[argyllcms] Re: Calibrating a projector with a X-Rite DTP92

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 04 Feb 2007 23:37:42 +1100

Frédéric wrote:

Last night, Mike and I made some tests to calibrate a video-projector, using a X-Rite DTP92 colormeter.

Pictures need to be over-saturated to look good when displayed on that projector, and this is what the profile does. On mid-saturated pictures, the results are pretty good. But on pictures containing already saturated areas, the results are very bad, especially in the red channel (the area is burned).

Looking at the profil using iccexamin first shows that the LUT is strangely corrected. The red and blue channels do not start from zero, but from 0.15 and 0.07 repectively :o/ It can easily be seen on a black desktop background, which becomes dark-red once the lut is loaded by dispwin.

The profile doesn't currently correct very well for offsets in the channels. The
calibration does a much better job of this. Did you calibrate or
just profile ?

Do you have any ideas/experiences of the problem? Do you think the DTP92 can't be used for such calibration? Is it very different from the DTP94, which gives good resultat on projectors? Maybe there is a piece of hardware missing on the 92, like a IR filter or so?

I did have a play with calibrating and profiling a projector at one stage,
but had a limited amount of time, and had a bug that was upsetting the result.
I haven't tried that type of thing out again.

I would imagine there could be spectral differences between the projector
and a typical CRT, which is what the DTP92 is set for. A projector with
dichroic filters and LCD, or a filter wheel and DLP + typical projector
lamp probably won't have a CRT spectrum.

The other aspect is the geometry and stray light aspect. You need to
fill the aperture of the instrument with the screen somehow, either
getting it close (without self shadowing of course), or use some sort
of telescopic adapter. Stray light may be an issue too, since it will
make the low end appear to be unresponsive.

I built a telescopic adapter to try this stuff out, although I
haven't characterized it's performance to any great degree.
The lens I got hold of is acrylic, and needs a fair amount of
correction at the blue end, to counteract the UV stabilizers in
the plastic. The idea is to collect more light at a distance, while
narrowing the instrument acceptance angle. I would have thought
that the DTP92 was well filtered in terms of keeping stray wavelengths out.

Graeme Gill.

Other related posts: