New texture: Remap (bring a value from one range into another range)
Three new textures for accessing the object ID: raw object ID (range 0 to 0xfffffffe), normalized object ID (range 0 to 1), object ID interpreted as color
LuxCore bug fixes:
Fixed a bug in OpenCL Bloom plugin
Fixed clamping of denoiser output, which was happening in some scene setups
The Linux binaries don't use SSSE3 instructions anymore so they now work on some old CPU
Fixed a bug in OpenCL code for imagemap wrap black and white
Fixed a difference between OpenCL and C++ code in SchlickBSDF_CoatingSampleF() affecting all glossy-related materials
Fixed a problem in variance clamping with TILEPATH
Fixed a problem in Glossy material OpenCL code
BlendLuxCore new features:
Add option to use the Bidir engine for viewport rendering (useful in scenes with very complex light paths, where RT Path preview does not work well)
New option in the math node: Divide
New textures in the utils section: Remap (bring a value from one range into another range), Object ID (can be used to randomize a material across multiple objects)
found a bug - when using the bidirectional engine, enabling the samplecount AOV stops the sampler from doing anything - 0k samples/second, 0 samples, however it still shows rays/sample correctly. Not sure if this is a blender plugin specific bug, or a bug with luxcore itself.
oldladyfear wrote: ↑Tue Oct 30, 2018 3:39 pm
found a bug - when using the bidirectional engine, enabling the samplecount AOV stops the sampler from doing anything - 0k samples/second, 0 samples, however it still shows rays/sample correctly. Not sure if this is a blender plugin specific bug, or a bug with luxcore itself.
BiDir has support only for a very small set of AOVs and it doesn't includes SAMPLECOUNT AOV. It sounds a problem inside the scope of Blender plugin.
Support LuxCoreRender project with salts and bounties
So how's the CPU usage for people using windows and beta1 with opencl? I'm still getting only 15-20% on cpu when rendering with opencl(alphav4 goes to 100%)
EDIT:
Also a small question, with beta1 in a relatively complex night scene I get:
CPU: 280K rays/sec, ~6.5 rays/sample, 780 samples in 8 minutes (or something very close to those numbers)
OpenCL(no cpu): 290K rays/sec, ~6.5 rays/sample, 860 samples in 8 minutes
with alpha v4 I thought since opencl works can use cpu I would get double(or near the performance) but I got
OpenCL(with cpu): 420K rays/sec, ~6.5 rays/sample, 1120 samples in 8 minutes (so far from double the performance)
I also noticed that it takes quite a bit of time to get to 400+ rays/sec, I probably would get more eventually but it was accelerating very very slow.
lacilaci wrote: ↑Wed Oct 31, 2018 7:50 am
So how's the CPU usage for people using windows and beta1 with opencl? I'm still getting only 15-20% on cpu when rendering with opencl(alphav4 goes to 100%)
Same here on Linux Mint....with opencl hybrid....cpu usage goes to max 21%.