Wow... still the same error, just like the azure builds on linux and windows too... wtf am I doing wrong here?
EDIT: I may have made an error with these latest 2 builds installation!!
EDIT2: Ok, so the very latest azure build and your just now I have indeed copied over to a wrong place in linux!! Idiot me... so your build definitely works, I'm gonna try the recent daily.
Wow... still the same error, just like the azure builds on linux and windows too... wtf am I doing wrong here?
EDIT: I may have made an error with these latest 2 builds installation!!
EDIT2: Ok, so the very latest azure build and your just now I have indeed copied over to a wrong place in linux!! Idiot me... so your build definitely works, I'm gonna try the recent daily.
So latest daily build doesn't work for me on linux
B.Y.O.B.'s build does work
EDIT: Also windows daily build has OIDN working now
Btw.
I've been "playing" with this all day lol... I canceled my corona license (after 4 years) and will use only luxcore probably.. Still some features missing so here and there cycles will have to help, but this denoiser is making work with luxcore super fast!
How's the bidir denoising investigation going on? I suspected image filtering, I saw ppl on github also talk about pixel filtering correlation. I always wondered why rendering bidir in luxcore is more blurry than pathtracing, something must be going on there...
lacilaci wrote: Tue Feb 05, 2019 6:28 pm
How's the bidir denoising investigation going on? I suspected image filtering, I saw ppl on github also talk about pixel filtering correlation. I always wondered why rendering bidir in luxcore is more blurry than pathtracing, something must be going on there...
Yes, BiDir pixel filtering is very different from Path one because the way BiDir works. If you check the GitHub discussion, we are investigating the problem. The latest discover is that Oidn is working fine with Mitsuba BiDir. I need to run some test to try to isolate the problem.
Support LuxCoreRender project with salts and bounties
lacilaci wrote: Tue Feb 05, 2019 6:28 pm
How's the bidir denoising investigation going on? I suspected image filtering, I saw ppl on github also talk about pixel filtering correlation. I always wondered why rendering bidir in luxcore is more blurry than pathtracing, something must be going on there...
Yes, BiDir pixel filtering is very different from Path one because the way BiDir works. If you check the GitHub discussion, we are investigating the problem. The latest discover is that Oidn is working fine with Mitsuba BiDir. I need to run some test to try to isolate the problem.
Yes I've seen some mentions about weighted sampling etc. I hope this can be resolved.
If you can solve this, you might also get ready to implement albedo and avg normal pass for bidir. And Simon could remove bcd from blender addon, or at least make it so that people don't enable it by accident(just a suggestion )
Hi! I'm the lead developer of OIDN, and I'm very excited to see that you're working on integrating it into LuxCore! I'm happy to help, so please let me know if you have any questions or concerns!
You're right that there should be a more detailed guideline for the albedo and normal AOVs, and there should be some example images for these as well. I'll try to add these ASAP (hopefully tomorrow) to the website and documentation.
voxelium wrote: Tue Feb 05, 2019 11:00 pm
Hi! I'm the lead developer of OIDN, and I'm very excited to see that you're working on integrating it into LuxCore! I'm happy to help, so please let me know if you have any questions or concerns!
You're right that there should be a more detailed guideline for the albedo and normal AOVs, and there should be some example images for these as well. I'll try to add these ASAP (hopefully tomorrow) to the website and documentation.
For the rest Oidn is working flawless with path tracing, we have only a couple of doubts about how some type of material should be handled on Albedo pass (for instance specular materials, or volumes, or opaque object, etc.).
Support LuxCoreRender project with salts and bounties
voxelium wrote: Tue Feb 05, 2019 11:00 pm
Hi! I'm the lead developer of OIDN, and I'm very excited to see that you're working on integrating it into LuxCore! I'm happy to help, so please let me know if you have any questions or concerns!
You're right that there should be a more detailed guideline for the albedo and normal AOVs, and there should be some example images for these as well. I'll try to add these ASAP (hopefully tomorrow) to the website and documentation.
Thanks a lot!
@B.Y.O.B. - Denoised pass doesn't have alpha in blender. My transparent background is black in the denoised pass, can be something done about that?