QTR-Quadtone RIP

Hello,

I have recently installed QTR (QTRGui on Windows XP) and started by
attempting to linearize and profile my Epson 2100. The documentation,
both included and on the web, is not always 100% clear, so I hope that
anyone can confirm my understanding of the software. In addition, if I
*do* understand things correctly, there seem to be a few issues with
the ICC profiles that are produced by QTR-create-ICC - more on that below.

1) It appears that QTR/QTRGui is a 'color-dumb' printer driver in the
sense that it is not aware of any ICC profiles associated with the
image data. It simply takes the numbers, processes them and sends them
to the printer. Correct?

2) Linearizing the QTR profile has the goal of mapping the input
values (numbers without meaning) to a perceptually uniform lightness
distribution. Steps in the input values correspond to proportionally
sized steps in the lightness range Lmin to Lmax. Correct?

Now on to the ICC profiles. The profiles contain 6 LUTs, 2 each for
the perceptual, colorimetric and saturation intents. Inspecting the
contents of the produced profiles, I noticed the following:

a) The BtoA0 table (perceptual Lab to printer) incorporates a black
point adjustment similar to (the same?) the black point adjustment as
usually applied in relative colorimetric transforms. The explicit
recommendation to use black point compensation in the QTR
documentation does not seem to be necessary - it makes no difference.

b) The AtoB1 table (colorimetric printer to Lab) correctly shows the
lightness values that correspond to QTR input numbers. Together with
the BtoA0 table, this ensures a correct soft proof (in perceptual mode
only - see below!)

So far, so good. However, the corresponding inverse tables (AtoB0
(printer-Lab perceptual) and BtoA1 (Lab-printer colorimetric)) appear
to be incorrect. First, BtoA1 is set equal to BtoA0, whereas it should
approximate a straight line from (Lmin,0) to (1,1). This miscalculated
table breaks the relative and absolute colorimetric rendering intents.
This is most clearly seen in the complete loss of dark tones when the
relative colorimetric intent is used in combination with black point
compensation.

Second, the AtoB0 table is set equal to to the AtoB1(colorimetric)
table. This has severe side-effects when images are converted to the
profile space and later converted again to another space. Funnily
enough, the QTR-create-ICC-RGB-bpc application *does* appear to make
the correct AtoB0 table, but in doing so it breaks the AtoB1 table
(used for soft proofing). I gather QTR-create-ICC-RGB-bpc.exe was made
to address issues with a certain application. Would a single
application that creates correct AtoB0 and AtoB1 tables fix it as well?

Can anyone confirm what I'm seeing - or tell me what I'm missing?

Thanks!
Simon