Datacolor User to User Support Group.

On Sep 2, 2009, at 11:19 PM, Bob Petruska wrote:

> 1.  Windows Vista Home Premium.
> 2.  NEC P221 LCD monitor calibrated using NEC's SpectraView II with  
> a NEC supplied I1 Display 2 calibrator that is special calibrated  
> for their wide gamut monitors.  This calibration is performed in the  
> monitor's hardware and stored in the LUTs.


Not a Datacolor product, but it certainly should produce a reasonably  
calibrated display... just be sure its not too bright or too dim for  
your viewing environment.

> 3.  Canon 6 ink I950 printer.  Using Canon Photo Paper Pro paper for  
> this test.

Not the latest and greatest printer, and not anything I'd use for long  
life prints, or for black and white, but it should produce reasonable  
color photo prints...
>
> Test....
>
> 1.  I set up a simple test target of objects with various colors,  
> bright whites, deep blacks, shades, etc. in a totally dark room  
> illuminated by tungsten lights.

Then you are not testing display to print, you are interjecting all  
sorts of complex scene and space conversion issue up front...


>   Using my Nikon D700 I manually measured/adjusted white balance and  
> shot in the  SRGB space.

There is no sRGB in RAW, RAW is RAW, at gamma 1.0 (not sRGB's 2.2) and  
without gamut clipping at the edges of the sRGB gamut.


>   The Nikon RAW file displayed on my NEC monitor in Nikon Capture  
> NX2 looks identical to what I see in real life.

Can't comment on propietary RAW import software, except to reccommend  
against it, in favor of industry standard methods, like Camera RAW and  
Lightroom...


>   I saved the file in NX2 as a TFF and viewed it Photoshop CS4 and  
> it matches identically to what I see in NX2.  So we can say that the  
> monitor is well calibrated and both applications display the proper  
> image.

Not sure that process guarantee's your display is appropriately  
calibrated, but you seem to come through a whole set of front end  
stuff and be happy with the result. So lets view it as a mountain,  
where all the source stuff is on the "up" side of the hill, and the  
corrected image in a Photoshop working space is at the summit, and  
what we really want to focus on is the "down" side, where we take a  
precorrected, tagged RGB image, and print it.

But of course there are still papers and inks and gamut limits in the  
file and gamut limits on the screen, and display black versus file  
black versus ink black, and display white, versus paper white, versus  
file white, and viewing light color, quality and brightness, and  
universal file view versus softproof view etc... so we aren't without  
complications, they are just complications on the appropriate side of  
the hill now...
>
> 2.  I ran the Spyder3Print calibration again using the Canon I950  I  
> turned off all color management in the Canon I950 driver, color  
> adjustment is set to manual, not much else to change there.  Printed  
> the Spyder3 test patterns, colors plus grays, scanned them in.   
> Printed the Spyder3 test photo and it had slightly more saturated  
> colors, slight red hue than what I see on the monitor, the B/W  
> sections were leaning more to a very dark gray tint than black.

Okay, this is in terms we can work with now. You are viewing the image  
on screen, in Spyder3Print, so its softproofed for you. You are  
printing it on some type of paper (would be handy to know what type),  
and seeing some visual differences between screen and print. Here are  
factors, causes, and reccommendations on that:

Display black is weak enough that softproofed black on screen looks  
weak, can be described as "lack of punch" (a matte paper issue,   
mostly). This can be improved by running your LCD at a higher  
brightness level, since choking it down to prepress levels drops the  
dynamic range a lot. Or it could be improved, inversely, but dimming  
your display, as running it lower produces a darker black. Sorry for  
the two inverse suggestions. Its not our display calibration product,  
so I can't say much more than that.

Next, most users are THRILLED with more color saturation in prints.  
They can always reduce it, but if its not there, there's no way to  
increase it. This is mostly on gloss and luster media, since matte has  
a smaller color gamut. No mention of your proofing light, but using an  
incandescent Solux bulb is the most common cause of "popped" prints.  
Other recommendations would be to change your intent from Saturation  
to Perceptual, and if its stll too punchy, to Relative Colorimeteric.  
Thats part of what the intents are there for.

On to your blacks. I can't interpret "leaning towards dark gray not  
black" but there is always going to be a color of black in any ink/ 
paper combination, and we can't really change that. So either you are  
speaking of the near blacks having a tint (tint means color), or you  
mean blacks aren't black enough to suit you, which isn't typically a  
profile issue, its a paper/ink/media setting issue. But the match of  
the softproof to the print is a printer profile issue, best adjusted  
at SooftProof black, by reducing the L* value.


>   I looked at the soft-proof .   The Spyder3 soft-proof colors  
> matched the monitor test photo but were slightly washed out, so I  
> lowered the black level in the soft-proof down to 1 as suggested by  
> David T. and the soft-proof matched the non-soft-proof test image  
> identically for color and contrast.

Matching softproof to nonsoftproof isn't the goal here. By zeroing out  
the L* value you have entirely removed the function that emulates  
paper black on screen. If your printed blacks are as dark as your  
display, that would be appropriate (likely with a dimmed LCD, and  
gloss/luster paper), but if you are printing on matte, removing this  
correction entirely will lie about your prints, and promise darker  
blacks than the paper and ink can manage. So adjust appropriately for  
your display and your media...

> Now this is nice that I have the test image and soft-proof image  
> matching, but neither match the printed photo.  If I edit the  
> profile to reduce the saturation I can get the printed photo to  
> match closer to the original test photo, but then the soft-proof  
> image doesn't match since any slider edits to the profile is also  
> going to change the soft-proof image in a negative way.  I did a lot  
> of profile editing and the printout still does not match the  
> original image as well as printing using the Canon Standard Driver.

No one, even a Canon fan like myself, has ever suggested that the  
"Canon standard driver" is a highly accurate color solution... good,  
but certainly not excellent.

First, you are adjusting in the wrong order. If you wish to edit the  
print output, do that first ( you may be the first person I even knew  
to edit in the direction of reducing the printers color gamut, but if  
thats what you want, then do it,  but do it carefully). Then, after  
that is done, make your softproof adjustments. But that may not solve  
the issue you are describing, which is sounding more and more like  
"metamerism" (AKA: color constancy)
>
> 3.  Now I went one extreme step further and found a new in factory  
> sealed box Canon Pro9000 MKII for $250 and bought that to compare in  
> this test.  That is an 8 ink printer (additional red and green inks  
> with the 6 of the I950).

I've never been impressed with that printer, though I love the pigment  
ink version, the Pro 9500. But I'm beginning to suspect that your  
issues lie in the dye ink/gloss paper end of things. That would be  
about the only way I could conceive of exceeding the desired gamut...

> Ran the same as 2 above, got the same results;  printout is slightly  
> color saturated, towards red tint, soft-proof slightly washed out  
> and fixed it with the soft-proof black level.  Canon Standard Driver  
> matches the original image very well.  I would like to say that this  
> printer is one huge printer and I really can't see any printout  
> detail differences against the I950.

As for color tint, we really can't talk about that without discussing  
what lighting is involved. I would certainly expect any profile (since  
they are targeted for 5000k) to look off color under incandescent  
light, and in fact my biggest complaint about the Canon Pro 9000, is  
that is a "pro" printer with all the metamerism issues that "amateur"  
dye printers have, the most obvious being ugly color tones and non- 
neutral grays under incandescent light. So perhaps the first thing to  
do here is to get a high CRI fluorescent light with about a 5000k  
color temperature (an Ott-Lite would do), and view your prints under  
that. If the neutrality, perceived color cast, saturation etc... are  
different under that light, then we've found the source of the  
problem: Canon dye inks "metamerising" (technically "showing color  
inconstancy") under non-5000k, low CRI light sources. No real color  
management solution for that one, other than trading your Pro 9000 in  
for a Pro 9500 with pigment inks.
>
> 4.  Now believing that something is not set correctly in Windows  
> Vista or my system that I just couldn't finger.  I pulled out my old  
> Windows XP system, ran the monitor calibration, attached both  
> printers, printed out the test sheets, scanned, adjusted the soft- 
> proof, and the printouts still show slight over saturation, red  
> tint, etc.  Again the Canon Standard Drivers match the monitor image  
> almost exactly.
>
> 5.  Put a very good SONY CRT monitor on the system, calibrated it  
> and the same results as 4 above.

You have certainly done your homework...
>
> I realize that there are many sliders to edit the profile but after  
> a few boxes of paper and many ink cartridges it is difficult to get  
> close to matching the printout to the display.  I do realize gamut  
> limitations but the Canon Standard Driver does a great job.

I refer to such results as "Eden"; lovely if you are there, but you  
can't go back again. If your color only matches by accident, there is  
no tool to make it match again later, when something changes...  
Thinking back to the days of Canon dye printers, I do seem to recall  
that the canned results tended to lean towards an incandescent match...
>
> I was expecting that the Spyder3Pint was going to make it painless  
> for me to generate profiles with different papers, I assumed you  
> just printed the charts, scanned, and bingo the as perfect as  
> possible ICC profile would be generated that would beat the Canon  
> Standard Driver every time.

Many users would describe their experience as being roughly what you  
expected. But lets see what a 5000k High CRI fluorescent proofing  
light does for the results. If it magically makes the prints look  
right, then I'd say you may simply not be happy until you get a  
pigment printer (or may simply give up an use canned profiles, if you  
can't afford a less problematic printer).
>
> I really don't know where to go from here.  I must be missing  
> something somewhere.
>
> I can use all the help and suggestions that anyone can give me!

Many comments, but only a couple suggestions, above... let us know how  
things turn out!

C. David Tobie
Global Product Technology Manager
Digital Imaging & Home Theater
CDTobie@...