Re: [Gimp-print-devel] Are dithers multi-shade aware?

[Robert K. <rl...@al...>]

On 6/29/21 9:28 AM, Walker Blackwell wrote:
> Dear devs,
> 
> I’m looking Gutenprint code and trying to determine if each channel is dithered independently of every other channel or if they take into account darker shades as no-go placement-points for lighter shades (or vis versa). 
> 
> Eg: For Vivid Magenta dots, no Light Magenta dots should be printed at the same coordinates. Is this what “masking” is in the code? If each channel is independently dithered I’d be interested in funding per-color channel-aware dithering . . .

It's not as easy as it sounds.  In 4.x (when it was called Gimp-Print), it dithered multiple shades
together by creating "virtual" shades, the product of the drop size and the relative darkness of the
ink.  For some things it worked quite well, but it had some problems that I don't remember off the
top of my head.

With the standard Ordered dither it's very hard to do this, since it's using a fixed mask.  That's
not the best way, but it's very fast and with small-drop, multi-shade inks, it's almost as smooth as
the best error diffusions that we've tried with none of the quirks of the latter.

Raph Levien came up with Even Tone Screening (which I implemented) and later came up with Even
Better Screening (which I didn't; it was rather thorny).  I don't remember whether there's any
handling of multi-shade inks in that.