Comment 8 for bug 953221

I tried it again with something that contained colour (yesterday's test was nearly a black-and-white sheet) and now I see that the large diagonal stripes also have wrong colours. The first triangle at the top has correct colors, but as the sheared pixels wrap around, they seem to have suffered large hue shifts, which is probably the result of making each pixel out of two different neighbouring pixels while confusing R,G,B channels.

I also just tried the «scanimage» command (from sane-utils) for the first time and it has the same problem.