Attempting to compress a texture using PVRTC2_4 crashes (while PVRTC2_2 and other formats pass.)
>./PVRTexToolCL -i test.png -o test.pvr -l -f PVRTC2_4 -q pvrtcbest
………*…*Assertion Failed:/usr/buildxl/buildroot/sdk/branch/UtilitiesSrc/PVRTexTool/Library/Source/Codecs/PVRTC/amtc/amtc_localised.c:8074:fabs(ModVariance) > 0.0
Need a debugger break on line /usr/buildxl/buildroot/sdk/branch/UtilitiesSrc/PVRTexTool/Library/Source/Codecs/PVRTC/amtc/amtc_localised.c:641
PVRTexToolCL: /usr/buildxl/buildroot/sdk/branch/UtilitiesSrc/PVRTexTool/Library/Source/Codecs/PVRTC/amtc/amtc_localised.c:8074: DetermineABUpdates: Assertion `fabs(ModVariance) > 0.0’ failed.
Aborted (core dumped)
I’ve tried the above with a PNG made of random noise and it worked, so there’s a specific problem compressing my texture (which is a spritesheet / atlas), which you can find here:
PVRTexToolCL version 3.40
Uses: PVRTexLib version 4.5 | JpegLib version 6b | Libpng version 1.5.12
i ve tried your sample and the tools effectively crash like you describe but Succeed using the GUI TOOLS
Since other format of compression are working, can it be a solution for you ?
I am using PVRTexTool GUI 4.4 on UBUNTU 32!!!
other format that i tried are taking age by command line but i think a solution is coming …i believe so
Thanks for the response, David. Unfortunately, I’m building our asset workflow, and I need the commandline to produce best quality results (slow is acceptable).
Also, the above settings crashesd the GUI for me. Steps are:
- Open test.png
- Click optimize transparent, bleed
- Click compress
- Choose “OpenGL ES 2” as the group
- Choose PVRTCII 4bpp
- Choose best quality
- Check generate mipmaps (no other checkboxes)
- Click encode
- Crash, terminal shows same output - DetermineABUpdates assertion failed
My workaround is to use PVRTC1_4 for now.
based on your pics, yes it does crash my GUI as well and command line , by selection an another png that i am using for my project it does nt not i am wondering about your setting of png creation then (nearest power of two etc etc)?
but sure a crash like this is not a normal behavior, only the Team can answer that one
Sorry for the delay in answering you - I’ve been investigating and we’ve managed to pin down the issue. I’ve added a temporary fix which will make it into REL3.2, but won’t be present beforehand. A permanent fix should hopefully be available before we actually release though, solving the issue properly. In the meantime, if you want I can provide you with an engineering drop that should solve the issue for now - if you email DevTech@imgtec.com, I can send you a copy.
Thank you Tobias. I’ll send you an email shortly.