Spaces:
Runtime error
Runtime error
zakarianasim073 commited on
Commit ·
aa12de4
1
Parent(s): 12a30c6
Fix requirements.txt and add ONNX Runtime
Browse files- README.md +41 -79
- requirements.txt +16 -44
README.md
CHANGED
|
@@ -1,23 +1,12 @@
|
|
| 1 |
-
---
|
| 2 |
-
title: Roop_unleashed4.3.2
|
| 3 |
-
emoji: 💩
|
| 4 |
-
colorFrom: green
|
| 5 |
-
colorTo: blue
|
| 6 |
-
sdk: gradio
|
| 7 |
-
sdk_version: 5.4.0
|
| 8 |
-
app_file: run.py
|
| 9 |
-
pinned: false
|
| 10 |
-
license: apache-2.0
|
| 11 |
-
---
|
| 12 |
# roop-unleashed
|
| 13 |
|
| 14 |
-
|
| 15 |
|
|
|
|
| 16 |
|
| 17 |
-
|
| 18 |
-
|
| 19 |
|
| 20 |
-
|
| 21 |
|
| 22 |
### Features
|
| 23 |
|
|
@@ -35,7 +24,6 @@ Uncensored Deepfakes for images and videos without training and an easy-to-use G
|
|
| 35 |
|
| 36 |
and lots more...
|
| 37 |
|
| 38 |
-
|
| 39 |
## Disclaimer
|
| 40 |
|
| 41 |
This project is for technical and academic use only.
|
|
@@ -44,40 +32,15 @@ Users of this software are expected to use this software responsibly while abidi
|
|
| 44 |
|
| 45 |
In the event of violation of the legal and ethical requirements of the user's country or region, this code repository is exempt from liability
|
| 46 |
|
| 47 |
-
Note: # Roop-Unleashed (Hugging Face Space)
|
| 48 |
-
|
| 49 |
-
This Space runs Roop-Unleashed with a Gradio interface.
|
| 50 |
-
|
| 51 |
-
---
|
| 52 |
-
|
| 53 |
-
## ⚙️ Runtime Configuration
|
| 54 |
-
|
| 55 |
-
We explicitly set the Python runtime to **3.10** using a `runtime.txt` file:
|
| 56 |
-
|
| 57 |
-
### Why Python 3.10?
|
| 58 |
-
- Hugging Face Spaces default to Python 3.13.
|
| 59 |
-
- Several dependencies (e.g., `torch==2.1.2`, `numpy==1.26.4`) do not yet provide wheels for Python 3.13.
|
| 60 |
-
- For compatibility and stability, we pinned the runtime to Python 3.10, which is widely supported by ML libraries.
|
| 61 |
-
|
| 62 |
-
---
|
| 63 |
-
|
| 64 |
-
## 📦 Dependencies
|
| 65 |
-
|
| 66 |
-
All dependencies are pinned in `requirements.txt` to avoid conflicts:
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
### Installation
|
| 72 |
|
| 73 |
Please refer to the [wiki](https://github.com/C0untFloyd/roop-unleashed/wiki).
|
| 74 |
|
| 75 |
#### macOS Installation
|
| 76 |
-
Simply run the following command. It will check and install all dependencies if necessary.
|
| 77 |
-
|
| 78 |
-
`/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/PJF16/roop-unleashed/master/installer/macOSinstaller.sh)`
|
| 79 |
|
|
|
|
| 80 |
|
|
|
|
| 81 |
|
| 82 |
### Usage
|
| 83 |
|
|
@@ -85,6 +48,7 @@ Simply run the following command. It will check and install all dependencies if
|
|
| 85 |
- Linux: `python run.py`
|
| 86 |
- macOS: `sh runMacOS.sh`
|
| 87 |
- Dockerfile:
|
|
|
|
| 88 |
```shell
|
| 89 |
docker build -t roop-unleashed . && docker run -t \
|
| 90 |
-p 7860:7860 \
|
|
@@ -95,19 +59,34 @@ Simply run the following command. It will check and install all dependencies if
|
|
| 95 |
roop-unleashed
|
| 96 |
```
|
| 97 |
|
| 98 |
-
<a target="_blank" href="https://colab.research.google.com/github/C0untFloyd/roop-unleashed/blob/main/roop-unleashed.ipynb">
|
| 99 |
-
<img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/>
|
| 100 |
-
</a>
|
| 101 |
-
|
| 102 |
-
|
| 103 |
Additional commandline arguments are currently unsupported and settings should be done via the UI.
|
| 104 |
|
| 105 |
> Note: When you run this program for the first time, it will download some models roughly ~2Gb in size.
|
| 106 |
|
|
|
|
| 107 |
|
|
|
|
| 108 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
|
| 110 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 111 |
|
| 112 |
**28.9.2024** v4.3.1
|
| 113 |
|
|
@@ -116,7 +95,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 116 |
- New swapping mode "All input faces"
|
| 117 |
- Average total fps displayed and setting for autorun
|
| 118 |
|
| 119 |
-
|
| 120 |
**16.9.2024** v4.2.8
|
| 121 |
|
| 122 |
- Bugfix: Starting roop-unleashed without NVIDIA gpu but cuda option enabled
|
|
@@ -126,7 +104,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 126 |
- Source Files Component cleared after face detection to release temp files
|
| 127 |
- Added masking and mouth restore options to virtual cam
|
| 128 |
|
| 129 |
-
|
| 130 |
**9.9.2024** v4.2.3
|
| 131 |
|
| 132 |
- Hotfix for gradio pydantic issue with fastapi
|
|
@@ -136,7 +113,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 136 |
- Hardcoded image output format for livecam to jpeg - might be faster than previous webp
|
| 137 |
- Chain events to be only executed if previous was a success
|
| 138 |
|
| 139 |
-
|
| 140 |
**5.9.2024** v4.2.0
|
| 141 |
|
| 142 |
- Added ability to move input & target faces order
|
|
@@ -147,18 +123,15 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 147 |
- Added CLI Arguments --server_share and --cuda_device_id
|
| 148 |
- Added webp image support
|
| 149 |
|
| 150 |
-
|
| 151 |
**15.07.2024** v4.1.1
|
| 152 |
|
| 153 |
- Bugfix: Post-processing after swapping
|
| 154 |
|
| 155 |
-
|
| 156 |
**14.07.2024** v4.1.0
|
| 157 |
|
| 158 |
- Added subsample upscaling to increase swap resolution
|
| 159 |
- Upgraded gradio
|
| 160 |
|
| 161 |
-
|
| 162 |
**12.05.2024** v4.0.0
|
| 163 |
|
| 164 |
- Bugfix: Unnecessary init every frame in live-cam
|
|
@@ -172,7 +145,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 172 |
- Added complete frame processing to extras tab
|
| 173 |
- Colorize, upscale and misc filters added
|
| 174 |
|
| 175 |
-
|
| 176 |
**22.04.2024** v3.9.0
|
| 177 |
|
| 178 |
- Bugfix: Face detection bounding box corrupt values at weird angles
|
|
@@ -184,7 +156,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 184 |
- Face masking only runs when there is a face detected
|
| 185 |
- Removed unnecessary toggle checkbox for text masking
|
| 186 |
|
| 187 |
-
|
| 188 |
**22.03.2024** v3.6.5
|
| 189 |
|
| 190 |
- Bugfix: Installer pulling latest update on first installation
|
|
@@ -192,7 +163,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 192 |
- Exposed erosion and blur amounts to UI
|
| 193 |
- Using same values for manual masking too
|
| 194 |
|
| 195 |
-
|
| 196 |
**20.03.2024** v3.6.3
|
| 197 |
|
| 198 |
- Bugfix: Workaround for Gradio Slider Change Bug
|
|
@@ -201,7 +171,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 201 |
- Show offset mask as overlay
|
| 202 |
- Changed layout for masking
|
| 203 |
|
| 204 |
-
|
| 205 |
**18.03.2024** v3.6.0
|
| 206 |
|
| 207 |
- Updated to Gradio 4.21.0 - requiring many changes under the hood
|
|
@@ -211,7 +180,6 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 211 |
- Removed unnecessary activate live-cam option
|
| 212 |
- Added time info to preview frame and changed frame slider event to allow faster changes
|
| 213 |
|
| 214 |
-
|
| 215 |
**10.03.2024** v3.5.5
|
| 216 |
|
| 217 |
- Bugfix: Installer Path Env
|
|
@@ -222,16 +190,13 @@ Additional commandline arguments are currently unsupported and settings should b
|
|
| 222 |
- Live Cam recoded to run separate from virtual cam and without blocking controls
|
| 223 |
- Swapping with only 1 target face allows selecting from several input faces
|
| 224 |
|
| 225 |
-
|
| 226 |
-
|
| 227 |
**08.01.2024** v3.5.0
|
| 228 |
|
| 229 |
- Bugfix: wrong access options when creating folders
|
| 230 |
-
- New auto rotation of horizontal faces, fixing bad landmark positions (expanded on
|
| 231 |
- Simple VR Option for stereo Images/Movies, best used in selected face mode
|
| 232 |
- Added RestoreFormer Enhancer - https://github.com/wzhouxiff/RestoreFormer
|
| 233 |
-
- Bumped up package versions for onnx/Torch etc.
|
| 234 |
-
|
| 235 |
|
| 236 |
**16.10.2023** v3.3.4
|
| 237 |
|
|
@@ -249,21 +214,18 @@ Initial Gradio Version - old TkInter Version now deprecated
|
|
| 249 |
- Updated readme for new gradio version
|
| 250 |
- Updated Colab
|
| 251 |
|
| 252 |
-
|
| 253 |
# Acknowledgements
|
| 254 |
|
| 255 |
Lots of ideas, code or pre-trained models borrowed from the following projects:
|
| 256 |
|
| 257 |
-
https://github.com/deepinsight/insightface<br
|
| 258 |
-
https://github.com/s0md3v/roop<br
|
| 259 |
-
https://github.com/AUTOMATIC1111/stable-diffusion-webui<br
|
| 260 |
-
https://github.com/Hillobar/Rope<br
|
| 261 |
-
https://github.com/TencentARC/GFPGAN<br
|
| 262 |
-
https://github.com/kadirnar/codeformer-pip<br
|
| 263 |
-
https://github.com/csxmli2016/DMDNet<br
|
| 264 |
-
https://github.com/glucauze/sd-webui-faceswaplab<br
|
| 265 |
-
https://github.com/ykk648/face_power<br
|
| 266 |
-
|
| 267 |
-
|
| 268 |
-
<br />
|
| 269 |
-
Thanks to all developers!
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# roop-unleashed
|
| 2 |
|
| 3 |
+
This is **unofficial** mirror of repository
|
| 4 |
|
| 5 |
+
Models: <https://codeberg.org/roop-unleashed/models>
|
| 6 |
|
| 7 |
+
[Changelog](#changelog) • [Usage](#usage)
|
|
|
|
| 8 |
|
| 9 |
+
Uncensored Deepfakes for images and videos without training and an easy-to-use GUI.
|
| 10 |
|
| 11 |
### Features
|
| 12 |
|
|
|
|
| 24 |
|
| 25 |
and lots more...
|
| 26 |
|
|
|
|
| 27 |
## Disclaimer
|
| 28 |
|
| 29 |
This project is for technical and academic use only.
|
|
|
|
| 32 |
|
| 33 |
In the event of violation of the legal and ethical requirements of the user's country or region, this code repository is exempt from liability
|
| 34 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
### Installation
|
| 36 |
|
| 37 |
Please refer to the [wiki](https://github.com/C0untFloyd/roop-unleashed/wiki).
|
| 38 |
|
| 39 |
#### macOS Installation
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
+
Simply run the following command. It will check and install all dependencies if necessary.
|
| 42 |
|
| 43 |
+
`/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/PJF16/roop-unleashed/master/installer/macOSinstaller.sh)"`
|
| 44 |
|
| 45 |
### Usage
|
| 46 |
|
|
|
|
| 48 |
- Linux: `python run.py`
|
| 49 |
- macOS: `sh runMacOS.sh`
|
| 50 |
- Dockerfile:
|
| 51 |
+
|
| 52 |
```shell
|
| 53 |
docker build -t roop-unleashed . && docker run -t \
|
| 54 |
-p 7860:7860 \
|
|
|
|
| 59 |
roop-unleashed
|
| 60 |
```
|
| 61 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
Additional commandline arguments are currently unsupported and settings should be done via the UI.
|
| 63 |
|
| 64 |
> Note: When you run this program for the first time, it will download some models roughly ~2Gb in size.
|
| 65 |
|
| 66 |
+
### Changelog
|
| 67 |
|
| 68 |
+
**03.01.2025** v4.4.1
|
| 69 |
|
| 70 |
+
Bugfix: Extras Tab
|
| 71 |
+
Bugfix: Preview Mask
|
| 72 |
+
Bugfix: DMDNet working on Cuda only
|
| 73 |
+
Updated Colab Script to also install CUDNN
|
| 74 |
|
| 75 |
+
**31.12.2024** v4.4.0 Hotfix
|
| 76 |
+
|
| 77 |
+
Bugfix: Updated Colab to use present Cuda Drivers
|
| 78 |
+
Bugfix: Live-Cam not working because of new face swapper
|
| 79 |
+
Set default swapping model back to Insightface
|
| 80 |
+
|
| 81 |
+
Happy New Year!
|
| 82 |
+
|
| 83 |
+
**30.12.2024** v4.4.0
|
| 84 |
+
|
| 85 |
+
- Added random face selection mode
|
| 86 |
+
- Added alternative face swapping model with 128px & 256 px output ([ReSwapper](https://github.com/somanchiu/ReSwapper/tree/main))
|
| 87 |
+
- Video repair added to Extras Tab
|
| 88 |
+
- Updated most packages to newer versions. CUDA >= 12.4 now required!
|
| 89 |
+
- Several minor bugfixes and QoL Changes
|
| 90 |
|
| 91 |
**28.9.2024** v4.3.1
|
| 92 |
|
|
|
|
| 95 |
- New swapping mode "All input faces"
|
| 96 |
- Average total fps displayed and setting for autorun
|
| 97 |
|
|
|
|
| 98 |
**16.9.2024** v4.2.8
|
| 99 |
|
| 100 |
- Bugfix: Starting roop-unleashed without NVIDIA gpu but cuda option enabled
|
|
|
|
| 104 |
- Source Files Component cleared after face detection to release temp files
|
| 105 |
- Added masking and mouth restore options to virtual cam
|
| 106 |
|
|
|
|
| 107 |
**9.9.2024** v4.2.3
|
| 108 |
|
| 109 |
- Hotfix for gradio pydantic issue with fastapi
|
|
|
|
| 113 |
- Hardcoded image output format for livecam to jpeg - might be faster than previous webp
|
| 114 |
- Chain events to be only executed if previous was a success
|
| 115 |
|
|
|
|
| 116 |
**5.9.2024** v4.2.0
|
| 117 |
|
| 118 |
- Added ability to move input & target faces order
|
|
|
|
| 123 |
- Added CLI Arguments --server_share and --cuda_device_id
|
| 124 |
- Added webp image support
|
| 125 |
|
|
|
|
| 126 |
**15.07.2024** v4.1.1
|
| 127 |
|
| 128 |
- Bugfix: Post-processing after swapping
|
| 129 |
|
|
|
|
| 130 |
**14.07.2024** v4.1.0
|
| 131 |
|
| 132 |
- Added subsample upscaling to increase swap resolution
|
| 133 |
- Upgraded gradio
|
| 134 |
|
|
|
|
| 135 |
**12.05.2024** v4.0.0
|
| 136 |
|
| 137 |
- Bugfix: Unnecessary init every frame in live-cam
|
|
|
|
| 145 |
- Added complete frame processing to extras tab
|
| 146 |
- Colorize, upscale and misc filters added
|
| 147 |
|
|
|
|
| 148 |
**22.04.2024** v3.9.0
|
| 149 |
|
| 150 |
- Bugfix: Face detection bounding box corrupt values at weird angles
|
|
|
|
| 156 |
- Face masking only runs when there is a face detected
|
| 157 |
- Removed unnecessary toggle checkbox for text masking
|
| 158 |
|
|
|
|
| 159 |
**22.03.2024** v3.6.5
|
| 160 |
|
| 161 |
- Bugfix: Installer pulling latest update on first installation
|
|
|
|
| 163 |
- Exposed erosion and blur amounts to UI
|
| 164 |
- Using same values for manual masking too
|
| 165 |
|
|
|
|
| 166 |
**20.03.2024** v3.6.3
|
| 167 |
|
| 168 |
- Bugfix: Workaround for Gradio Slider Change Bug
|
|
|
|
| 171 |
- Show offset mask as overlay
|
| 172 |
- Changed layout for masking
|
| 173 |
|
|
|
|
| 174 |
**18.03.2024** v3.6.0
|
| 175 |
|
| 176 |
- Updated to Gradio 4.21.0 - requiring many changes under the hood
|
|
|
|
| 180 |
- Removed unnecessary activate live-cam option
|
| 181 |
- Added time info to preview frame and changed frame slider event to allow faster changes
|
| 182 |
|
|
|
|
| 183 |
**10.03.2024** v3.5.5
|
| 184 |
|
| 185 |
- Bugfix: Installer Path Env
|
|
|
|
| 190 |
- Live Cam recoded to run separate from virtual cam and without blocking controls
|
| 191 |
- Swapping with only 1 target face allows selecting from several input faces
|
| 192 |
|
|
|
|
|
|
|
| 193 |
**08.01.2024** v3.5.0
|
| 194 |
|
| 195 |
- Bugfix: wrong access options when creating folders
|
| 196 |
+
- New auto rotation of horizontal faces, fixing bad landmark positions (expanded on PR 364)
|
| 197 |
- Simple VR Option for stereo Images/Movies, best used in selected face mode
|
| 198 |
- Added RestoreFormer Enhancer - https://github.com/wzhouxiff/RestoreFormer
|
| 199 |
+
- Bumped up package versions for onnx/Torch etc.
|
|
|
|
| 200 |
|
| 201 |
**16.10.2023** v3.3.4
|
| 202 |
|
|
|
|
| 214 |
- Updated readme for new gradio version
|
| 215 |
- Updated Colab
|
| 216 |
|
|
|
|
| 217 |
# Acknowledgements
|
| 218 |
|
| 219 |
Lots of ideas, code or pre-trained models borrowed from the following projects:
|
| 220 |
|
| 221 |
+
https://github.com/deepinsight/insightface<br/>
|
| 222 |
+
https://github.com/s0md3v/roop<br/>
|
| 223 |
+
https://github.com/AUTOMATIC1111/stable-diffusion-webui<br/>
|
| 224 |
+
https://github.com/Hillobar/Rope<br/>
|
| 225 |
+
https://github.com/TencentARC/GFPGAN<br/>
|
| 226 |
+
https://github.com/kadirnar/codeformer-pip<br/>
|
| 227 |
+
https://github.com/csxmli2016/DMDNet<br/>
|
| 228 |
+
https://github.com/glucauze/sd-webui-faceswaplab<br/>
|
| 229 |
+
https://github.com/ykk648/face_power<br/>
|
| 230 |
+
|
| 231 |
+
Thanks to all developers!
|
|
|
|
|
|
requirements.txt
CHANGED
|
@@ -1,48 +1,20 @@
|
|
| 1 |
--extra-index-url https://download.pytorch.org/whl/cu124
|
| 2 |
-
numpy
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
torch; sys_platform =
|
| 10 |
-
|
| 11 |
-
torchvision; sys_platform =
|
| 12 |
-
|
| 13 |
-
onnxruntime
|
| 14 |
-
onnxruntime-
|
| 15 |
-
|
|
|
|
| 16 |
ftfy
|
| 17 |
regex
|
| 18 |
pyvirtualcam
|
| 19 |
-
pydantic
|
| 20 |
-
gradio
|
| 21 |
-
huggingface_hub>=0.25.0
|
| 22 |
-
|
| 23 |
-
# Core UI framework
|
| 24 |
-
gradio[oauth,mcp]
|
| 25 |
-
|
| 26 |
-
# Hugging Face integration
|
| 27 |
-
huggingface_hub
|
| 28 |
-
spaces
|
| 29 |
-
|
| 30 |
-
# Deep learning stack
|
| 31 |
-
torch
|
| 32 |
-
torchvision
|
| 33 |
-
|
| 34 |
-
# Image/video processing
|
| 35 |
-
opencv-python
|
| 36 |
-
numpy
|
| 37 |
-
Pillow
|
| 38 |
-
|
| 39 |
-
# Web server & networking
|
| 40 |
-
uvicorn
|
| 41 |
-
websockets
|
| 42 |
-
|
| 43 |
-
# Utilities
|
| 44 |
-
protobuf
|
| 45 |
-
click
|
| 46 |
-
|
| 47 |
-
# If roop is external, uncomment the next line:
|
| 48 |
-
# git+https://github.com/s0md3v/roop.git
|
|
|
|
| 1 |
--extra-index-url https://download.pytorch.org/whl/cu124
|
| 2 |
+
numpy==1.26.4
|
| 3 |
+
gradio==5.9.1
|
| 4 |
+
opencv-python-headless==4.10.0.84
|
| 5 |
+
onnx==1.16.1
|
| 6 |
+
insightface==0.7.3
|
| 7 |
+
albucore==0.0.16
|
| 8 |
+
psutil==5.9.6
|
| 9 |
+
torch==2.5.1+cu124; sys_platform != 'darwin'
|
| 10 |
+
torch==2.5.1; sys_platform == 'darwin'
|
| 11 |
+
torchvision==0.20.1+cu124; sys_platform != 'darwin'
|
| 12 |
+
torchvision==0.20.1; sys_platform == 'darwin'
|
| 13 |
+
onnxruntime==1.20.1; sys_platform == 'darwin' and platform_machine != 'arm64'
|
| 14 |
+
onnxruntime-silicon==1.20.1; sys_platform == 'darwin' and platform_machine == 'arm64'
|
| 15 |
+
onnxruntime-gpu==1.20.1; sys_platform != 'darwin'
|
| 16 |
+
tqdm==4.66.4
|
| 17 |
ftfy
|
| 18 |
regex
|
| 19 |
pyvirtualcam
|
| 20 |
+
pydantic==2.10.4
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|