Looks like regression from 5.2.1 from unknown bug with which Bug 77525 is also resolved. Bibisect could be useful. Tested in Windows.
Steps to Reproduce:
1. open attachment 113502 [details] from Bug 89473
Images shown as black squares.
Images shown normally.
User Profile Reset:
1. Right click on the image and select "Compress Graphic"
2. In the Compress Graphic dialogue click "Calculate"
Should this be done automatically on fileopen?
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:38.0) Gecko/20100101 Firefox/38.0
Confirmed when OpenGL is enabled
Build ID: f4ca1573fcf445164c068c1046ab5d084e1b005f
CPU Threads: 4; OS Version: Linux 4.2; UI Render: GL; VCL: gtk2;
Locale: ca-ES (ca_ES.UTF-8); Calc: group
I can reproduce it in
Build ID: de093506bcdc5fafd9023ee680b8c60e3e0645d7
Created attachment 128741 [details]
Timur - Compare with LO versions
Created attachment 128742 [details]
Timur - OpenGL report
Can the difference in version stem from OpenGL? I attach mine.
Contrary to bug 77525, this was problem from 5.2.1. up to master 5.3, but then recently started to work again.
Master tested with:
TinderBox: Win-x86@42, Branch:master, Time: 2016-10-26 (works OK)
Build ID: bbd44f8f89839b5abb4ec6c7ea195431de5b2f48
TinderBox: Win-x86@39, Branch:master, Time: 2016-11-15 (problem again)
Build ID: 074f0ab1d76f16fe92493868e2f2de75e67792ef
Since flopping, I raise to "high". In bug 77525 I asked for bibisectRequest and that should reveal both.
After driver reinstallation for AMD Radeon 6450, I have OpenGL blacklisted and I can't reproduce this, only if I force OpenGL.
Attachment 113502 [details] (from bug 89473) is a MS Word .doc file that contains a set of diagrams as PNG images.
Extracting the images, the ImageMagik "identify -verbose" shows them to be "Bilevel" with depth of 8/1-bit in sRGB color space--but with an alpha color, background color, and border color defined.
With OpenGL rendering enabled--inserting any of these odd PNGs to document canvas (Writer, Draw, Impress) renders each as a black rectangle. While processing with GPU enabled, or just with CPU, the image is well formed to canvas--and on print/export.
Removing regression as this has clearly been an issue with our OpenGL rendering in VCL when implemented for 4.4.0
Windows 10 Pro 64-bit en-US
nVidia GTX 750Ti w/driver 22.214.171.12470
Version: 126.96.36.199 (x64)
Build ID: 7074905676c47b82bbcfbea1aeefc84afe1c50e1
CPU Threads: 8; OS Version: Windows 6.19; UI Render: GL; Layout Engine: new;
Locale: en-US (en_US); Calc: group
Build ID: 2670ca3fc597decae78499d1397539668eb84e5e
CPU Threads: 8; OS Version: Windows 6.19; UI Render: GL;
TinderBox: Win-x86@42, Branch:master, Time: 2017-01-31_05:32:46
Locale: en-US (en_US); Calc: CL
At least easy to reproduce on Windows builds, extract the PNGs from attachement 113502 and insert into any new document. With OpenGL rendering enabled, result in the black rectangles for the images. With CPU, or GPU rendering--bitmaps are rendered to canvas no issue.
Not sure of behavior on Linux / macOS will check those a bit deeper.
Sidebar question though is how unusual/broken is this set of PNGs? They seem kind of scarce.
@Tomaž, could you have a look?
*** Bug 89473 has been marked as a duplicate of this bug. ***
Created attachment 131026 [details]
tweakpng details for one of the images from attachment 113502 [details]
The tweakpng utility is a helpful validator, and shows the PNGs we are having issues with are 8bit, paletted "bilevel" PNG.
As here with PNG images extracted from attachment 113502 [details] and also now duplicated with attachment 131025 [details] "bilevel" PNG from dupe bug 89473
Created attachment 131027 [details]
tweakpng details for one of the images from attachment 131025 [details]
tweakpng shows this problem image is also 8bit, paletted with two colors--so another "bilevel" PNG.
Hmmm... rendering PNGs is still tricky 11yrs later ;-) https://bz.apache.org/ooo/show_bug.cgi?id=60481
I can reproduce this with 6.1.something, but it works with with recent master, so apparently this has got fixed somewhen between 6.1 and 6.3. Closing.
I confirm that it's OK with 6.3+ while not with 6.2.2. WFM then.