Commit History
-
Optimize Gallery UI and enable Ctrl+V image pasting
- Remove custom upload/delete buttons in favor of native gr.Gallery features. - Add support for pasting images (Ctrl+V) directly into the gallery. - Refactor code to remove unused functions and simplify the layout. - Improve console logging: Group execution times by image to reduce redundancy and improve readability. - Optimize ZIP handling: Conditionally create face-related ZIPs only when face restoration is enabled. Added logic to remove empty ZIP files (<= 22 bytes) before returning results. - Fix filename bug: Reset self.modelInUse at the start of inference to prevent model name suffix accumulation across multiple runs.314497f
-
Update the PyTorch version to CUDA 12.8 in the requirements.
31305bd
-
feat: Add option to force PNG output for best quality
f48354a
-
feat(dataops): Proactively split large tiles in auto_split_upscale to prevent CUDA OOM errors.
aa7cfca -
Fixed the handling behavior when face_model is set to None.
129b135
-
Download output is now in zip file format.
645bf54
-
Upgraded Gradio version to 5.25.1.
b4e0c45 -
Fixed the duplicate filename checking function during inference.
Previously, serial numbers were automatically added to duplicate filenames, but they were incorrectly added after the file extension. This has now been fixed. Upgraded Gradio version to 5.25.1.2cb2e7b
-
Add progress display during the inference phase
99141c3
-
Added support for using gallery as an image input.
You can now upload multiple images at once. Due to the slow execution in the online environment, the default image limit is set to 5. You can modify this limit by setting input_images_limit. If set to -1, there will be no limit. ex. python app.py --input_images_limit -1a01cc5b -
Add display of execution time for the upscale and face enhancer models.
Improve the progress display of auto_split_upscale execution.20b04f8
-
Referring to the Spandrel project, the hyperparameter settings of the upscale model have been updated.
Added support for parsing SRFormer models from the SRFormer project by author HVision-NKU.69d49d2
-
Added support for parsing MoSR models from the MoSR project by author umzi2.
cc44a11 -
Added support for parsing ATD models from the ATD project by author LabShuHangGU.
8cbd0a5 -
Added support for parsing DRCT models from the DRCT project by author ming053l.
308a135
-
Update the calling method of the HAT Module.
ac599ac
-
Update the Markdown content displaying upscale model tables.
bb6a756
-
Improve RRDB and ESRGAN model loading methods.
Set torch.backends.cudnn options to avoid get black image for RTX16xx card086ab89
-
Add two option parameters: Face eye dist threshold and Face detection only center.
Add several SRVGGNet models created by Phhofm. Face eye dist threshold: A threshold to filter out faces with too small an eye distance (e.g., side faces). Face detection only center: If set to True, only the face closest to the center of the image will be kept.c59704a
-
Overall improvements with added support for various face restoration models and upscale models.
01. The inference uses the auto_split_upscale mechanism, auto_split_upscale is in the dataops.py file, and its source is from the ESRGAN project forked by authors joeyballentine and BlueAmulet. 02. The face model now supports RestoreFormer++, authored by wzhouxiff. 03. The face model now supports CodeFormer, authored by sczhou. 04. The face model now supports GPEN, authored by yangxy. 05. Added support for parsing older RRDB models from the ESRGAN project. 06. Added support for parsing DAT models from the DAT project by author zhengchen1999. 07. Added support for parsing HAT models from the HAT project by author XPixelGroup. 08. Added support for parsing RealPLKSR models from the PLKSR project by author dslisleedh & neosr-project. 09. In the interface, you can select the type of face detection: "retinaface_resnet50", "YOLOv5l", or "YOLOv5n". The default is "retinaface_resnet50". 10. In the interface, you can set whether to include the model name in the output image file names …debfe4b
-
The inference uses the auto_split_upscale mechanism.
1. auto_split_upscale is in the dataops.py file, and its source is from the ESRGAN project forked by authors joeyballentine and BlueAmulet. 2. The face model now supports RestoreFormer++, authored by wzhouxiff. 3. Added support for parsing older RRDB models from the ESRGAN project. 4. Added support for parsing DAT models from the DAT project by author zhengchen1999. 5. Added support for parsing HAT models from the HAT project by author XPixelGroup. 6. Added support for parsing RealPLKSR models from the PLKSR project by author dslisleedh & neosr-project.c85d0ce
-
1. Change model download from the wget command to the requests module functionality.
2. Add inbrowser=True parameter to Gradio. 3. Upgrade the project's Gradio version to 5.8.0 4. Replace facexlib with the improved version implemented by sczhou.1. Add weights_only=True parameter to torch.load to avoid FutureWarning. 5. Modify GFPGAN to a custom version to avoid encountering the error: ModuleNotFoundError: No module named 'torchvision.transforms.functional_tensor'. 6. Change **basicsr** to the GitHub **BasicSR** version to avoid encountering the error: `ModuleNotFoundError: No module named 'torchvision.transforms.functional_tensor'`. 7. Face Restoration and RealESR can now be freely combined in various ways, or one can be set to "None" to use only the other model.0f09e67
-
Update app.py
d56014c
-
Update app.py
379bb4d -
Update README.md
b896792 -
Create new file
c2b36ce -
Create new file
0510534 -
Create new file
5ed81fc -
initial commit
9d99838