lllyasviel / lllyasviel/ControlNet

Character consistency

Open
#219 2 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

Spoiler alert: I'm a noob
ControlNet is the first real method to control SD. I was thinking if with this method there would be a way to teach the network to keep the generated character. I'll give you an example:
dog with blue fur => save subject => riding a motorcycle
I know there are already methods to teach the network about the subject, like dreambooth etc.
but I thought since the network already knows the subject 'cause he created it in the first place, isn't there a way theorically to tell the network to keep overall subject traits without training it and waste time.
This imho would unleash the power of SD. Ranging from creating storyboard, to notime character design creation
Ps. sorry for my bad english

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

The issue discusses ControlNet, Stable Diffusion, and DreamBooth but names no files, tests, or entry points. Start by reviewing how ControlNet handles conditioning and how existing subject-preservation methods work. Done would require a concrete approach and reproducible criteria for preserving a generated character without additional training.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai, 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.