ziqihuangg/ReVersion

• highreadme#1

  Reposition the README's opening to clarify the core task

  Why:

  CURRENT

  # ReVersion (SIGGRAPH Asia, 2024)
  
  [](https://ziqihuangg.github.io/papers/2024SigAsia-ReVersion.pdf)
  [](https://arxiv.org/abs/2303.13495)
  [](https://ziqihuangg.github.io/projects/reversion.html)
  [](https://www.youtube.com/watch?v=pkal3yjyyKQ)
  
  [](https://huggingface.co/spaces/Ziqi/ReVersion)
  
  This repository contains the implementation of the following paper:
  > **ReVersion: Diffusion-Based Relation Inversion from Images**<br>
  > Ziqi Huang<sup>∗</sup>, Tianxing Wu<sup>∗</sup>, Yuming Jiang, Kelvin C.K. Chan, Ziwei Liu<br>
  
  From MMLab@NTU affiliated with S-Lab, Nanyang Technological University
  
  ## :open_book: Overview
  
  We propose a new task, **Relation Inversion**: Given a few exemplar images, where a relation co-exists in every image, we aim to find a relation prompt **<R>** to capture this interaction, and apply the relation to new entities to synthesize new scenes. The above images are generated by our **ReVersion** framework.

  COPY-PASTE FIX

  Copy

  # ReVersion (SIGGRAPH Asia, 2024)
  
  **ReVersion introduces a novel diffusion-based framework for Relation Inversion, enabling the extraction and application of visual relationships from example images to synthesize new scenes.**
  
  [](https://ziqihuangg.github.io/papers/2024SigAsia-ReVersion.pdf)
  [](https://arxiv.org/abs/2303.13495)
  [](https://ziqihuangg.github.io/projects/reversion.html)
  [](https://www.youtube.com/watch?v=pkal3yjyyKQ)
  
  [](https://huggingface.co/spaces/Ziqi/ReVersion)
  
  This repository contains the implementation of the following paper:
  > **ReVersion: Diffusion-Based Relation Inversion from Images**<br>
  > Ziqi Huang<sup>∗</sup>, Tianxing Wu<sup>∗</sup>, Yuming Jiang, Kelvin C.K. Chan, Ziwei Liu<br>
  
  From MMLab@NTU affiliated with S-Lab, Nanyang Technological University
  
  ## :open_book: Overview
  
  We propose a new task, **Relation Inversion**: Given a few exemplar images, where a relation co-exists in every image, we aim to find a relation prompt **<R>** to capture this interaction, and apply the relation to new entities to synthesize new scenes. The above images are generated by our **ReVersion** framework.

  OpenDoneDismiss
• mediumtopics#2

  Add more specific topics related to the task and application

  Why:

  CURRENT

  aigc, diffusion-model, gen-ai, image-generation, relation-modeling, stable-diffusion

  COPY-PASTE FIX

  Copy

  aigc, diffusion-model, gen-ai, image-generation, relation-modeling, stable-diffusion, scene-generation, visual-relations, relation-learning, image-synthesis, generative-ai-for-vision

  OpenDoneDismiss
• lowlicense#3

  Clarify the project's license in the README

  Why:

  COPY-PASTE FIX

  Copy

  Add a section or line in the README, for example, under the 'Overview' or 'Updates' section, stating: 'This project is released under the terms specified in the [LICENSE](LICENSE) file. Please refer to the file for full details.'

  OpenDoneDismiss

Category GEO backends resolved for this scan: google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• CATEGORY QUERY

  How to extract and apply object relationships from example images for new scene generation?

  you: not recommended
  AI recommended (in order):

  1. Graph R-CNN
  2. Neural Motifs
  3. Scene Graph Transformer (SGT)
  4. PyTorch Geometric (PyG)
  5. NetworkX
  6. LayoutGAN
  7. LayoutVAE
  8. SG2IM (Scene Graph to Image Synthesis)
  9. GigaGAN
  10. StyleGAN

  AI recommended 10 alternatives but never named ziqihuangg/ReVersion. This is the gap to close.

  Show full AI answer
• CATEGORY QUERY

  Seeking a generative AI model to synthesize new images based on learned interactions from provided examples.

  you: not recommended
  AI recommended (in order):

  1. Stable Diffusion (Stability-AI/StableDiffusion)
  2. DreamStudio
  3. Midjourney
  4. DALL-E 3
  5. ChatGPT Plus
  6. Microsoft Copilot
  7. Adobe Firefly
  8. Imagen
  9. StyleGAN (NVlabs/stylegan3)

  AI recommended 9 alternatives but never named ziqihuangg/ReVersion. This is the gap to close.

  Show full AI answer

• Metadata completeness

  pass

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of ziqihuangg/ReVersion?

  pass
  ✓AI named ziqihuangg/ReVersion explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• If a team adopts ziqihuangg/ReVersion in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI named ziqihuangg/ReVersion explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• In one sentence, what problem does the repo ziqihuangg/ReVersion solve, and who is the primary audience?

  pass
  ✓AI named ziqihuangg/ReVersion explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?