bmild/nerf

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

10927 stars · 1430 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

The official TensorFlow implementation of NeRF (Neural Radiance Fields), the ECCV 2020 paper representing scenes as neural radiance fields for novel view synthesis. It trains a small fully-connected network to map 5D coordinates to color and opacity, enabling differentiable volume rendering of new views of a single scene.

Use cases

• render novel views of a scene from a sparse set of photos
• reproduce the NeRF paper results
• train a neural radiance field on custom real or synthetic images
• learn how NeRF works with the tiny_nerf Colab notebook
• research baseline for neural scene representation experiments

When to choose

• you want the original reference implementation of NeRF for research or education
• you have a single GPU and can tolerate hours-to-days of training per scene
• you need the exact code behind the ECCV 2020 paper

When to avoid

• you need fast, production-quality NeRF training - modern implementations (Instant-NGP, Nerfstudio) are far faster
• you need TensorFlow 2.x or PyTorch - this code targets TensorFlow 1.15
• you want real-time rendering or large multi-scene training

Facets

library · maturity maintenance

machine-learning deep-learning graphics simulation computer-vision graphics deep-learning machine-learning python nerf neural-radiance-fields view-synthesis tensorflow volume-rendering 3d-reconstruction research-code gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="bmild/nerf")