Hi everyone, first sorry for being vague but the project I’m on is restrictive from NDA. I was wondering if I could get some help with getting setup with Dynamic Global Illumination. I followed a few tutorials and can’t seem to get any results from them. Most of which were using an older version of Unreal Engine; (I’ve also read through the Unreal Engine Documentation).
I played around with the Light Propagation Volume, and other settings; both in the light and Post Process settings. I’m also perplexed on how too activate Emissive Lighting. There’s a certain lighting setup I need that requires me to use a rectangular plane model, and use that for a light source. I’m using 4.12 and “r.LightPropagationVolume” doesn’t seem to be in the “ConsoleVariables.ini” file, (as most people go there to activate this in older versions).
And it only supports directional light, and for emissive you will need to enable the option in the material to have it emit light for LPV.
Note that LPV is not very good, performance is bad and quality is bad, it will not give as good of results as Enlighten, though it’s actually dynamic whereas Enlighten requires a large amount of preprocessing and doesn’t work with dynamic objects.
VXGI might work more like what you’d expect from a dynamic GI . The thing is you’ll need to have an NVIDIA GPU (EDIT: nope) and be on NVIDIAs branch of Unreal Engine, which might or might not work for you given your hardware, deadline, project technical constraints as well as your experience on building software from source.
Actually no you don’t, at least not the UE4 version. VXGI runs on Direct Compute rather than CUDA. It will probably run better on a Nvidia card since it is their tech, but it will also work on AMD without any modifications.
Oh wow, good to know! Was under that impression, I’ve only had NV cards for a while now so didn’t bother to check. Curious as to what the perf difference is between AMD/NV. None? 1:2? 1:5?
I only have a 980 so I am not sure on the exact numbers. Technically there shouldn’t be any difference as both cards support Direct Compute, so if there is a difference it shouldn’t be very much (10% max imo).
Don’t quote me on that, only a guess, but DC is DC so it shouldn’t matter. We had a Nvidia dev answering questions & implementing it all over in this thread if you want to read more (includes other Nvidia specific tech like Flex, WaveWorks, Hairworks, etc as well):
Thanks you for the insightful information. I’m weary to ask for this setup on a test project for R&D (due to time constraints, and System Admins/Developers calling the shots), but that doesn’t stop me from a little R&D at home on the weekends, no?
Thanks for the clarification! as I have an AMD card at home (but Nvidia for work). It will be interesting to try it out on both cards.
