Page 20 of 23
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Wed Jul 24, 2019 9:03 am
by Dade
epilectrolytics wrote: Wed Jul 24, 2019 8:32 am
But in LuxcoreUI I still get the "too bright caustics" problem:
I get this:
How do you get bright caustics ?
epilectrolytics wrote: Wed Jul 24, 2019 8:32 am
I wanted to provide a screenshot but LuxCoreUI crashed when I tried to open the render engine editor window.
I should have fixed this problem.
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Wed Jul 24, 2019 9:20 am
by epilectrolytics
Dade wrote: Wed Jul 24, 2019 9:03 am
I get this:
caust.jpg
How do you get bright caustics ?
It happens randomly and I found no way to reproduce it with certainty:
When switching around render engines and samplers in Luxcore sometimes caustics get too bright.
Mostly they work normally though like in the image you posted.
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Wed Jul 24, 2019 9:30 am
by Sharlybg
It happens randomly and I found no way to reproduce it with certainty:
When switching around render engines and samplers in Luxcore sometimes caustics get too bright.
Mostly they work normally though like in the image you posted.
isn't reinhard tone mapper involved somewhere ?
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Wed Jul 24, 2019 9:36 am
by epilectrolytics
Sharlybg wrote: Wed Jul 24, 2019 9:30 am
isn't reinhard tone mapper involved somewhere ?
Nope:
Code: Select all
film.imagepipelines.0.1.type = "TONEMAP_LINEAR"
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Mon Jul 29, 2019 8:23 am
by epilectrolytics
Dade wrote: Wed Jul 24, 2019 9:03 am
epilectrolytics wrote: Wed Jul 24, 2019 8:32 am
I wanted to provide a screenshot but LuxCoreUI crashed when I tried to open the render engine editor window.
I should have fixed this problem.
Yep, LightCPU now doesn't crash any more.
Playing around with the scene I notice that BiDir caustics are darker than Hybrid caustics (after ~500 samples).
And when I run Hybrid for longer (trying to find out when the SDS-paths converge) sooner or later overexposed caustics kick in (bottom pic, early example).
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Mon Jul 29, 2019 9:56 am
by FarbigeWelt
epilectrolytics wrote: Mon Jul 29, 2019 8:23 am
And when I run Hybrid for longer (trying to find out when the SDS-paths converge) sooner or later overexposed caustics kick in (bottom pic, early example).
Screenshot2.jpg
???
Your pictures compare pathcpu sobol and random. Sobol (270 sampkes) looks much brighter than random (512 samples). Who knows why?
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Mon Jul 29, 2019 10:48 am
by epilectrolytics
When light tracing is activated it does not show up in the render info, it will only display "PathCPU" or "PathOCL".
I have not noticed any difference between Sobol or Random sampler regarding the "overexposed caustics" problem but will test that.
These pictures are just examples to demonstrate the various differences in caustic brightness with the hybrid engine, which should show consistent brightness under all conditions IMHO.
Edit:
I just played around with different combinations for half an hour and everything worked normally until I changed hybrid partition from 0.9 to 0.8, resulting in too bright caustics. But after the next start all is well again
It is a really nasty thing because it can not be reproduced with certain settings.
I don't even know if it happens with Sobol only because whenever I switch samplers the statistics window disappears with reboot and I forget to reactivate it when making a screenshot.
Edit2:
Just caught it from Blender with Metro sampler.
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Mon Jul 29, 2019 1:00 pm
by epilectrolytics
Dade wrote: Wed Jul 24, 2019 9:03 am
How do you get bright caustics ?
Finally I could create a file with settings where the caustics always show the brightness issue, at least on my PC (Win 1903, LuxCore build from 7/27/2019).
In only works in Blender with Hybrid PathOCL (when switching to PathCPU I got normal behavior but not tested for longer rendering).
Caustics start looking normal but somewhere during the first 1000 samples brightness increases.
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Mon Jul 29, 2019 6:14 pm
by kintuX
epilectrolytics wrote: Mon Jul 29, 2019 1:00 pm
Dade wrote: Wed Jul 24, 2019 9:03 am
How do you get bright caustics ?
Finally I could create a file with settings where the caustics always show the brightness issue, at least on my PC (Win 1903, LuxCore build from 7/27/2019).
In only works in Blender with Hybrid PathOCL (when switching to PathCPU I got normal behavior but not tested for longer rendering).
Caustics start looking normal but somewhere during the first 1000 samples brightness increases.
testanim3.blend.zip
Seems there's even a difference between PT CPU vs PT OCL.
But in LuxCoreUI CPU looks brighter.
Re: Hybrid Back/Forward path tracing (aka BiDir without MIS)
Posted: Mon Jul 29, 2019 10:07 pm
by kintuX
epilectrolytics wrote: Mon Jul 29, 2019 1:00 pm
Dade wrote: Wed Jul 24, 2019 9:03 am
How do you get bright caustics ?
Finally I could create a file with settings where the caustics always show the brightness issue, at least on my PC (Win 1903, LuxCore build from 7/27/2019).
In only works in Blender with Hybrid PathOCL (when switching to PathCPU I got normal behavior but not tested for longer rendering).
Caustics start looking normal but somewhere during the first 1000 samples brightness increases.
testanim3.blend.zip
OK, found your bug
Rough glass with Roughness @ 0.007 
Read the forums sometime. Think, workout with your brain, get some fresh air and do something better with your life
Thanks
Dade! for pointing it out all around
testanim3.blend.zip