LuxCoreRender Forums

Ondra Karlik (corona) on caustics: https://www.youtube.com/watch?v=OEU-Ksg9wls Quite funny as it is exactly what I have done (with the additional scenario of heterogeneous CPU/GPU workload). The only missing part is the SDS paths and, while they seem to have used vertex merging, I'm planning, at th...

[...] 1 c result evaluation after all threads finished generates Z new large mutations start conditions [...] Samples can have a very variable cost and, in your scenario, all threads will be "hostages" of (i.e. will have to wait) the one with the more time consuming work: the CPU load wil...

lacilaci wrote: ↑Wed Aug 28, 2019 11:52 am But pathguiding is for cpu only, right?

Yes (and, for instance, for indirect light, PhotonGI is a lot faster than any path guiding can ever be).

Exactly. The question is, why ate they more than available and render still run. If the stats are accurate, swap most likely. The performance would probably suffer pretty badly, though VRAM-RAM-swapping? Never observed with my old graphic cards. In much earlier LuxCoreRender versions not enough VRA...

acasta69 wrote: ↑Tue Aug 27, 2019 9:54 am I'll also need access to the BlendLuxCore repo to set up the pipeline files.

You (and all platform/OS maintainers) should have now access to BlendLuxCore repository.

Now, if I remember correctly those lines have something to do with how gpu samples stuff and they disappear after a while... Yes it is due to Sobol "quasi-random" nature and how the work is partitioned in OpenCL (i.e. in strips of 16 pixels). The patterns should disappear like normal nois...

The problem is quite easy to explain: the eye ray hits a front face however it doesn't detect the hit with the very near (on the edges) back face (due to numerical precision). The result is a volume segment between the front face and the far camera clipping plane instead of between the front and bac...

I don't know how Blender works but are you using an array of cubes instead of a single big cube


In this case, you are going to have a lot of problem with all the overlapped faces, (lack of) numerical precision will do a mess of your scene. Why aren't you using a single big cube ?

I have a fairly simple scene with volumetric caustics that I've been rendering at a fixed number of samples, and I thought I would try using the "noise level" halt condition, but I have no idea how to pick a sensible target value. I started with the default value of 5, and the render kept...

This is for large amount of emitters and other situations don't benefit from this? It is useful as soon as you have more than one light source and the light sources illuminated area doesn't overlap a lot. For instance, this is a rendering with 4 spot lights and default light strategy, 8 samples/pix...