GaussianSplatting.jl

3D Gaussian Splatting for Real-Time Radiance Field Rendering in Julia

bicycle.mp4

Requirements

• Julia 1.10.
• AMD GPU (ROCm) or Nvidia (CUDA) capable machine.

Usage

0. Install GaussianSplatting.jl package:

] add https://github.com/JuliaNeuralGraphics/GaussianSplatting.jl.git

• AMD GPU:

1. Add necessary packages: ] add AMDGPU

2. Run:

julia> using AMDGPU, GaussianSplatting

julia> GaussianSplatting.gui("path-to-colmap-dataset-directory"; scale=1)

• Nvidia GPU:

1. Add necessary packages: ] add CUDA, cuDNN

2. Run:

julia> using CUDA, cuDNN, GaussianSplatting

julia> GaussianSplatting.gui("path-to-colmap-dataset-directory"; scale=1)

GPU selection

• AMD GPU:

  julia> using Flux
  
  julia> Flux.gpu_backend!("AMDGPU")

  Additionally, set soft & hard memory limit to help Julia GC:

  julia> AMDGPU.soft_memory_limit!("80 %")
  
  julia> AMDGPU.hard_memory_limit!("80 %")

• Nvidia GPU:

  julia> using Flux
  
  julia> Flux.gpu_backend!("CUDA")

Datasets

Download one of the reference datasets from the MIP-NeRF-360: https://jonbarron.info/mipnerf360/