[Bug 1133893] New: Incorrect colors being displayed by "display" for a 10-bit LUT (it seems)
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 Bug ID: 1133893 Summary: Incorrect colors being displayed by "display" for a 10-bit LUT (it seems) Classification: openSUSE Product: openSUSE Distribution Version: Leap 15.0 Hardware: Other OS: SUSE Other Status: NEW Severity: Major Priority: P5 - None Component: X11 Applications Assignee: bnc-team-screening@forge.provo.novell.com Reporter: Ulrich.Windl@rz.uni-regensburg.de QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- Created attachment 803995 --> http://bugzilla.opensuse.org/attachment.cgi?id=803995&action=edit Windows-Shot of incorrect display When I use ImageMagick-7.0.7.34-lp150.2.26.1.x86_64 to display an image, the colors look significantly wrong, and the image is partly transparent. However when I open the same image on the same display with GIMP or Firefox, the image looks correct. The only special thing seems to be my "HDR" display featuring a 10-bit lookup-table (LUT) for my ATI Radeon: "ATI Baffin [Radeon RX 460/560D / Pro 450/455/460/555/555X/560/560X]" Driver: "amdgpu" The visual probably is like this: visual: visual id: 0x5af class: TrueColor depth: 32 planes available colormap entries: 1024 per subfield red, green, blue masks: 0x3ff00000, 0xffc00, 0x3ff significant bits in color specification: 10 bits -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #803995|Windows-Shot of incorrect |Window-Shot of incorrect description|display |display -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c1 --- Comment #1 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- Created attachment 803996 --> http://bugzilla.opensuse.org/attachment.cgi?id=803996&action=edit Correct display of the image in Firefox (60.6.1) Some additional message that may be helpful for understanding the problem: (II) AMDGPU(0): Pixel depth = 24 bits stored in 4 bytes (32 bpp pixmaps) (II) AMDGPU(0): Using 8 bits per RGB (8 bit DAC) (--) Depth 24 pixmap format is 32 bpp (--) AMDGPU(0): Chipset: "Radeon RX 560 Series" (ChipID = 0x67ef) (there are no explicit messages referring to a 10 bit LUT or 1024 LUT entries) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c2 --- Comment #2 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- Created attachment 803997 --> http://bugzilla.opensuse.org/attachment.cgi?id=803997&action=edit Partial screenshot form Wikipedia (DE): Example where Firefox gets some colors wrong, too Maybe I should add that for some web pages, colors even look wrong in Firefox, but I could not find a condition that triggers it. The effect is that orange-like tones are mapped to greenish-tones, resulting in green faces, sometimes (see Julian Assange). While "display" never gets the colors right, Firefox mostly gets the colors right (see Amrita Sher-Gils). -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Summary|Incorrect colors being |Radeon RX 560: Incorrect |displayed by "display" for |colors being displayed by |a 10-bit LUT (it seems) |"display" for a 10-bit LUT | |(it seems) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c4 --- Comment #4 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- (In reply to Petr Gajdos from comment #3)
Reading comment 2 it does not seem to be ImageMagick specific. I may help, just let me know how.
In ImageMagick the effect is 100% reproducible, while I'm unsure what actually triggers the effect in FireFox (Before filing this bug I had requested some enhancement for Firefox that might help understanding the problem; see https://bugzilla.mozilla.org/show_bug.cgi?id=1535860). Most prominent are green faces in Firefox, but not all faces have that effect -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c11 --- Comment #11 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- With Mesa-18.3.2-lp151.22.4.x86_64 the problem is still there, but not all apps are affected. I suspect some older applications cannot handle "pixmap format is 32 bpp" correctly. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c13 --- Comment #13 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- Mesa-18.3.2 seems to be the most current update. So where should Mesa 19 come from? Also: Isn't Mesa used for 3D graphics only? 3D (OpenGL) graphics doe not show color problems, but 2D graphics do (in some applications, not all). -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c14 Stefan Dirsch <sndirsch@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(Ulrich.Windl@rz.u | |ni-regensburg.de) | --- Comment #14 from Stefan Dirsch <sndirsch@suse.com> --- Ouch. This is still Leap 15.0. My fault. There is no Mesa 19 available and planned for Leap 15.0. Mesa driver is also used for 2D acceleration via GLAMOR. But things are different than in boo#1113211 (comment #5), where VAAPI drivers are coming from Mesa-libva package ... -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c15 --- Comment #15 from Stefan Dirsch <sndirsch@suse.com> --- Ok. Meanwhile I believe it's just the software/libraries not working correctly here. In any case I suggest testing your 30bit monitor also with a machine running Tumbleweed. Things may have improved a lot since Leap 15.0. And Leap 15.1 won't be much better I'm afraid. Mainly only hardware support has been updated on Leap 15.1. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c21 Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(Ulrich.Windl@rz.u | |ni-regensburg.de) | --- Comment #21 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- (In reply to Stefan Dirsch from comment #20)
Not sure, what you mean with monitor profile. Could you explain, please? I haven't seen you mentioning this in Firefox bugreport.
Maybe see attachment 9083126 in https://bugzilla.mozilla.org/show_bug.cgi?id=1564585 -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c22 Stefan Dirsch <sndirsch@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags| |needinfo?(Ulrich.Windl@rz.u | |ni-regensburg.de) --- Comment #22 from Stefan Dirsch <sndirsch@suse.com> --- Ok. Seems there are ICC monitor profiles one can copy from Windows/MacOS and apply with xiccd or the appropriate tool on KDE/GNOME https://wiki.archlinux.org/index.php/ICC_profiles Never heard about this before. I suggest to try also without applying this profile in order to figure out if the ICC profile is the issue here. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c23 Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(Ulrich.Windl@rz.u | |ni-regensburg.de) | --- Comment #23 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- (In reply to Stefan Dirsch from comment #22)
Never heard about this before. I suggest to try also without applying this profile in order to figure out if the ICC profile is the issue here.
Would this answer your question:? (from comment https://bugzilla.mozilla.org/show_bug.cgi?id=1564585#c13)
Made the following test (meanwhile Firefox ESR 68.1.0): Disable monitor profile, load the test page: Image looks OK. Re-enable profile while Firefox is active, then reload the page: Image still looks OK Restart Firefox, then load the test page: Image look wrong (green face)
-- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 Stefan Dirsch <sndirsch@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Summary|Radeon RX 560: Incorrect |Radeon RX 560 with applied |colors being displayed by |Monitor ICC profile: |"display" for a 10-bit LUT |Incorrect colors being |(it seems) |displayed by "display" for | |a 10-bit LUT (it seems) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c24 Stefan Dirsch <sndirsch@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags| |needinfo?(Ulrich.Windl@rz.u | |ni-regensburg.de) --- Comment #24 from Stefan Dirsch <sndirsch@suse.com> --- Yes, may I ask how you load the ICC profile? I wouldn't rule out a bug in this loader. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c25 Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(Ulrich.Windl@rz.u | |ni-regensburg.de) | --- Comment #25 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- (In reply to Stefan Dirsch from comment #24)
Yes, may I ask how you load the ICC profile? I wouldn't rule out a bug in this loader.
Actually I'm unsure: Using GNOME, I guess it's colord loading the profile(s). -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1133893 http://bugzilla.opensuse.org/show_bug.cgi?id=1133893#c26 Stefan Dirsch <sndirsch@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sndirsch@suse.com Component|X.Org |GNOME Assignee|gfx-bugs@suse.de |gnome-bugs@suse.de QA Contact|gfx-bugs@suse.de |qa-bugs@suse.de --- Comment #26 from Stefan Dirsch <sndirsch@suse.com> --- Hmm. You know how to disable the ICC monitor profile, but you don't know, who enables/loads it? Ok. Let our GNOME developers have a look into this. I suggest to have a deeper look into https://wiki.archlinux.org/index.php/ICC_profiles in order to understand this better, e.g. run colord, xiccd, xcalib manually in a simple desktop like icewem, which doesn't support automatic? loading of ICC monitor profiles. -- You are receiving this mail because: You are on the CC list for the bug.
participants (2)
-
bugzilla_noreply＠novell.com -
bugzilla_noreply＠suse.com