Human portrait generation outputs have extremely poor quality and are unusable
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1.6k
- Forks
- 174
- PR merge metrics
- No merged PRs in 30d
Description
All generated human portraits show terrible overall quality. Details are severely blurred, facial features are distorted and chaotic, with weird facial structures, messy hair and unnatural body contours. The generated results are visually unacceptable and cannot be used for any practical scenarios.
No special complex prompts were used. Common portrait prompts consistently produce low-quality broken figures. Need optimization for human generation capability. :(
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the reported failures with the common portrait prompts described in the issue and record the resulting image-quality problems. The issue names no files, tests, or entry points, so trace the generation path from the reproduction before defining a fix. Done means human portraits no longer show the reported blur, distorted features, messy hair, and unnatural body contours.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-vision, machine-learning
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100