Coming back to this issue as it appearantly won't be fixed upstream anytime soon, does the LOOP fix the issue? Or does it produce new issues? If it works out I could just temporarily set the default to LOOP for the debian package.
Lennart Am 11.07.2013 18:36, schrieb Schrober: > Package: libtxc-dxtn-s2tc0 > Severity: normal > Version: 0~git20121227-1 > Forwarded: https://github.com/divVerent/s2tc/issues/2 > X-Debbugs-CC: divver...@xonotic.org > > The default setting which enables color refinement causes edges towards alpha > to get a weird "broken pixel" effect. > > To reproduce: > > convert input.png input.tga > export S2TC_REFINE_COLORS=ALWAYS > s2tc_compress -i input.tga -o output.dds -t DXT5 > convert output.dds output.png > s2tc_decompress -i output.dds -o output.tga > > Input and output image are attached (you may need to scale it to see the 1 > pixel line problem on the bottom and right side). This problem doesn't happen > with libtxc_dxtn.so from an actual S3TC implementation like > http://cgit.freedesktop.org/~mareko/libtxc_dxtn/ and doesn't happen when > disabling S2TC_REFINE_COLORS by setting it to export > S2TC_REFINE_COLORS=NEVER. > It is important to understand that S2TC_REFINE_COLORS=ALWAYS is always > enabled > by default and the user has to deactivate it manually. Maybe it is better to > change it to S2TC_REFINE_COLORS=LOOP by default? > > This problem was introduced in one of those commits: > d7caf4ce1bd5df6c98a1a59a122a4a63d9916bd7 > 35e0718f1325a482cdf2795d054f3f49779747a6 > 38acfdc032ae17ac97bb547dfae89ad6160ee798 > b3f0fb2b88f700d5e32b3f7e2ac004c3f51121df > > I think 38acfdc032ae17ac97bb547dfae89ad6160ee798 ("alpha 0 pixels are always > important (think color filtering, unexpected blend functions)") caused the > problem because compiling it with the lap fix from > b3f0fb2b88f700d5e32b3f7e2ac004c3f51121df ("another LAB") was the first commit > causing the problem. The previous commit > d7caf4ce1bd5df6c98a1a59a122a4a63d9916bd7 ("fix file opening bug in > s2tc_decompress") together with b3f0fb2b88f700d5e32b3f7e2ac004c3f51121df > ("another LAB") did not have this problem -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org