lllyasviel / lllyasviel/ControlNet

ControlNet high contrast image

Open
#530 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
34.1k
Forks
3k
PR merge metrics
No merged PRs in 30d

Description

Hi, I have been training ControlNet in a style transfer manner.
But the generated outputs colors are somewhat dimmed.

Should I normalize image channelwise ?

I have been training with 2 GPUs w/ batch size = 2 and accumulate_grad = 4
Should I increase the batch size ?

GT image
![image](https://github.com/lllyasviel/ControlNet/assets/58447982/c4a00891-cb63-46a9-a917-6b62a8f2d5cd)

Output image
![image](https://github.com/lllyasviel/ControlNet/assets/58447982/4028ca86-6703-4f51-b5b9-e9df85d64a0c)

Contributor guide

No contributing guide indexed for this repository

Research direction

The issue names no files, tests, or entry points; begin by reviewing the linked ground-truth and output images and reproducing the reported dimmed colors with the stated training setup. Done is not defined because the report asks for advice about channel normalization and batch size rather than a concrete repository change.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
15/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.