[argyllcms] Re: targen options

Gerhard Fuernkranz wrote:
> Btw, I found the test prints I had archived.
> You can take a look at http://home.arcor.de/gfuer/tmp12116/
> In the device RGB image sent to the printer driver, the gray
> wedge on the left side has pretty smooth RGB gradients.
> But in scanned_print1.jpg you can see horiziontal stripes,
> i.e. the printed color changes rapidly as a function of the
> gray level (at the dark end there is a color "jump" at approx.
> every 2 RGB levels). Since these stripes only occur in color
> gradients, and not in areas with a uniform color, it is evident
> that they are caused by the printer's color response (and not
> e.g. congested nozzles).

Unfortunately, the scans don't really convey as much as examining
a print in real life. In my experience at Colorbus, we found it
difficult to get (reasonably) smooth looking neutrals unless we
controlled the output values better than 8 bits/component (and
this was for a CMYK driver). That meant setting up 12 bit screens,
as well as 12 bit device values in the calibration system. We
got reasonable control over the device through the 12 bit
calibration systems with 8 bits/component from the RIP, but
the output was not quite flawless (but superior to what would
be expected for non-professional output). My impression about
the remaining imperfections was that they were due to a combination
of the 8 bit linearised channel quantisation, limited resolution
of the ICC transforms, limited measurements in the device profiling
etc.
There was a huge difference in smoothness between the conventional
ICC workflow (source + destination profile passively linked), vs.
the smart CMM workflow (ie. Argyll icclink -G, transform using
device link). This was to be expected, since the conventional workflows
concatenate the "bumpiness" of the two ICC profiles (particularly
the output B2A), whereas the smart CMM workflow deals directly with
the more accurate, better behaved A2B tables only.

> (Actually I'm disappointed that professionals write such
> a rugged printer driver - I cannot imagine that such flaws
> were not avoidable. I do not think, that e.g. gimp-print
> behaves as badly. Unfortunately it has no support for
> my printer in high-resulution photo mode.)

It's a little tricky creating a separation (RGB->CMYK) that
meets all the competing requirements, and you may be seeing
the basic limitations of the technique chosen. I've had a go
at such a task, and while I achieved very acceptable result
for an office type color printer, the results were a little bit
worse in terms of smoothness than a direct CMYK workflow
(ie.
   sRGB -input-profile-> PCS -output-profile-> dRGB -separation-> CMYK
vs
   sRGB -input-profile-> PCS -output-profile-> CMYK
).

Graeme Gill.

Other related posts: