# omerbt/MultiDiffusion

Official Pytorch Implementation for "MultiDiffusion: Fusing Diffusion Paths for Controlled Image Generation" presenting "MultiDiffusion" (ICML 2023)

Repository: https://github.com/omerbt/MultiDiffusion
Canonical: https://ross.abutalabs.com/products/multidiffusion
Homepage: https://multidiffusion.github.io/
Language: Jupyter Notebook
License Family: other
Topics: diffusion-models, generative-model, image-generation, stable-diffusion, text-to-image, multidiffusion, icml
Last push: 2023-09-21T12:27:10+00:00

## Health v2 (maintenance only)
Score: 31/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 93
- inputs: {"age_days": 1312, "days_push": 1077, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1066, forks 63 (observed 2026-08-28T04:03:27.033620+00:00)

## What it is
Official PyTorch implementation of MultiDiffusion (ICML 2023), a training-free framework that fuses multiple diffusion paths over a pre-trained Stable Diffusion model for controllable image generation. It supports applications such as text-to-panorama synthesis at wide aspect ratios and spatially guided generation using segmentation masks or bounding boxes, and is integrated as a pipeline in HuggingFace diffusers.

## Use cases
- generate panorama images from text prompts
- control text-to-image generation with segmentation masks
- generate images with a desired aspect ratio without retraining
- guide stable diffusion output with bounding boxes
- apply spatially grounded edits to a pre-trained diffusion model
- reproduce the MultiDiffusion paper results

## When to choose
- You want training-free, spatially controllable generation on top of an existing Stable Diffusion checkpoint
- You need panorama or mask/bounding-box guided synthesis as described in the paper
- You are doing research on controllable diffusion and want the reference implementation

## When to avoid
- You need a production-supported, licensed dependency - the repo has no license and research-grade code
- You only need standard text-to-image inference without spatial controls, where plain diffusers is simpler
- You expect active maintenance or bug-fix releases; the project has seen no updates since 2023

## Facets
- artifact type: library
- maturity: maintenance
- function: stable-diffusion, machine-learning, deep-learning
- domain: artificial-intelligence, machine-learning, deep-learning, image-processing
- platform: python, cross-platform
- tags: stable-diffusion, text-to-image, diffusion-models, panorama-generation, controllable-generation, training-free, research-code, icml-2023, image-generation, spatial-guidance, gpu

## Member repositories
- omerbt/MultiDiffusion (main) score 31

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:27.033620+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 2026-08-30T06:55:26.275765+00:00, confidence not recorded.
  - readme: https://github.com/omerbt/MultiDiffusion (fetched 2026-08-28T04:03:27.033620+00:00, sha b35dbbc8e8b5)
  - homepage: https://multidiffusion.github.io/ (fetched 2026-08-29T12:57:30.745788+00:00, sha bee2f49e54ee)
- Data as of 2026-08-30T08:39:29.467469+00:00.