RenderKit/oidn

Health v2 · maintenance only

Adoption not part of the score

2089 stars · 198 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Intel Open Image Denoise is an open-source library of high-performance, deep-learning-based denoising filters for images rendered with ray tracing. It reduces Monte Carlo noise in path-traced renders, cutting required samples per pixel by orders of magnitude, and integrates via a simple C/C++ API.

Use cases

• denoise noisy path-traced renders with few samples per pixel
• integrate AI denoising into a custom ray tracing renderer
• use albedo and normal AOV buffers to preserve render detail
• denoise renders on CPU or GPU across Intel, NVIDIA, and AMD hardware
• train a custom denoising model for a specific renderer or scene type
• speed up final-frame rendering in production pipelines

When to choose

• you render with ray tracing or path tracing and need to cut sample counts
• you want an award-winning, permissively licensed (Apache-2.0) denoiser
• you need CPU and multi-vendor GPU support including Apple silicon
• you want pre-trained filters plus the option to train custom models

When to avoid

• your noise is not Monte Carlo rendering noise (e.g. photo denoising)
• you need a turnkey GUI tool rather than an embeddable library
• your platform lacks SSE4.1 or a supported GPU

Facets

library · maturity stable

image-processing machine-learning graphics graphics computer-vision deep-learning cpp cross-platform windows denoising ray-tracing monte-carlo rendering deep-learning-filters c-api game-development gpu linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="RenderKit/oidn")