-
Supports all 4 standard lights (point, spot, directional and sky)
- However stationary skylight is not supported so if you have one in your scene it will still be baked since it is treated as a static one
- Advanced lighting features such as soft shadows (SourceLength and SourceRadius) and IES are not respected when calculating indirect lighting. However you may still see some effect of them since their direct lighting is calculated on CPU
- According to some tests the direct lighting is low quality sometimes because Lightmass quality level is set to “Preview”. While this problem is actually irrelevant to GPULightmass, you’ll probably want to set the quality level to “Production” to avoid such kind of mysterious problems.
-
Supports baking of standard surface lightmaps, volumetric lightmaps (also with faster voxelization), and sparse volume lighting samples
-
Lightmass parameters and quality settings specified in the editor are ignored except for Num Indirect Lighting Bounces
- Current the number of samples (quality settings) is hardcoded in the program and you cannot change it since I haven’t found an reliable way to expose them to the user
- AO Mask generation is not supported
-
GPULightmass should work with Swarm distributed rendering naturally as long as you’ve correctly set up the environments on all machines. (esp. a sufficiently new NVIDIA driver)
-
GPULightmass uses Brute Force as its Primary GI Engine and some form of radiosity caching as its Secondary GI Engine, which is very similar to vray.
- Brute force is much more accurate than Irradiance Caching (Lightmass’s primary GI engine) but it is also slower. It is also more sensitive to lightmap resolution since it actually calculates each lightmap texel, while irradiance caching is more or less resolution independent. You may want to tune down your lightmap resolutions firstly then slowly tune them up when working with GPULightmass.
[/quote]
- Brute force is much more accurate than Irradiance Caching (Lightmass’s primary GI engine) but it is also slower. It is also more sensitive to lightmap resolution since it actually calculates each lightmap texel, while irradiance caching is more or less resolution independent. You may want to tune down your lightmap resolutions firstly then slowly tune them up when working with GPULightmass.
Hi @aydanyr, This is taken from the first post in this thread, nothing has really changed. I don’t really understand a lot of these statements. Information is sparse and takes a lot of work to read the CUDA code.
There is only sparse comments in the code and the documentation is for Epic Lightmass. Much of these settings as described do not apply to LGPU.
My skill is in converting the LGPU code to support new versions of Unreal Engine 5.
Although you say rectangular lights work they are not in code. Everyone on this thread uses work around described earlier and it wont be added as its need an array using up much need VRAM spaces most of the time this wont be used so emissive lights as described earlier .