PhotonGI caustic cache re-factoring
Re: PhotonGI caustic cache re-factoring
Light tracing (or the cache) is used only "under" the sphere because the caustic is only there, over the sphere, normal path tracing will be used and your light source is very hard to find for a path tracer: just render with normal path tracing to have an idea.provisory wrote: Sun Sep 29, 2019 10:27 am Are you sure that caustic cache works with volumes?
It didn't seem to me in my scene, so I made a simple test case.
BiDir + metro:
CCvsVol-bidir.jpg
PT Sobol + Light Tracing:
CCvsVol-lt.jpg
PT Sobol + Caustic Cache:
CCvsVol-cc.jpg
About the cache, the problem is here:
Code: Select all
[LuxCore][0.613] PhotonGI visibility lookup radius: 0.0169757
[LuxCore][0.613] SceneVisibility trace thread count: 12
[LuxCore][2.121] SceneVisibility hit rate: 0.2408%
[LuxCore][2.153] SceneVisibility total entries: 1063240

The cache can cover only a finite amount of space, try to enclose your scene in a box or to use a box volume

Re: PhotonGI caustic cache re-factoring
Product design showcase might often be happening in a very open scene, can't we have bounced photons killed once they exit scene bounding box? Wouldn't this help focus cache where it needs to happen?Dade wrote: Sun Sep 29, 2019 1:35 pm
The scene visibility has an hit rate of 0.2408% but this isn't really surprising: you are trying to cover an infinite 3D space with cache entries. Your volume is the world default and you have an open scene![]()
The cache can cover only a finite amount of space, try to enclose your scene in a box or to use a box volume![]()
Re: PhotonGI caustic cache re-factoring
The above scene bounding box is really small (i.e. it would truncate the visible volume scattering under the sphere, with no volume caustic at all). However this is a quite corner case: the use of default scene volume imply a infinite scene size, this is somewhat incompatible with using a finite (volume) cache.lacilaci wrote: Sun Sep 29, 2019 1:52 pm Product design showcase might often be happening in a very open scene, can't we have bounced photons killed once they exit scene bounding box? Wouldn't this help focus cache where it needs to happen?
Re: PhotonGI caustic cache re-factoring
Hi,
I'm new to Luxcore and just started testing the last build.
What settings should I modify to get light passing through water plane when the camera is below it?
I'm new to Luxcore and just started testing the last build.
What settings should I modify to get light passing through water plane when the camera is below it?

Re: PhotonGI caustic cache re-factoring
You have to set the camera volume to "water" otherwise the ray will start with default scene volume (I assume it is "air"). In general, you have to be careful setting up the object volumes or you can end with some odd result.evacnoc wrote: Sun Sep 29, 2019 6:49 pm What settings should I modify to get light passing through water plane when the camera is below it?![]()
Re: PhotonGI caustic cache re-factoring
I have added the option to use the caustic cache only for SDS paths:
And this is a rendering with Hybrid forward/backward path tracing and caustic cache used only for SDS paths:
This is some serious beauty
Code: Select all
path.photongi.caustic.useonlyforsds = 1
This is some serious beauty

Re: PhotonGI caustic cache re-factoring
Oh, yes...Dade wrote: Mon Sep 30, 2019 8:46 am I have added the option to use the caustic cache only for SDS paths:
And this is a rendering with Hybrid forward/backward path tracing and caustic cache used only for SDS paths:Code: Select all
path.photongi.caustic.useonlyforsds = 1
sds.jpg
This is some serious beauty![]()
Can I assume that since cache is sds only it will be faster?
Re: PhotonGI caustic cache re-factoring
What an achievement you have done so farThis is some serious beauty

This need to be benchmarked.Can I assume that since cache is sds only it will be faster?
Re: PhotonGI caustic cache re-factoring
What do you mean ? Faster compared to ? The cache look up is done only for SDS paths so there are a lot less cache look up (I assume you mean this). However the cache creation time (and update) is exactly the same (a caustic is a caustic and I can not know if it will be seen trough a specular surface or not).lacilaci wrote: Mon Sep 30, 2019 8:51 am Can I assume that since cache is sds only it will be faster?
This is a v2.3 manifesto:
If you remove glass dispersion from the first one, you have a Cycles rendering
