[argyllcms] Re: ArgyllCMS: shadow details and blue color issues. What I'm doing incorrectly?

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 15 Oct 2012 17:47:50 +1100

Xavi wrote:
> On my case the softproof and the print are very very similar. On prints,
> using perceptual intent, I see the blue color as a light purple color, as I
> see on the screen when I softproof with Photoshop CS6. Also checked on
> prints with 2 models of spectros using spot read.

        I'm guessing that the image you are using is from the
DQ-Tool Monitor Reference File by Fujifilm Europe GmbH / Jens Rubb.
If so, it's an unlabelled (ie. no colorspace) RGB file.
The color of the "blue" fan bland is something like .09, .14, .55 RGB,
so it is not a pure RGB blue, but is tending a little towards purple
in the original, and all four of the soft proofs hint at this.

> Also the overall color is changed to a magenta cast, when I use perceptual
> intent. Using the saturation intent the blue color is less purple color and
> is more vivid but skin colors, for example, are more reddish and excessive
> saturated.

I can't verify this - the profile through perceptual seems very self consistent
and neutral, ie:

icclu -fb -ip Argyllv140_396parxes48Gris3Blanc.icm
50.000000 0.000000 0.000000 [Lab] -> Lut -> 0.345472 0.388520 0.513978 [RGB]

icclu -ff -ir Argyllv140_396parxes48Gris3Blanc.icm
0.345472 0.388520 0.513978 [RGB] -> Lut -> 52.334969 -0.072152 -0.770024 [Lab]

< 1 delta E hue & saturation error seems reasonable through a perceptual
B2A Lut.

It tends towards cyan as it nears the black point, due to Argyll's
"extend and bend" neutral mapping.

> I have created a collage with 3 cropped photos. Please, tell me if you see
> the reddish/magenta cast on color skin. And also, if you agree with me that
> the blue color of the sky, seems a blue/purple very light and not saturated:
> http://i436.photobucket.com/albums/qq82/aaruizz/prcptualvsRC_zps09f2fe42.jpg
> On the other hand I see very natural colors, on the right image, softproofed
> using the same profile, but with the colorimetric intent.

You soft proof does look somewhat reddish - but then I'm not sure how
your softproof workflow is functioning. For instance, you are using BPC
which could be having an effect on the neutrals, given the very non-neutral
black point of your colorspace - ie. Argyll maintains neutrality down
the neutral axis until it nears the black point, at which stage it
becomes cyan tint, to match the black point. If the BPC then maps the
cyan black to neutral with the mapping being in a straight line to the white
point, then the neutral mid tones will get shoved in a megenta direction.

Note that typical Argyll workflows won't use BPC and are not created to
anticipate the effects of applying BPC, because you can do a more
sophisticated mapping using the collink gamut mapping machinery. My typical
softproof workflow is as follows:

# Create proofing transform from output to AdobeRGB
collink -v -ila -G -qh -cpp -dmt Argyllv140_396parxes48Gris3Blanc.icm 

# Create proof image back in AdobeRGB space
cctiff Argyllv140_396parxes48Gris3Blanc_v.icm skin_sky_p.jpg skin_sky_pv.jpg

The above softproof (not using BPC) doesn't show such a tint.

> I am doing all the profiles with a Canon Pro9000MKII with third party inks
> (Hobbicolors UW). I will do,  from zero, new profiles for another Canon
> MP540, that uses only four different color cartridges (CMYK) and also I will
> start creating profiles, from zero, for the Pro9000MKII but using cartridges
> refilled with other third party ink (OCP), that uses a very neutral black
> color but that their gamut is more limited vs Hobbicolors Ultra Wide Gamut
> inks.

I think a neutral black color will give better visual results, and leave less
room for workflow details to affect the outputt.

Graeme Gill.

Other related posts: