timntorres
timntorres
As per #2921, Hypernet preview images now support infotext if the user so desires. Tested and works with... ✅ ...both enabled and disabled checkbox “Read parameters (prompt, etc...) from txt2img...
Fixes #3524 and #292 and #177. Uses PIL's [exif_transpose](https://pillow.readthedocs.io/en/stable/reference/ImageOps.html#PIL.ImageOps.exif_transpose) to consider the orientation of photos taken with smartphones. ✅Works with img2img ✅Works with Inpaint, no mask ✅Works with Inpaint, drawn...
Assumes #3516 is indeed a bug and fixes it. Specifically, introduces a case that always updates image ([line 329](https://github.com/timntorres/stable-diffusion-webui/blob/fa062a52c023efa0f2c447a428ef71c3e07b2dd5/modules/ui.py#L329)) when training. I considered a one line fix, but three conditions...
This fixes the bug described by @SeverianVoid both in [a comment on 5245c7a](https://github.com/AUTOMATIC1111/stable-diffusion-webui/commit/5245c7a4935f67b677da0f5a1fc2b74c074aa0e2#commitcomment-87547378) and in [a comment on issue 2921](https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/2921#issuecomment-1287523905).
Fulfills #3186. The requested changes are entirely symmetric to #3189, #3309, and #3625; just for embeddings instead of hypernets. I ran some light testing and can confirm these changes work...
Fulfills #3766. My first attempt both at writing a feature and changing the UI. Thorough reviews welcome and testing needed. **Especially if your workflow involves high step counts or Generate...
Implements #3601 but for Hypernetworks. This, as with @patrickas's code, appends a new parameters block to `settings.txt` each time the training function is called. A few improvements over #3601: -...
Addresses #3825. Saves a generation's Hypernetwork strength to infotext (both PNG chunks/EXIF data and .txts). Also loads it with PNG info's "Send to..." tabs. Also made it so infotext explicitly...
This pull request consists of two parts. 1\) Addresses #4028. Allows user to save a copy of image/images before highres fix is applied. Saving "-before-highres-fix" is tested and works with......
Fixes #4209: ~~Saving before highres fix now saves images *before* upscaling when `use_scale_latent_for_hires_fix` is enabled.~~ (Previous sentence was patched in f674c48.) **Saving before highres fix now supports infotext.** To get...