Re: [Gimp-print-devel] Canon driver

[Andy T. <th...@ph...>]

Robert L Krawitz wrote:

>    Date: Mon, 08 May 2000 09:01:03 +0200
>    From: Andy Thaller <th...@ph...>
>
>    I've tested the canon driver's output this weekend, though not with
>    every dither algorithm. 360x360 and 720x720 dpi seem fine, though
>    slighty mistuned in respect to brightnes and contrast. 360x360 with
>    variable dot sizes shows extreme color corruption (what can I do
>    here?) and 1440x720 is way too dark.
>
> 1) For 1440x720, you need to divide v->density by 2.  Apparently the
> Canon printers are like the Epson printers in this regard -- they're
> really 720x720, and 1440 is done in two passes.
>
> 2) You probably want to fiddle with the_levels.
>
>   double        the_levels[] = { 0.5, 0.75, 1.0 };
>
> This indicates that there are three dot sizes, of size .5, .75, and
> 1.0 respectively.  However, I doubt that those being a bit off would
> cause even moderately severe color corruption.  Most likely the format
> of the output is wrong.  The dither routine normally generates the
> output as concatenated rows, with the lowest bit plane first.  It's
> possible (by means of dither_set_c_ranges and such) to change that
> ordering around.  If Canon takes each pixel as a full unit, you need a
> folding routine like that in print-escp2.c

The driver already had the folding routine but somehow the pixel order got
mixed somewhere during the last changes...
Now it's fixed and every mode works fine again - except the color balance,
this I want to leave to someone else.

Andy.