mention xcalib as an alternative

[vn971]

Hi!
Sorry, this is not an issue in the pure sense of the word. Instead I want to mention an alternative, xcalib. Example usage:

xcalib -i -a

You may note, if you wish, this alternative in README. Initially, I found your project with the desire to invert colors by "value" (along with hue/saturation). Had to read the source code and install to make sure what your program does exactly.

[zoltanp]

Hi, I've added to the readme file xcalib as alternative program. Actually I've created this utility because xcalib has not been working on my dual-monitor setup: it would invert the colors of a single (random) monitor, but not both.

Should I close this issue?

[lkraav]

Yeah @zoltanp that's exactly where xcalib failed for me. Tyvm for creating xrandr-invert-colors.

When patching documentation, perhaps we should also include a note about things like redshift, that vary screen color temperature throughout the day. It will immediately reset whatever xcalib-invert-colors tries to do.

[zoltanp]

@lkraav Redshift is already in acknowledgements, because I've reused some code from it. If you have ideas for improving the documentation, proof-of-concept text, patches, pull requests are all welcome.

[vn971]

Ah, I didn't know about this and that's interesting to know. Thanks!

[vn971]

@zoltanp a question not related to this issue. Since you seem to understand the Xserver architecture in some depth, could you also know a way to invert colors by "value" (keep the color but invert "brightness")? I tried to explain what I want here:

https://superuser.com/questions/742338/invert-screen-colors-by-value

[zoltanp]

@vn971 I have only basic knowledge about the functioning of the Xserver; for the xrandr-invert-colors I've just copied/adapted what redshift does: modify the gamma look-up tables.

This is where the new gamma tables are generated:

https://github.com/zoltanp/xrandr-invert-colors/blob/master/xrandr-invert-colors.c#L197

This is where the new gamma tables are applied:

https://github.com/zoltanp/xrandr-invert-colors/blob/master/xrandr-invert-colors.c#L201

The gamma tables have a length, and they consist of a 3 vectors for R, G and B color components. The tables can be considered as a color-by-color pixel transformation applied to the whole image. Thus the value of one color cannot affect the value of a different color. I'm sceptical if the processing/filtering you would like to perform is possible with this constraint.

Maybe with a different API, arbitrary transformation/filtering is possible, but I don't know exactly how. Apparently compiz has (had?) a plugin that can perform arbitrary filtering of display colors:

http://askubuntu.com/questions/211617/how-can-i-recover-the-compiz-color-filter-plugin-in-ubuntu-12-10

http://askubuntu.com/questions/538896/compiz-plugins-filter-ordering

[vn971]

@zoltanp Thank you very much for your answer!

Yes, [unfortunately,] RandR can't apply color transformation matrices. They really seem to have independent color "function", which makes by-value inversion impossible. (I studied your code, too, as well as xcalib-s.) Sad to know that there is no straightforward work-around known.

Ok, anyway, thanks again for your answer and view on the subject!