proofing and rendering intent, and I think I understand some things a little better:

From what I've read, the profile also contains the info on which the conversion is based. In the printer profile for instance paper color and such.

Perceptual rendering intent will 'fit' (compress) all colors of the image color space into the target (printer) color space. Regardless of whether the extremer colors are actually in the image.
So if you have an image full of pastels, well within the printers Gamut, they will still be compressed. I hadn't realised that.
That explains the rather big shifts in the skin tones.

Relative colorimetric shifts the white point like Perceptual does, but clips instead of compresses out of gamut colors. If there are no out of gamut colors, (pastels again), it won't alter the image apart from the shift, so is the best choice.

The white point is shifted because "255, 255, 255" on the monitor looks different then "no ink" on the paper. The shifted whitepoint makes that color difference visible. (Otherwhise 'white' had to be printed)

One thing I *still* don't completely get is the Gamut Warning.
I thought it was showing 'out of Gamut' colors. But when I use 'perceptual' rendering, there *are no* out of gamut colors AFAIK: All is 'compressed to fit'.
If the gamut warning gives an indication what colors *where* out of gamut, so were 'compressed', I'd expect no change in Gamut warning between perceptual and relative colorimetric intent.

Edit: Must have been sleepy when I wrote this: I tried again, and there is no change in Gamut Warning between perceptual and Relative Colorimetric...

Best explanation Yahoo found me until now is something along the lines of "Beware, these colors might print screwy' :lol:

On the proof looking vastly different from the original: (From one of Bruce Frasers' articles (http://www.creativepro.com/story/feature/15310.html)): "As a result, your first reaction when checking paper white may be that your image just died before your eyes. I've become accustomed to looking away from the monitor when I check Paper White so that I don't see the change happen."

I'll have a further look into this, since I'm curious, and like to understand this.
Now if anyone has seen UncleDoug lately.... Help? ;)

More articles: Bruce Fraser (http://www.creativepro.com/story/feature/10150.html), Dry Creek Photo (http://www.drycreekphoto.com/Learn/profiles.htm#SoftProofing) and a lot of pdfs from DigitalDog (http://www.digitaldog.net/tips/) (very usefull!)

[René Damkot]

On the Gamut Warning: Got a better explanation from UncleDoug (Thanks!)

René

The gamut warning indicates what colors will need to be edited, or compressed/clipped, in order to get the image "into gamut".

Lets say you have an image that was converted from RAW and set to ProPhotoRGB.
When you toggle gamut warning on, it is comparing the ProPhotoRGB numbers to the Proof Setup numbers and "greying out" the colors that will either be clipped or compressed when converting to the Proof Space.
Now, if you edit the image, which is still in the ProPhotoRGB space, say by desaturating the entire image, the grey of the gamut warning will slowly shrink and finally disappear once all of the colors are "in gamut" vs. the proof space.

Does this help and make sense?