lllyasviel / lllyasviel/stable-diffusion-webui-forge
[Feature Request]: Never OOM for SVD and Zero123
- Dominant language
- Python
- Stars
- 13k
- Forks
- 1.7k
- PR merge metrics
- No merged PRs in 30d
Description
### Is there an existing issue for this?
- [X] I have searched the existing issues and checked the recent builds/commits
### What would your feature do ?
Currently I am using GTX 1050 Ti 4GB, Forge UI works pretty well in SDXL, but to render a higres image in txt2img I always use the checkbox in Never OOM --> Enabled for VAE (always tiled). Even without hires fix, using SDXL Lightning model 4 steps, I can render an image of 1600x1300 in about 2.59 minutes. This is not currently possible in Automatic1111.
Now, I am asking the same "Never OOM" feature to be implemented in "SVD" and "Zero123" tab, it will help the higres quality.
Thanks for your support, keep the awesome work
:)
### Proposed workflow
1. Go to SVD and Zero123
2. Press the "Never OOM"
3. You get the highres video and image
### Additional information
_No response_
Contributor guide
No contributing guide indexed for this repository
Research direction
Start at the SVD and Zero123 tabs and trace how the existing Never OOM option is handled for VAE tiling. Implement equivalent controls for both workflows and verify that enabling them supports the requested high-resolution video and image generation without out-of-memory failures.
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
- Mostly clear
- Newbie friendliness
- 25/100