Created attachment 7684 [details] Simple 2-spot PDF If I'm interpreting these results correctly, TIFFSEP always seems to antialias against solid "black", even if the adjacent spot isn't solid. In this example, Spot #1 is a field of 50%. Spot #2 is a flower on top of it. When rasterized (result.s0.tif), the field of black is at 50%, except the pixels coincident with the flower, where some of the pixels are actually darker than 50%. This bug may be associated with Bug 687674 and possibly Bug 692303. gs -sDEVICE=tiffsep -dBATCH -dSAFER -dNOPAUSE -dGraphicsAlphaBits=4 -dTextAlphaBits=4 -r72 -sOutputFile="result.tif" AntiAliasTest.pdf This was tested against the master (configured with --with-libiconv=no).
Created attachment 7685 [details] 50% field spot output
This appears to be fixed in current code, possibly as a result of fixing some of the other problems referenced above.