Feedback & Support
Help us improve by sharing your thoughts or reporting issues.
Open
9
In Progress
0
Planned
1
Closed
8
Recent Feedback
View All →Pony XL Prediction Fail
Pony XL will not intalise, as above it prediction fails. I did drop a note in discord a couple of days ago. I know you are busy and it's an experimental. If you could give it a kick, It would be much appreciated. If it's already on your to do list, please ignore this comment. Many thanks.
IC-Light V2 Quality Difference on Pixel Dojo
I've been playing around with IC-Light V2 and noticed a pretty big difference in quality between Pixel Dojo and the version on Hugging Face Spaces here: https://huggingface.co/spaces/lllyasviel/iclight-v2. They also talk about it on GitHub: https://github.com/lllyasviel/IC-Light/discussions/98. On Pixel Dojo, the images look a bit lower quality and sometimes distorted, especially faces. Plus, the biggest image I can get out of Pixel Dojo is 1024x1024. I'm not sure if Pixel Dojo is using a different version of the model or if there's something else going on, but the one on Hugging Face seems to be the "Flux-based IC-Light Model with 16ch VAE and native high resolution" and it looks way better. Would it be possible to get that version on Pixel Dojo? It'd make the IC-Light feature much more awesome to use! Thanks!
Enhancements for PuLID Characters
- Multi-Image Support (1 → 4 images) to improve identity accuracy. - Step Limit Increase (20 → 50 steps). - Add Kolors Character as a complementary model.
Samplers and schedulers
It would be nice to be able to choose between different samplers and schedulers for all available bases. Creative upscaler lacks of settings too.
log the start and end time of generations
log the start and end time of generations (for upscaling, image to video, and text to video) basically for anything that takes a long time. Then in the UI we can show an average generation time for a better user experience.
I lose a lot of generated images
Hi, I keep losing images using img2img. We discussed this before and you said it happens when you do updates. That makes sense because it only happens sporadically. Sometimes days go by and I don't lose any, other days I'll lose 4 in a row. The weird part is that the page appears to work fine. I can upload and edit prompt and parameters, and even generate the image. But a moment after the completed image appears, the page deletes everything and goes back to default. I estimate that I've lost at least 40 images over the last couple of months. Is there any way to remedy or prevent this problem? Any help or extra credits appreciated! Best, Kevin
Folders in My Media Gallery
Add folders for organizing images. Ideally sub folders or nested folders as well.
Show all images trained with a specific LoRA
For a given LoRA model, provide a way to show all images that were generated with it. A way to filter in the My Media page.
Bulk operations in the My Media gallery
Allow bulk delete, bulk download, and bulk public/private toggles in the My Media Gallery.
Some upgrades to my images are needed
from discord: Would you be able to put a next arrow in the Personal gallery, it becomes a poor u/x when you want to look through multiple photos, and you have to close one, then click on another to open. Also adding a "show more" for really long prompts should be added as well as a simple copy button to easily copy and paste prompts. There's some more TLC, but these are just a few suggestions to improve quality of life and user experience.