Stable Diffusion GUI 0.65 + Real-ESRGAN App Update. (Patreon)
Content
Hi all, been dealing with a few personal problems this last few days, so didn't managed to work as much I would like, but still got some new updates ready.
Real-ESRGAN:
There been some comments that the link was down. I accidentally had deleted it. Took this opportunity to clean the code a little and add an extra model to upscale anime videos. There still more updates I want to make, but with the code clean, it will be easier to update it soon:
https://drive.google.com/file/d/1b1Py9YGpfTDRb5OjP9_nh6Kpp6CmxP0d/view?usp=share_link
Stable Diffusion:
Link:
https://drive.google.com/drive/folders/1ZzYYwSH0ZzjhkpjKxn3YC2lGKAUXUbLJ?usp=share_link
New Stuff:
- You can download the Stable Diffusion 1.5 model directly from the GUI. This still in Beta test. Will add more models soon. The model will download to the cache folder from torch, I may change this to the model_sd folder later on.
- New schedulers from Karras. Now that the code is ready, it will be super easy to add more Karras Schedulers in the future. dpmpp sde and dpmpp 2m are community favorites. For now they work with prompt2img and a little with img2img.
- You can use .safetensors in the models and vae.
- After selecting a folder for Dreambooth training, you can press "Check all tags from the selected folder", that way you can see what tags will be used for each image.
- "Auto use Output as Input", if you are using multiple img2img, this may be useful.
- Dreambooth only use images to train and ignore other extensions.
Working right now:
- Was gonna to add "prior preservation" to dreambooth training, but there still some problems and would take some days just to fix this, will appear on the next build.
- Making Textural Inversion tokens work.
- Add Lora support to models.
- Add upscale models support.
- InstructPix2Pix
Dreambooth tags:
As default all images will use the "Exclusive prompt" as tags. You can override one image using the "ImageFile" + ".txt" and write the image prompt on the txt.
You can also create sub-folders and the tags for all images inside the sub-folder will be the sub-folder name.
Use the "Check all tags from the selected folder" button to check if the tags are correct.
Here a nice site to find some models: