Created attachment 144839 [details] Video demonstrating bug Smudge engine paints hard black squares on a grayscale image surrounding the layer's alpha. STEPS TO REPRODUCE 1. create/convert image to grayscale using default color profile OR convert a layer's color profile to grayscale 2. paint with any brush that smudges on the grayscale image or layer. OBSERVED RESULT Big black squares surrounding the brush EXPECTED RESULT Smudging as seen in an RGB image SOFTWARE/OS VERSIONS Windows: macOS: Linux/KDE Plasma: (available in About System) KDE Plasma Version: KDE Frameworks Version: Qt Version: ADDITIONAL INFORMATION
Hi, John! Thank you for your report. I'll take a look at this bug.
The bug is somehow related to the conversion algorithm graya8 -> graya16. For some reason, the alpha is set to 100% during this conversion. It might be that the conversion takes the path via alpha color space :(
A possibly relevant merge request was started @ https://invent.kde.org/graphics/krita/-/merge_requests/1386
Git commit 5b3b1de7fcbc033b1bf9dfbcd20a0dd53a5a98a3 by Dmitry Kazakov. Committed on 28/03/2022 at 06:19. Pushed by dkazakov into branch 'master'. Fix conversions between GRAYA8<->GRAYA16 The color conversion system had a bug, which caused the conversion to happen via ALPHA8 color space (GRAYA8<->ALPHA16<->GRAYA16). That basically dropped the alpha channel of the GRAYA color space. It was caused by the presence of custom connections in the alpha color space. The patch changes the way how connection path is searched. It implements NodeCapabilities structure, which tracks capabilities of the nodes used in the connection path. M +0 -32 libs/pigment/KoColorConversionAlphaTransformation.cpp M +0 -4 libs/pigment/KoColorConversionAlphaTransformation.h M +13 -27 libs/pigment/KoColorConversionSystem.cpp M +0 -10 libs/pigment/KoColorConversionSystem.h M +73 -50 libs/pigment/KoColorConversionSystem_p.h M +1 -11 libs/pigment/KoColorConversionTransformationFactory.h M +0 -8 libs/pigment/KoCopyColorConversionTransformation.cpp M +0 -2 libs/pigment/KoCopyColorConversionTransformation.h M +103 -20 libs/pigment/tests/TestColorConversionSystem.cpp M +8 -0 libs/pigment/tests/TestColorConversionSystem.h M +4 -0 plugins/color/lcms2engine/LcmsRGBP2020PQColorSpace.h M +0 -30 plugins/color/lcms2engine/LcmsRGBP2020PQColorSpaceTransformation.h https://invent.kde.org/graphics/krita/commit/5b3b1de7fcbc033b1bf9dfbcd20a0dd53a5a98a3
Git commit 83a4e3010777f5c1c2b7ae385a0c2f096ae6a676 by Dmitry Kazakov. Committed on 31/03/2022 at 13:43. Pushed by dkazakov into branch 'krita/5.0'. Fix conversions between GRAYA8<->GRAYA16 The color conversion system had a bug, which caused the conversion to happen via ALPHA8 color space (GRAYA8<->ALPHA16<->GRAYA16). That basically dropped the alpha channel of the GRAYA color space. It was caused by the presence of custom connections in the alpha color space. The patch changes the way how connection path is searched. It implements NodeCapabilities structure, which tracks capabilities of the nodes used in the connection path. M +0 -32 libs/pigment/KoColorConversionAlphaTransformation.cpp M +0 -4 libs/pigment/KoColorConversionAlphaTransformation.h M +13 -27 libs/pigment/KoColorConversionSystem.cpp M +0 -10 libs/pigment/KoColorConversionSystem.h M +73 -50 libs/pigment/KoColorConversionSystem_p.h M +1 -11 libs/pigment/KoColorConversionTransformationFactory.h M +0 -8 libs/pigment/KoCopyColorConversionTransformation.cpp M +0 -2 libs/pigment/KoCopyColorConversionTransformation.h M +103 -20 libs/pigment/tests/TestColorConversionSystem.cpp M +8 -0 libs/pigment/tests/TestColorConversionSystem.h M +4 -0 plugins/color/lcms2engine/LcmsRGBP2020PQColorSpace.h M +0 -30 plugins/color/lcms2engine/LcmsRGBP2020PQColorSpaceTransformation.h https://invent.kde.org/graphics/krita/commit/83a4e3010777f5c1c2b7ae385a0c2f096ae6a676