[argyllcms] Re: [argyllcms] Re: rgb printer calibration

  • From: Nikolay Pokhilchenko <nikolay_po@xxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 27 May 2010 17:27:30 +0400

Sun, 23 May 2010 21:52:01 +0200 Radek Doulik wrote:

> On Sun, 2010-05-23 at 21:34 +0200, edmund ronald wrote:
> > Maybe printing in 16 bit space thru the printer driver might help?
> >
> > One could generate better values close to white that way.
>
> Yup, we use 16bit path until we reach the 8bit driver. Everything looks
> nice and smooth on display with hardware LUT, but the proof shows issues
> in soft shadows. So I am quite sure the original images are OK.

Radek, there is no problem with CMS, there is the problem with printing 
equipment. The solution is to add some uniform noise to the image. You should 
do the next:

1. Print a gradient strip (gray for example) across the long sheet side (You 
may print on plain paper - it's no sense for this test), from 0 to 255 (8-bit 
scale).
2. Look at the gradient and define the gradient bads edges. Mark them with a 
pencil.
3. Measure the largest step between bands with the ruler.
4. Compute the band width in percents from the full scale (255). Possibly it 
will be 1-bit of 8-bit-scale or more.

That way You may know the amount of dithering to eliminate the banding.
The next steps:

5. Generate the target images with printtarg, they should be bitmap (TIFF). The 
charts should have the same resolution as Your printing mode have (for example 
720 dpi).
6. Add the uniform noise to the chart images. The amount of noise should be the 
half of the measured band width at the step 4 (for example, 0.5%).
7. Print the target images with noise.
8. Measure targets and build the profile.

Now You can process Your images:

9. Convert Your images with the new "noisy" profile.
10. Resize images to the printing resolution, as the target was and if the 
image have different.
11. Add the noise to the images as at the step 6.
12. Print images in the same resolutions, as the targets was printed.

Now You can see, that the lights are smooth, but the overall image become less 
detailed and, in some degree, noisy.
Such workaround I've successfully used with a non-UV printer while printing on 
the uncoated glass.

Other related posts:

  • » [argyllcms] Re: [argyllcms] Re: rgb printer calibration - Nikolay Pokhilchenko