# ShichenLiu/SoftRas

Project page of paper "Soft Rasterizer: A Differentiable Renderer for Image-based 3D Reasoning"

Repository: https://github.com/ShichenLiu/SoftRas
Canonical: https://ross.abutalabs.com/products/softras
Language: Python
License: MIT
License Family: permissive
Topics: differentiable-rendering, 3d-reconstruction, computer-graphics, geometry-processing
Last push: 2025-11-24T06:47:24+00:00

## Health v2 (maintenance only)
Score: 56/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 53, release rhythm 35, longevity 100
- inputs: {"age_days": 2710, "days_push": 282, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1298, forks 159 (observed 2026-08-28T04:04:17.287964+00:00)

## What it is
SoftRas is a PyTorch-based differentiable renderer that treats rasterization as a differentiable aggregating process over mesh triangles, enabling gradient back-propagation from images to mesh vertices and attributes. It supports rendering colorized meshes and training 3D reconstruction models from images without 3D supervision.

## Use cases
- render 3D meshes differentiably in PyTorch
- reconstruct 3D meshes from single RGB images without 3D supervision
- backpropagate image-based losses to mesh vertices and textures
- train single-view mesh reconstruction networks on multi-view image datasets
- generate soft, blurred, or transparent renderings of meshes for research

## When to choose
- you need gradients flowing from rendered images back to 3D mesh geometry or texture
- you are doing image-based 3D reconstruction or inverse graphics research
- you want a differentiable alternative to standard rasterizers in a PyTorch pipeline

## When to avoid
- you need a fast production renderer without gradient requirements
- you need support for modern PyTorch versions or recent CUDA toolkits out of the box
- you need features like ray tracing, global illumination, or complex material models

## Facets
- artifact type: library
- maturity: maintenance
- function: graphics, machine-learning, image-processing, simulation
- domain: computer-vision, graphics, deep-learning, machine-learning
- platform: python
- tags: differentiable-rendering, pytorch, 3d-reconstruction, rasterizer, cuda, iccv-2019, mesh-rendering, gpu, linux

## Member repositories
- ShichenLiu/SoftRas (main) score 56

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:17.287964+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:53:48.709205+00:00, confidence not recorded.
  - readme: https://github.com/ShichenLiu/SoftRas (fetched 2026-08-28T04:04:17.287964+00:00, sha 546b5e50cc1f)
- Data as of 2026-08-30T08:39:29.467469+00:00.