# liuyuan-pal/SyncDreamer

[ICLR 2024 Spotlight] SyncDreamer: Generating Multiview-consistent Images from a Single-view Image

Repository: https://github.com/liuyuan-pal/SyncDreamer
Canonical: https://ross.abutalabs.com/products/syncdreamer
Homepage: https://liuyuan-pal.github.io/SyncDreamer/
Language: Python
License: MIT
License Family: permissive
Topics: 3d-reconstruction, diffusion-models, generative-model, single-view-reconstruction
Last push: 2025-10-26T16:16:49+00:00

## Health v2 (maintenance only)
Score: 50/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 49, release rhythm 35, longevity 78
- inputs: {"age_days": 1093, "days_push": 311, "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 1045, forks 50 (observed 2026-08-28T04:03:21.369147+00:00)

## What it is
SyncDreamer is a synchronized multiview diffusion model that generates multiview-consistent images from a single-view image, released with inference, training, and evaluation code plus pretrained checkpoints. It enables downstream 3D reconstruction with NeuS or NeRF without SDS loss, supporting tasks like novel-view synthesis, text-to-3D, and image-to-3D.

## Use cases
- generate consistent views of an object from one photo
- reconstruct a 3D model from a single image
- turn sketches or paintings into 3D models
- novel view synthesis from a single image
- create multiview training data for NeRF or NeuS
- image-to-3D asset generation for games

## When to choose
- you need geometry- and color-consistent multiview images from one input image
- you want to feed generated views directly into NeRF/NeuS reconstruction without SDS optimization
- you want a research-grade model with training code and pretrained weights
- you have a GPU (10GB+ VRAM) for inference

## When to avoid
- you need real-time or CPU-only inference
- you want a polished end-user application rather than research code
- you cannot supply an elevation angle or foreground mask for the input image
- you need a pretrained model without elevation input, which is not yet released

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, image-processing, computer-vision
- domain: computer-vision, image-processing, graphics, artificial-intelligence
- platform: python
- tags: diffusion-models, novel-view-synthesis, multiview-generation, 3d-reconstruction, image-to-3d, research-code, gpu, linux

## Member repositories
- liuyuan-pal/SyncDreamer (main) score 50

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:21.369147+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 2026-08-30T07:02:06.257729+00:00, confidence not recorded.
  - readme: https://github.com/liuyuan-pal/SyncDreamer (fetched 2026-08-28T04:03:21.369147+00:00, sha d94fb2245e2c)
  - homepage: https://liuyuan-pal.github.io/SyncDreamer/ (fetched 2026-08-29T13:03:19.772774+00:00, sha 2ae4b7b859f3)
- Data as of 2026-08-30T08:39:29.467469+00:00.