# yerfor/Real3DPortrait

Real3D-Portrait: One-shot Realistic 3D Talking Portrait Synthesis; ICLR 2024 Spotlight; Official code

Repository: https://github.com/yerfor/Real3DPortrait
Canonical: https://ross.abutalabs.com/products/real3dportrait
Language: Python
License: MIT
License Family: permissive
Topics: nerf, one-shot, talking-face-generation
Last push: 2024-10-18T09:11:18+00:00

## Health v2 (maintenance only)
Score: 26/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 67
- inputs: {"age_days": 943, "days_push": 684, "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 1091, forks 131 (observed 2026-08-28T04:03:32.961706+00:00)

## What it is
Official PyTorch implementation of Real3D-Portrait, an ICLR 2024 Spotlight paper for one-shot realistic 3D talking portrait synthesis. It generates high-reality talking head videos from a single source image driven by audio, using NeRF-based plane representations and audio-to-motion models.

## Use cases
- generate a talking head video from a single photo and audio clip
- synthesize realistic 3D talking portraits one-shot
- animate a still portrait image with speech audio
- create talking avatar videos from one reference image
- research NeRF-based talking face generation
- train audio-to-motion and image-to-plane models for talking heads

## When to choose
- you need one-shot talking portrait synthesis from a single image
- you want high video realism in audio-driven talking heads
- you need a research baseline or codebase for 3D talking face papers
- you want to train or fine-tune the full pipeline with released training code

## When to avoid
- you need a production-ready, easy-to-deploy avatar API
- you lack a GPU or cannot set up a complex Conda environment with third-party models
- you only need lightweight 2D face animation without 3D reconstruction
- you want talking style control, for which MimicTalk is more suitable

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, image-processing, video-processing, speech-recognition, graphics
- domain: deep-learning, computer-vision, artificial-intelligence
- platform: python
- tags: talking-face-generation, nerf, one-shot, 3d-portrait, audio-driven, pytorch, research-code, iclr-2024, video, audio, linux, gpu

## Member repositories
- yerfor/Real3DPortrait (main) score 26

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:32.961706+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 2026-08-30T06:49:10.701302+00:00, confidence not recorded.
  - readme: https://github.com/yerfor/Real3DPortrait (fetched 2026-08-28T04:03:32.961706+00:00, sha 48d437ab514d)
- Data as of 2026-08-30T08:39:29.467469+00:00.