In fb/fbcompose.c the implementation of convolution filters treats the convolution coefficients as integers instead of fixed-point values. The result is that convolution filters with fractional co-efficients are impossible to specify. I'll attach a patch that fixes the problem.
Created attachment 3285 [details] [review] Proposed patch to fix the problem This patch shifts the result of multiplying to convolution coefficients with the source component values in order to convert from a fixed-point representation.
Sorry about the phenomenal bug spam, guys. Adding xorg-team@ to the QA contact so bugs don't get lost in future.
This appears to be fixed by 4365d16c8629e824973ee0c357efdfbfd28b672c.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.