lllyasviel / lllyasviel/ControlNet

A model that reconstructs the same image. Is it a dumb idea?

Open
#131 6 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
34.1k
Forks
3k
PR merge metrics
No merged PRs in 30d

Description

First of all, I apologize if I'm asking nonsense. My doubt arises from the need to edit the same image, not a transformed one. Projects like img2img alternative test (automatic1111 script) have pointed to this for a while.
So... Can't you just make a model based on rebuilding the same image for inference? Maybe there's a simpler way and I'm asking nonsense, but the last few days I've seen the possibility of using that to edit videos using an input in img2img as modifier and compositor.
However it can be achieved, it would be an extremely powerful feature.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reviewing the existing ControlNet inference approach and the referenced img2img alternative test from the issue. Clarify whether reconstruction of the same image is technically feasible and define the expected image-editing or video-compositing behavior before identifying implementation and validation work.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.