Merge branch 'evolve' into facemasking
This commit is contained in:
Art Gourieff
commit
348685cb06
150
README.md
150
README.md
@ -1,11 +1,17 @@
|
||||
<div align="center">
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/raw/main/sd-webui-reactor/ReActor_logo_red.png?raw=true" alt="logo" width="180px"/>
|
||||
<img src="https://github.com/Gourieff/Assets/raw/main/sd-webui-reactor/ReActor_logo_NEW_EN.png?raw=true" alt="logo" width="180px"/>
|
||||
|
||||
|
||||
|
||||
<a href='https://ko-fi.com/gourieff' target='_blank'><img height='33' src='https://storage.ko-fi.com/cdn/kofi3.png?v=3' border='0' alt='Buy Me a Coffee at ko-fi.com' /></a>
|
||||
|
||||
|
||||
<a href="https://boosty.to/artgourieff" target="_blank">
|
||||
<img src="https://lovemet.ru/www/boosty.jpg" width="108" alt="Support Me on Boosty"/>
|
||||
<br>
|
||||
<sup>
|
||||
Support This Project
|
||||
</sup>
|
||||
</a>
|
||||
|
||||
<hr>
|
||||
|
||||
[](https://github.com/Gourieff/sd-webui-reactor/commits/main)
|
||||
@ -22,45 +28,91 @@
|
||||
|
||||
---
|
||||
<b>
|
||||
<a href="#installation">Installation</a> | <a href="#features">Features</a> | <a href="#usage">Usage</a> | <a href="#api">API</a> | <a href="#troubleshooting">Troubleshooting</a> | <a href="#updating">Updating</a> | <a href="#comfyui">ComfyUI</a> | <a href="#disclaimer">Disclaimer</a>
|
||||
<a href="#latestupdate">What's new</a> | <a href="#installation">Installation</a> | <a href="#features">Features</a> | <a href="#usage">Usage</a> | <a href="#api">API</a> | <a href="#troubleshooting">Troubleshooting</a> | <a href="#updating">Updating</a> | <a href="#comfyui">ComfyUI</a> | <a href="#disclaimer">Disclaimer</a>
|
||||
</b>
|
||||
</div>
|
||||
|
||||
---
|
||||
|
||||
<table>
|
||||
<tr>
|
||||
<td width="134px">
|
||||
<a href="https://boosty.to/artgourieff" target="_blank">
|
||||
<img src="https://lovemet.ru/www/boosty.jpg" width="108" alt="Support Me on Boosty"/>
|
||||
<br>
|
||||
<sup>
|
||||
Support This Project
|
||||
</sup>
|
||||
</a>
|
||||
</td>
|
||||
<td>
|
||||
ReActor is an extension for Stable Diffusion WebUI that allows a very easy and accurate face-replacement (face swap) in images. Based on <a href="https://github.com/Gourieff/ReActor-UI" target="_blank">ReActor-UI</a>.
|
||||
</td>
|
||||
<td width="144px">
|
||||
<a href="https://paypal.me/artgourieff" target="_blank">
|
||||
<img src="https://www.paypalobjects.com/digitalassets/c/website/logo/full-text/pp_fc_hl.svg" width="108" alt="Support Me via PayPal"/>
|
||||
<br>
|
||||
<sup>
|
||||
Donate to This Project
|
||||
</sup>
|
||||
</a>
|
||||
</td>
|
||||
</tr>
|
||||
</table>
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/raw/main/sd-webui-reactor/demo_crop.jpg?raw=true" alt="example"/>
|
||||
|
||||
<a name="latestupdate">
|
||||
|
||||
## What's new in the latest updates
|
||||
|
||||
<details>
|
||||
<summary><a>Click to expand</a></summary>
|
||||
|
||||
### 0.7.0 <sub><sup>BETA2
|
||||
|
||||
- X/Y/Z is improved! One more parameter is ready: you can now select several face models to create a variation of swaps to choose the best one!
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-05.jpg?raw=true" alt="0.7.0-whatsnew-05" width="100%"/>
|
||||
|
||||
To use "Face Model" axis - you should enable ReActor and choose any face model as the Source:<br>
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-07.jpg?raw=true" alt="0.7.0-whatsnew-07" width="50%"/><img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-06.jpg?raw=true" alt="0.7.0-whatsnew-06" width="50%"/>
|
||||
|
||||
Full size demo image: [xyz_demo_2.png](https://raw.githubusercontent.com/Gourieff/Assets/main/sd-webui-reactor/xyz_demo_2.png)
|
||||
|
||||
### 0.7.0 <sub><sup>BETA1
|
||||
|
||||
- X/Y/Z Script support (up to 3 axes: CodeFormer Weight, Restorer Visibility, Face Mask Correction)
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-03.jpg?raw=true" alt="0.7.0-whatsnew-03" width="100%"/>
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-04.jpg?raw=true" alt="0.7.0-whatsnew-04" width="100%"/>
|
||||
|
||||
Full size demo image: [xyz_demo.png](https://raw.githubusercontent.com/Gourieff/Assets/main/sd-webui-reactor/xyz_demo.png)
|
||||
|
||||
__Don't forget to enable ReActor and set any source (to prevent "no source" error)__
|
||||
|
||||
### 0.7.0 <sub><sup>ALPHA1
|
||||
|
||||
- You can now blend faces to build blended face models ("Tools->Face Models->Blend") - due to popular demand
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-01.jpg?raw=true" alt="0.7.0-whatsnew-01" width="100%"/><img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-02.jpg?raw=true" alt="0.7.0-whatsnew-02" width="100%"/>
|
||||
|
||||
- CUDA 12 Support in the Installer script for 1.17.0 ORT-GPU library
|
||||
- New tab "Detection" with "Threshold" and "Max Faces" parameters
|
||||
|
||||
### 0.6.1 <sub><sup>BETA3
|
||||
|
||||
- 'Force Upscale' option inside the 'Upscale' tab: ReActor will run the Upscaler even if there's no face is detected (FR https://github.com/Gourieff/sd-webui-reactor/issues/116)
|
||||
- ReActor shows filenames of source images in-process when the multiple images mode or the folder mode (random as well) is selected
|
||||
|
||||
### 0.6.1 <sub><sup>BETA2
|
||||
|
||||
- 'Save original' option works fine now when you select 'Multiple Images' or 'Source Folder'
|
||||
- Random Mode for 'Source Folder'
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/random_from_folder_demo_01.jpg?raw=true" alt="0.6.1-whatsnew-01" width="100%"/>
|
||||
|
||||
### 0.6.0
|
||||
|
||||
- New Logo
|
||||
- Adaptation to A1111 1.7.0 (appropriate GFPGAN loader)
|
||||
- New URL for the main model file
|
||||
- UI reworked
|
||||
- You can now load several source images (with reference faces) or set the path to the folder containing faces images
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/multiple_source_images_demo_01.png?raw=true" alt="0.6.0-whatsnew-01" width="100%"/>
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/multiple_source_images_demo_02.png?raw=true" alt="0.6.0-whatsnew-02" width="100%"/>
|
||||
|
||||
### 0.5.1
|
||||
|
||||
- You can save face models as "safetensors" files (stored in `<sd-web-ui-folder>\models\reactor\faces`) and load them into ReActor, keeping super lightweight face models of the faces you use;
|
||||
- "Face Mask Correction" option - if you encounter some pixelation around face contours, this option will be useful;
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/face_model_demo_01.jpg?raw=true" alt="0.5.0-whatsnew-01" width="100%"/>
|
||||
|
||||
</details>
|
||||
|
||||
## Installation
|
||||
|
||||
[Automatic1111](#a1111) | [Vladmandic SD.Next](#sdnext) | [Google Colab SD WebUI](#colab)
|
||||
[A1111 WebUI / WebUI-Forge](#a1111) | [SD.Next](#sdnext) | [Google Colab SD WebUI](#colab)
|
||||
|
||||
<a name="a1111">If you use [AUTOMATIC1111 web-ui](https://github.com/AUTOMATIC1111/stable-diffusion-webui/):
|
||||
<a name="a1111">If you use [AUTOMATIC1111 SD WebUI](https://github.com/AUTOMATIC1111/stable-diffusion-webui/) or [SD WebUI Forge](https://github.com/lllyasviel/stable-diffusion-webui-forge):
|
||||
|
||||
1. (For Windows Users):
|
||||
- Install **Visual Studio 2022** (Community version, for example - you need this step to build some of dependencies):
|
||||
@ -68,8 +120,8 @@
|
||||
- OR only **VS C++ Build Tools** (if you don't need the whole Visual Studio) and select "Desktop Development with C++" under "Workloads -> Desktop & Mobile":
|
||||
https://visualstudio.microsoft.com/visual-cpp-build-tools/
|
||||
- OR if you don't want to install VS or VS C++ BT - follow [this steps (sec. VIII)](#insightfacebuild)
|
||||
2. In web-ui, go to the "Extensions" tab and use this URL `https://github.com/Gourieff/sd-webui-reactor` in the "Install from URL" tab and click "Install"
|
||||
3. Please, wait for several minutes until the installation process will be finished
|
||||
2. In web-ui, go to the "Extensions" tab, load "Available" extensions and type "ReActor" in the search field or use this URL `https://github.com/Gourieff/sd-webui-reactor` in the "Install from URL" tab - and click "Install"
|
||||
3. Please, wait for several minutes until the installation process will be finished (be patient, don't interrupt the process)
|
||||
4. Check the last message in your SD-WebUI Console:
|
||||
* If you see the message "--- PLEASE, RESTART the Server! ---" - so, do it, stop the Server (CTRL+C or CMD+C) and start it again - or just go to the "Installed" tab, click "Apply and restart UI"
|
||||
* If you see the message "Done!", just reload the UI
|
||||
@ -82,7 +134,7 @@
|
||||
3. Go to (Windows)`automatic\venv\Scripts` or (MacOS/Linux)`automatic/venv/bin`, run Terminal or Console (cmd) for that folder and type `activate`
|
||||
4. Run `pip install insightface==0.7.3`
|
||||
5. Run SD.Next, go to the "Extensions" tab and use this URL `https://github.com/Gourieff/sd-webui-reactor` in the "Install from URL" tab and click "Install"
|
||||
6. Please, wait for several minutes until the installation process will be finished
|
||||
6. Please, wait for several minutes until the installation process will be finished (be patient, don't interrupt the process)
|
||||
7. Check the last message in your SD.Next Console:
|
||||
* If you see the message "--- PLEASE, RESTART the Server! ---" - stop the Server (CTRL+C or CMD+C) or just close your console
|
||||
8. Go to the `automatic\extensions\sd-webui-reactor` directory - if you see there `models\insightface` folder with the file `inswapper_128.onnx`, just move the file to the `automatic\models\insightface` folder
|
||||
@ -90,8 +142,8 @@
|
||||
|
||||
<a name="colab">If you use [Cagliostro Colab UI](https://github.com/Linaqruf/sd-notebook-collection):
|
||||
|
||||
1. In active WebUI, go to the "Extensions" tab and use this URL `https://github.com/Gourieff/sd-webui-reactor` in the "Install from URL" tab and click "Install"
|
||||
2. Please, wait for several minutes until the installation process will be finished
|
||||
1. In active WebUI, go to the "Extensions" tab, load "Available" extensions and type "ReActor" in the search field or use this URL `https://github.com/Gourieff/sd-webui-reactor` in the "Install from URL" tab - and click "Install"
|
||||
2. Please, wait for several minutes until the installation process will be finished (be patient, don't interrupt the process)
|
||||
3. When you see the message "--- PLEASE, RESTART the Server! ---" (in your Colab Notebook Start UI section "Start Cagliostro Colab UI") - just go to the "Installed" tab and click "Apply and restart UI"
|
||||
4. Enjoy!
|
||||
|
||||
@ -103,10 +155,12 @@
|
||||
- Ability to **save original images** (made before swapping)
|
||||
- **Face restoration** of a swapped face
|
||||
- **Upscaling** of a resulting image
|
||||
- Saving ans loading **Safetensors Face Models**
|
||||
- **Facial Mask Correction** to avoid any pixelation around face contours
|
||||
- Ability to set the **Postprocessing order**
|
||||
- **100% compatibility** with different **SD WebUIs**: Automatic1111, SD.Next, Cagliostro Colab UI
|
||||
- **Fast performance** even with CPU, ReActor for SD WebUI is absolutely not picky about how powerful your GPU is
|
||||
- **CUDA** acceleration support from the version 0.5.0
|
||||
- **CUDA** acceleration support since version 0.5.0
|
||||
- **[API](/API.md) support**: both SD WebUI built-in and external (via POST/GET requests)
|
||||
- **ComfyUI [support](https://github.com/Gourieff/comfyui-reactor-node)**
|
||||
- **Mac M1/M2 [support](https://github.com/Gourieff/sd-webui-reactor/issues/42)**
|
||||
@ -198,7 +252,7 @@ Please, check the path where "inswapper_128.onnx" model is stored. It must be in
|
||||
7. Then one-by-one:
|
||||
- `pip install insightface==0.7.3`
|
||||
- `pip install onnx`
|
||||
- `pip install onnxruntime-gpu>=1.16.1`
|
||||
- `pip install "onnxruntime-gpu>=1.16.1"`
|
||||
- `pip install opencv-python`
|
||||
- `pip install tqdm`
|
||||
8. Type `deactivate`, you can close your Terminal or Console and start your SD WebUI, ReActor should start OK - if not, welcome to the Issues section.
|
||||
@ -222,7 +276,7 @@ Probably, you need to disable the "SD-CN-Animation" extension (or perhaps some a
|
||||
|
||||
This error may occur if there's smth wrong with the model file `inswapper_128.onnx`
|
||||
|
||||
Try to download it manually from [here](https://github.com/facefusion/facefusion-assets/releases/download/models/inswapper_128.onnx)
|
||||
Try to download it manually from [here](https://huggingface.co/datasets/Gourieff/ReActor/resolve/main/models/inswapper_128.onnx)
|
||||
and put it to the `stable-diffusion-webui\models\insightface` replacing existing one
|
||||
|
||||
### **VI. "ValueError: This ORT build has ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'] enabled" OR "ValueError: This ORT build has ['AzureExecutionProvider', 'CPUExecutionProvider'] enabled"**
|
||||
@ -233,7 +287,7 @@ and put it to the `stable-diffusion-webui\models\insightface` replacing existing
|
||||
4. Then:
|
||||
- `python -m pip install -U pip`
|
||||
- `pip uninstall -y onnxruntime onnxruntime-gpu onnxruntime-silicon onnxruntime-extensions`
|
||||
- `pip install onnxruntime-gpu>=1.16.1`
|
||||
- `pip install "onnxruntime-gpu>=1.16.1"`
|
||||
|
||||
If it didn't help - it seems that you have another extension reinstalling `onnxruntime` when SD WebUI checks requirements. Please see your extensions list. Some extensions can causes reinstalling of `onnxruntime-gpu` to `onnxruntime<1.16.1` every time SD WebUI runs.<br>ORT 1.16.0 has a bug https://github.com/microsoft/onnxruntime/issues/17631 - don't install it!
|
||||
|
||||
@ -246,7 +300,7 @@ If it didn't help - it seems that you have another extension reinstalling `onnxr
|
||||
5. Then:
|
||||
- `python -m pip install -U pip`
|
||||
- `pip uninstall protobuf`
|
||||
- `pip install protobuf>=3.20.3`
|
||||
- `pip install "protobuf>=3.20.3"`
|
||||
|
||||
If this method doesn't help - there is some other extension that has a wrong version of protobuf dependence and SD WebUI installs it on a startup requirements check
|
||||
|
||||
@ -255,10 +309,10 @@ If this method doesn't help - there is some other extension that has a wrong ver
|
||||
### **VIII. (For Windows users) If you still cannot build Insightface for some reasons or just don't want to install Visual Studio or VS C++ Build Tools - do the following:**
|
||||
|
||||
1. Close (stop) your SD WebUI Server if it's running
|
||||
2. Download and put [prebuilt Insightface package](https://github.com/Gourieff/sd-webui-reactor/raw/main/example/insightface-0.7.3-cp310-cp310-win_amd64.whl) into the stable-diffusion-webui (or SD.Next) root folder (where you have "webui-user.bat" file)
|
||||
3. From stable-diffusion-webui (or SD.Next) root folder run CMD and `.\venv\Scripts\activate`
|
||||
4. Then update your PIP: `python -m pip install -U pip`
|
||||
5. Then install Insightface: `pip install insightface-0.7.3-cp310-cp310-win_amd64.whl`
|
||||
2. Download and put [prebuilt Insightface package](https://github.com/Gourieff/Assets/raw/main/Insightface/insightface-0.7.3-cp310-cp310-win_amd64.whl) into the stable-diffusion-webui (or SD.Next) root folder where you have "webui-user.bat" file or (A1111 Portable) "run.bat"
|
||||
3. From stable-diffusion-webui (or SD.Next) root folder run CMD and `.\venv\Scripts\activate`<br>OR<br>(A1111 Portable) Run CMD
|
||||
4. Then update your PIP: `python -m pip install -U pip`<br>OR<br>(A1111 Portable)`system\python\python.exe -m pip install -U pip`
|
||||
5. Then install Insightface: `pip install insightface-0.7.3-cp310-cp310-win_amd64.whl`<br>OR<br>(A1111 Portable)`system\python\python.exe -m pip install insightface-0.7.3-cp310-cp310-win_amd64.whl`
|
||||
6. Enjoy!
|
||||
|
||||
### **IX. 07-August-23 Update problem**
|
||||
@ -292,7 +346,7 @@ For the installation instruction follow the [ReActor Node repo](https://github.c
|
||||
|
||||
This software is meant to be a productive contribution to the rapidly growing AI-generated media industry. It will help artists with tasks such as animating a custom character or using the character as a model for clothing etc.
|
||||
|
||||
The developers of this software are aware of its possible unethical applicaitons and are committed to take preventative measures against them. We will continue to develop this project in the positive direction while adhering to law and ethics.
|
||||
The developers of this software are aware of its possible unethical application and are committed to take preventative measures against them. We will continue to develop this project in the positive direction while adhering to law and ethics.
|
||||
|
||||
Users of this software are expected to use this software responsibly while abiding the local law. If face of a real person is being used, users are suggested to get consent from the concerned person and clearly mention that it is a deepfake when posting content online. **Developers and Contributors of this software are not responsible for actions of end-users.**
|
||||
|
||||
|
||||
140
README_RU.md
140
README_RU.md
@ -1,10 +1,16 @@
|
||||
<div align="center">
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/raw/main/sd-webui-reactor/ReActor_logo_red.png?raw=true" alt="logo" width="180px"/>
|
||||
<img src="https://github.com/Gourieff/Assets/raw/main/sd-webui-reactor/ReActor_logo_NEW_RU.png?raw=true" alt="logo" width="180px"/>
|
||||
|
||||
|
||||
|
||||
|
||||
<a href='https://ko-fi.com/gourieff' target='_blank'><img height='33' src='https://storage.ko-fi.com/cdn/kofi3.png?v=3' border='0' alt='Buy Me a Coffee at ko-fi.com' /></a>
|
||||
<a href="https://boosty.to/artgourieff" target="_blank">
|
||||
<img src="https://lovemet.ru/www/boosty.jpg" width="108" alt="Поддержать проект на Boosty"/>
|
||||
<br>
|
||||
<sup>
|
||||
Поддержать проект
|
||||
</sup>
|
||||
</a>
|
||||
|
||||
<hr>
|
||||
|
||||
@ -21,47 +27,89 @@
|
||||
|
||||
---
|
||||
<b>
|
||||
<a href="#installation">Установка</a> | <a href="#features">Возможности</a> | <a href="#usage">Использование</a> | <a href="#api">API</a> | <a href="#troubleshooting">Устранение проблем</a> | <a href="#updating">Обновление</a> | <a href="#comfyui">ComfyUI</a> | <a href="#disclaimer">Ответственность</a>
|
||||
<a href="#latestupdate">Что нового</a> | <a href="#installation">Установка</a> | <a href="#features">Возможности</a> | <a href="#usage">Использование</a> | <a href="#api">API</a> | <a href="#troubleshooting">Устранение проблем</a> | <a href="#updating">Обновление</a> | <a href="#comfyui">ComfyUI</a> | <a href="#disclaimer">Ответственность</a>
|
||||
</b>
|
||||
</div>
|
||||
|
||||
---
|
||||
|
||||
<table>
|
||||
<tr>
|
||||
<td width="134px">
|
||||
<a href="https://boosty.to/artgourieff" target="_blank">
|
||||
<img src="https://lovemet.ru/www/boosty.jpg" width="108" alt="Поддержать проект на Boosty"/>
|
||||
<br>
|
||||
<sup>
|
||||
Поддержать проект
|
||||
</sup>
|
||||
</a>
|
||||
</td>
|
||||
<td>
|
||||
ReActor это расширение для Stable Diffusion WebUI, которое позволяет делать простую и точную замену лиц на изображениях. Сделано на основе <a href="https://github.com/Gourieff/ReActor-UI" target="_blank">ReActor-UI</a>.
|
||||
</td>
|
||||
<td width="144px">
|
||||
<a href="https://paypal.me/artgourieff" target="_blank">
|
||||
<img src="https://www.paypalobjects.com/digitalassets/c/website/logo/full-text/pp_fc_hl.svg" width="108" alt="Поддержать проект через PayPal"/>
|
||||
<br>
|
||||
<sup>
|
||||
Помочь проекту
|
||||
</sup>
|
||||
</a>
|
||||
</td>
|
||||
</tr>
|
||||
</table>
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/raw/main/sd-webui-reactor/demo_crop.jpg?raw=true" alt="example"/>
|
||||
|
||||
<a name="latestupdate">
|
||||
|
||||
## Что нового в последних обновлениях
|
||||
|
||||
<details>
|
||||
<summary><a>Нажмите, чтобы посмотреть</a></summary>
|
||||
|
||||
### 0.7.0 <sub><sup>BETA2
|
||||
|
||||
- X/Y/Z опция улучшена! Добавлен ещё один параметр: теперь вы можете выбрать несколько моделей лиц для создания вариации замененных лиц, чтобы выбрать наилучшие!
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-05.jpg?raw=true" alt="0.7.0-whatsnew-05" width="100%"/>
|
||||
|
||||
Чтобы использовать ось "Face Model" - активируйте РеАктор и выбирите любую модель лица в качестве источника:<br>
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-07.jpg?raw=true" alt="0.7.0-whatsnew-07" width="50%"/><img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-06.jpg?raw=true" alt="0.7.0-whatsnew-06" width="50%"/>
|
||||
|
||||
Полноразмерное демо-изображение: [xyz_demo_2.png](https://raw.githubusercontent.com/Gourieff/Assets/main/sd-webui-reactor/xyz_demo_2.png)
|
||||
|
||||
### 0.7.0 <sub><sup>BETA1
|
||||
|
||||
- Поддержка X/Y/Z скрипта (до 3-х параметров: CodeFormer Weight, Restorer Visibility, Face Mask Correction)
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-03.jpg?raw=true" alt="0.7.0-whatsnew-03" width="100%"/>
|
||||
|
||||
Полноразмерное демо-изображение: [xyz_demo.png](https://raw.githubusercontent.com/Gourieff/Assets/main/sd-webui-reactor/xyz_demo.png)
|
||||
|
||||
### 0.7.0 <sub><sup>ALPHA1
|
||||
|
||||
- По многочисленным просьбам появилась возможность строить смешанные модели лиц ("Tools->Face Models->Blend")
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-01.jpg?raw=true" alt="0.7.0-whatsnew-01" width="100%"/><img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/0.7.0-whatsnew-02.jpg?raw=true" alt="0.7.0-whatsnew-02" width="100%"/>
|
||||
|
||||
- Поддержка CUDA 12 в скрипте установщика для библиотеки ORT-GPU версии 1.17.0
|
||||
- Новая вкладка "Detection" с параметрами "Threshold" и "Max Faces"
|
||||
|
||||
### 0.6.1 <sub><sup>BETA3
|
||||
|
||||
- Опция 'Force Upscale' внутри вкладки 'Upscale': апскейл выполнится, даже если не было обнаружено ни одного лица (FR https://github.com/Gourieff/sd-webui-reactor/issues/116)
|
||||
- Отображение имён файлов используемых изображений, когда выбрано несколько изображений или папка (а также режим случайного изображения)
|
||||
|
||||
### 0.6.1 <sub><sup>BETA2
|
||||
|
||||
- Опция 'Save original' теперь работает правильно, когда вы выбираете 'Multiple Images' или 'Source Folder'
|
||||
- Добавлен режим выбора случайного изображения для 'Source Folder'
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/random_from_folder_demo_01.jpg?raw=true" alt="0.6.1-whatsnew-01" width="100%"/>
|
||||
|
||||
### 0.6.0
|
||||
|
||||
- Новый логотип
|
||||
- Адаптация к версии A1111 1.7.0 (правильная загрузка GFPGAN)
|
||||
- Новая ссылка для файла основной модели
|
||||
- UI переработан
|
||||
- Появилась возможность загружать несколько исходных изображений с лицами или задавать путь к папке, содержащей такие изображения
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/multiple_source_images_demo_01.png?raw=true" alt="0.6.0-whatsnew-01" width="100%"/>
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/multiple_source_images_demo_02.png?raw=true" alt="0.6.0-whatsnew-02" width="100%"/>
|
||||
|
||||
### 0.5.1
|
||||
|
||||
- Теперь можно сохранять модели лиц в качестве файлов "safetensors" (находятся в `<sd-web-ui-folder>\models\reactor\faces`) и загружать их с ReActor, храня супер легкие модели лиц, которые вы чаще всего используете;
|
||||
- Новые опция "Face Mask Correction" - если вы сталкиваетесь с пикселизацией вокруг контуров лица, эта опция будет полезной;
|
||||
|
||||
<img src="https://github.com/Gourieff/Assets/blob/main/sd-webui-reactor/face_model_demo_01.jpg?raw=true" alt="0.5.0-whatsnew-01" width="100%"/>
|
||||
|
||||
</details>
|
||||
|
||||
<a name="installation">
|
||||
|
||||
## Установка
|
||||
|
||||
[Automatic1111](#a1111) | [Vladmandic SD.Next](#sdnext) | [Google Colab SD WebUI](#colab)
|
||||
[A1111 WebUI / WebUI-Forge](#a1111) | [SD.Next](#sdnext) | [Google Colab SD WebUI](#colab)
|
||||
|
||||
<a name="a1111">Если вы используете [AUTOMATIC1111 Web-UI](https://github.com/AUTOMATIC1111/stable-diffusion-webui/):
|
||||
<a name="a1111">Если вы используете [AUTOMATIC1111 SD WebUI](https://github.com/AUTOMATIC1111/stable-diffusion-webui/) или [SD WebUI Forge](https://github.com/lllyasviel/stable-diffusion-webui-forge):
|
||||
|
||||
1. (Для пользователей Windows):
|
||||
- Установите **Visual Studio 2022** (Например, версию Community - этот шаг нужен для правильной компиляции библиотеки Insightface):
|
||||
@ -69,8 +117,8 @@
|
||||
- ИЛИ только **VS C++ Build Tools** (если вам не нужен весь пакет Visual Studio), выберите "Desktop Development with C++" в разделе "Workloads -> Desktop & Mobile":
|
||||
https://visualstudio.microsoft.com/visual-cpp-build-tools/
|
||||
- ИЛИ если же вы не хотите устанавливать что-либо из вышеуказанного - выполните [следующие шаги (пункт VIII)](#insightfacebuild)
|
||||
2. Внутри SD Web-UI перейдите во вкладку "Extensions" и вставьте ссылку `https://github.com/Gourieff/sd-webui-reactor` в "Install from URL" и нажмите "Install"
|
||||
3. Пожалуйста, подождите несколько минут, пока процесс установки полностью не завершится
|
||||
2. Внутри SD Web-UI перейдите во вкладку "Extensions", загрузите список доступных расширений (вкладка "Available") и введите "ReActor" в строке поиска или же вставьте ссылку `https://github.com/Gourieff/sd-webui-reactor` в "Install from URL" - и нажмите "Install"
|
||||
3. Пожалуйста, подождите несколько минут, пока процесс установки полностью не завершится (наберитесь терпения, не прерывайте процесс)
|
||||
4. Проверьте последнее сообщение в консоли SD-WebUI:
|
||||
* Если вы видите "--- PLEASE, RESTART the Server! ---" - остановите Сервер (CTRL+C или CMD+C) и запустите его заново - ИЛИ же перейдите во вкладку "Installed", нажмите "Apply and restart UI"
|
||||
* Если вы видите "Done!", просто перезагрузите UI, нажав на "Reload UI"
|
||||
@ -83,7 +131,7 @@
|
||||
3. Перейдите в (Windows)`automatic\venv\Scripts` или (MacOS/Linux)`automatic/venv/bin`, запустите Терминал или Консоль (cmd) для данной папки и выполните `activate`
|
||||
4. Выполните `pip install insightface==0.7.3`
|
||||
5. Запустите SD.Next, перейдите во вкладку "Extensions", вставьте эту ссылку `https://github.com/Gourieff/sd-webui-reactor` в "Install from URL" и нажмите "Install"
|
||||
6. Пожалуйста, подождите несколько минут, пока процесс установки полностью не завершится
|
||||
6. Пожалуйста, подождите несколько минут, пока процесс установки полностью не завершится (наберитесь терпения, не прерывайте процесс)
|
||||
7. Проверьте последнее сообщение в консоли SD.Next:
|
||||
* Если вы видите "--- PLEASE, RESTART the Server! ---" - остановите Сервер (CTRL+C или CMD+C) или просто закройте консоль
|
||||
8. Перейдите в директорию `automatic\extensions\sd-webui-reactor` - если вы видите там папку `models\insightface` с файлом `inswapper_128.onnx` внутри, переместите его в папку `automatic\models\insightface`
|
||||
@ -91,8 +139,8 @@
|
||||
|
||||
<a name="colab">Если вы используете [Cagliostro Colab UI](https://github.com/Linaqruf/sd-notebook-collection):
|
||||
|
||||
1. В активном WebUI, перейдите во вкладку "Extensions", вставьте ссылку `https://github.com/Gourieff/sd-webui-reactor` в "Install from URL" и нажмите "Install"
|
||||
2. Пожалуйста, подождите некоторое время, пока процесс установки полностью не завершится
|
||||
1. В активном WebUI перейдите во вкладку "Extensions", загрузите список доступных расширений (вкладка "Available") и введите "ReActor" в строке поиска или же вставьте ссылку `https://github.com/Gourieff/sd-webui-reactor` в "Install from URL" - и нажмите "Install"
|
||||
2. Пожалуйста, подождите некоторое время, пока процесс установки полностью не завершится (наберитесь терпения, не прерывайте процесс)
|
||||
3. Когда вы увидите сообщение "--- PLEASE, RESTART the Server! ---" (в секции "Start UI" вашего ноутбука "Start Cagliostro Colab UI") - перейдите во вкладку "Installed" и нажмите "Apply and restart UI"
|
||||
4. Готово!
|
||||
|
||||
@ -106,6 +154,8 @@
|
||||
- Функция **сохранения оригинального изображения** (сгенерированного до замены лица)
|
||||
- **Восстановление лица** после замены
|
||||
- **Увеличение размера** полученного изображения
|
||||
- Сохранение и загрузка **Моделей Лиц типа Safetensors**
|
||||
- **Коррекция Маски Лица** для предотвращения какой-либо пикселизации вокруг контуров лиц
|
||||
- Возможность задать **порядок постобработки**
|
||||
- **100% совместимость** с разными **SD WebUI**: Automatic1111, SD.Next, Cagliostro Colab UI
|
||||
- **Отличная производительность** даже с использованием ЦПУ, ReActor для SD WebUI абсолютно не требователен к мощности вашей видеокарты
|
||||
@ -205,7 +255,7 @@ Inpainting также работает, но замена лица происх
|
||||
7. Далее:
|
||||
- `pip install insightface==0.7.3`
|
||||
- `pip install onnx`
|
||||
- `pip install onnxruntime-gpu>=1.16.1`
|
||||
- `pip install "onnxruntime-gpu>=1.16.1"`
|
||||
- `pip install opencv-python`
|
||||
- `pip install tqdm`
|
||||
8. Выполните `deactivate`, закройте Терминал или Консоль и запустите SD WebUI, ReActor должен запуститься без к-л проблем - если же нет, добро пожаловать в раздел "Issues".
|
||||
@ -229,7 +279,7 @@ Inpainting также работает, но замена лица происх
|
||||
|
||||
Эта ошибка появляется, если что-то не так с файлом модели `inswapper_128.onnx`.
|
||||
|
||||
Скачайте вручную по ссылке [here](https://github.com/facefusion/facefusion-assets/releases/download/models/inswapper_128.onnx)
|
||||
Скачайте вручную по ссылке [here](https://huggingface.co/datasets/Gourieff/ReActor/resolve/main/models/inswapper_128.onnx)
|
||||
и сохраните в директорию `stable-diffusion-webui\models\insightface`, заменив имеющийся файл.
|
||||
|
||||
### **VI. "ValueError: This ORT build has ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'] enabled" ИЛИ "ValueError: This ORT build has ['AzureExecutionProvider', 'CPUExecutionProvider'] enabled"**
|
||||
@ -240,7 +290,7 @@ Inpainting также работает, но замена лица происх
|
||||
4. Затем:
|
||||
- `python -m pip install -U pip`
|
||||
- `pip uninstall -y onnxruntime onnxruntime-gpu onnxruntime-silicon onnxruntime-extensions`
|
||||
- `pip install onnxruntime-gpu>=1.16.1`
|
||||
- `pip install "onnxruntime-gpu>=1.16.1"`
|
||||
|
||||
Если это не помогло - значит какое-то другое расширение переустанавливает `onnxruntime` всякий раз, когда SD WebUI проверяет требования пакетов. Внимательно посмотрите список активных расширений. Некоторые расширения могут вызывать переустановку `onnxruntime-gpu` на версию `onnxruntime<1.16.1` при каждом запуске SD WebUI.<br>ORT 1.16.0 выкатили с ошибкой https://github.com/microsoft/onnxruntime/issues/17631 - не устанавливайте её!
|
||||
|
||||
@ -253,7 +303,7 @@ Inpainting также работает, но замена лица происх
|
||||
5. Затем:
|
||||
- `python -m pip install -U pip`
|
||||
- `pip uninstall protobuf`
|
||||
- `pip install protobuf>=3.20.3`
|
||||
- `pip install "protobuf>=3.20.3"`
|
||||
|
||||
Если это не помгло - значит, есть к-л другое расширение, которое использует неподходящую версию пакета protobuf, и SD WebUI устанавливает эту версию при каждом запуске.
|
||||
|
||||
@ -262,10 +312,10 @@ Inpainting также работает, но замена лица происх
|
||||
### **VIII. (Для пользователей Windows) Если вы до сих пор не можете установить пакет Insightface по каким-то причинам или же просто не желаете устанавливать Visual Studio или VS C++ Build Tools - сделайте следующее:**
|
||||
|
||||
1. Закройте (остановите) SD WebUI Сервер, если он запущен
|
||||
2. Скачайте готовый [пакет Insightface](https://github.com/Gourieff/sd-webui-reactor/raw/main/example/insightface-0.7.3-cp310-cp310-win_amd64.whl) и сохраните его в корневую директорию stable-diffusion-webui (или SD.Next) - туда, где лежит файл "webui-user.bat"
|
||||
3. Из корневой директории откройте Консоль (CMD) и выполните `.\venv\Scripts\activate`
|
||||
4. Обновите PIP: `python -m pip install -U pip`
|
||||
5. Затем установите Insightface: `pip install insightface-0.7.3-cp310-cp310-win_amd64.whl`
|
||||
2. Скачайте готовый [пакет Insightface](https://github.com/Gourieff/Assets/raw/main/Insightface/insightface-0.7.3-cp310-cp310-win_amd64.whl) и сохраните его в корневую директорию stable-diffusion-webui (или SD.Next) - туда, где лежит файл "webui-user.bat" или (A1111 Portable) "run.bat"
|
||||
3. Из корневой директории откройте Консоль (CMD) и выполните `.\venv\Scripts\activate`<br>ИЛИ<br>(A1111 Portable) Откройте Консоль (CMD)
|
||||
4. Обновите PIP: `python -m pip install -U pip`<br>ИЛИ<br>(A1111 Portable)`system\python\python.exe -m pip install -U pip`
|
||||
5. Затем установите Insightface: `pip install insightface-0.7.3-cp310-cp310-win_amd64.whl`<br>ИЛИ<br>(A1111 Portable)`system\python\python.exe -m pip install insightface-0.7.3-cp310-cp310-win_amd64.whl`
|
||||
6. Готово!
|
||||
|
||||
### **IX. Ошибка обновления 07-Август-23**
|
||||
|
||||
@ -17,7 +17,7 @@ finally:
|
||||
print(im)
|
||||
|
||||
img_bytes = io.BytesIO()
|
||||
im.save(img_bytes, format='PNG')
|
||||
im.save(img_bytes, format='PNG')
|
||||
img_base64 = base64.b64encode(img_bytes.getvalue()).decode('utf-8')
|
||||
|
||||
# ReActor arguments:
|
||||
@ -31,7 +31,7 @@ args=[
|
||||
1, #5 Restore visibility value
|
||||
True, #7 Restore face -> Upscale
|
||||
'4x_NMKD-Superscale-SP_178000_G', #8 Upscaler (type 'None' if doesn't need), see full list here: http://127.0.0.1:7860/sdapi/v1/script-info -> reactor -> sec.8
|
||||
2, #9 Upscaler scale value
|
||||
1.5, #9 Upscaler scale value
|
||||
1, #10 Upscaler visibility (if scale = 1)
|
||||
False, #11 Swap in source image
|
||||
True, #12 Swap in generated image
|
||||
@ -43,6 +43,15 @@ args=[
|
||||
False, #18 Source Image Hash Check, True - by default
|
||||
False, #19 Target Image Hash Check, False - by default
|
||||
"CUDA", #20 CPU or CUDA (if you have it), CPU - by default
|
||||
True, #21 Face Mask Correction
|
||||
1, #22 Select Source, 0 - Image, 1 - Face Model, 2 - Source Folder
|
||||
"elena.safetensors", #23 Filename of the face model (from "models/reactor/faces"), e.g. elena.safetensors, don't forger to set #22 to 1
|
||||
"C:\PATH_TO_FACES_IMAGES", #24 The path to the folder containing source faces images, don't forger to set #22 to 2
|
||||
None, #25 skip it for API
|
||||
True, #26 Randomly select an image from the path
|
||||
True, #27 Force Upscale even if no face found
|
||||
0.6, #28 Face Detection Threshold
|
||||
2, #29 Maximum number of faces to detect (0 is unlimited)
|
||||
]
|
||||
|
||||
# The args for ReActor can be found by
|
||||
|
||||
@ -8,7 +8,7 @@ curl -X POST \
|
||||
"source_faces_index": [0],
|
||||
"face_index": [2],
|
||||
"upscaler": "None",
|
||||
"scale": 1,
|
||||
"scale": 1.5,
|
||||
"upscale_visibility": 1,
|
||||
"face_restorer": "CodeFormer",
|
||||
"restorer_visibility": 1,
|
||||
@ -19,5 +19,11 @@ curl -X POST \
|
||||
"gender_target": 0,
|
||||
"save_to_file": 1,
|
||||
"result_file_path": "",
|
||||
"device": "CUDA"
|
||||
"device": "CUDA",
|
||||
"mask_face": 1,
|
||||
"select_source": 1,
|
||||
"face_model": "elena.safetensors",
|
||||
"source_folder": "C:/faces",
|
||||
"random_image": 1,
|
||||
"upscale_force": 1
|
||||
}'
|
||||
|
||||
@ -4,7 +4,7 @@
|
||||
"source_faces_index": [0],
|
||||
"face_index": [2],
|
||||
"upscaler": "None",
|
||||
"scale": 1,
|
||||
"scale": 1.5,
|
||||
"upscale_visibility": 1,
|
||||
"face_restorer": "CodeFormer",
|
||||
"restorer_visibility": 1,
|
||||
@ -15,5 +15,11 @@
|
||||
"gender_target": 0,
|
||||
"save_to_file": 1,
|
||||
"result_file_path": "",
|
||||
"device": "CUDA"
|
||||
"device": "CUDA",
|
||||
"mask_face": 1,
|
||||
"select_source": 1,
|
||||
"face_model": "elena.safetensors",
|
||||
"source_folder": "C:/faces",
|
||||
"random_image": 1,
|
||||
"upscale_force": 1
|
||||
}
|
||||
Binary file not shown.
58
install.py
58
install.py
@ -12,30 +12,30 @@ except:
|
||||
try:
|
||||
from modules.paths import models_path
|
||||
except:
|
||||
model_path = os.path.abspath("models")
|
||||
models_path = os.path.abspath("models")
|
||||
|
||||
|
||||
BASE_PATH = os.path.dirname(os.path.realpath(__file__))
|
||||
|
||||
req_file = os.path.join(BASE_PATH, "requirements.txt")
|
||||
|
||||
models_dir_old = os.path.join(models_path, "roop")
|
||||
models_dir = os.path.join(models_path, "insightface")
|
||||
|
||||
# Check the deprecated 'roop' folder
|
||||
if os.path.exists(models_dir_old):
|
||||
if not os.listdir(models_dir_old) and (not os.listdir(models_dir) or not os.path.exists(models_dir)):
|
||||
os.rename(models_dir_old, models_dir)
|
||||
else:
|
||||
import shutil
|
||||
for file in os.listdir(models_dir_old):
|
||||
shutil.move(os.path.join(models_dir_old, file), os.path.join(models_dir, file))
|
||||
try:
|
||||
os.rmdir(models_dir_old)
|
||||
except Exception as e:
|
||||
print(f"OSError: {e}")
|
||||
# DEPRECATED:
|
||||
# models_dir_old = os.path.join(models_path, "roop")
|
||||
# if os.path.exists(models_dir_old):
|
||||
# if not os.listdir(models_dir_old) and (not os.listdir(models_dir) or not os.path.exists(models_dir)):
|
||||
# os.rename(models_dir_old, models_dir)
|
||||
# else:
|
||||
# import shutil
|
||||
# for file in os.listdir(models_dir_old):
|
||||
# shutil.move(os.path.join(models_dir_old, file), os.path.join(models_dir, file))
|
||||
# try:
|
||||
# os.rmdir(models_dir_old)
|
||||
# except Exception as e:
|
||||
# print(f"OSError: {e}")
|
||||
|
||||
model_url = "https://github.com/facefusion/facefusion-assets/releases/download/models/inswapper_128.onnx"
|
||||
model_url = "https://huggingface.co/datasets/Gourieff/ReActor/resolve/main/models/inswapper_128.onnx"
|
||||
model_name = os.path.basename(model_url)
|
||||
model_path = os.path.join(models_dir, model_name)
|
||||
|
||||
@ -75,16 +75,18 @@ if not os.path.exists(models_dir):
|
||||
if not os.path.exists(model_path):
|
||||
download(model_url, model_path)
|
||||
|
||||
print("ReActor preheating...", end=' ')
|
||||
# print("ReActor preheating...", end=' ')
|
||||
|
||||
last_device = None
|
||||
first_run = False
|
||||
available_devices = ["CPU", "CUDA"]
|
||||
|
||||
try:
|
||||
last_device_log = os.path.join(BASE_PATH, "last_device.txt")
|
||||
with open(last_device_log) as f:
|
||||
for el in f:
|
||||
last_device = el.strip()
|
||||
last_device = f.readline().strip()
|
||||
if last_device not in available_devices:
|
||||
last_device = None
|
||||
except:
|
||||
last_device = "CPU"
|
||||
first_run = True
|
||||
@ -95,31 +97,39 @@ with open(req_file) as file:
|
||||
install_count = 0
|
||||
ort = "onnxruntime-gpu"
|
||||
import torch
|
||||
cuda_version = None
|
||||
try:
|
||||
if torch.cuda.is_available():
|
||||
if first_run:
|
||||
cuda_version = torch.version.cuda
|
||||
print(f"CUDA {cuda_version}")
|
||||
if first_run or last_device is None:
|
||||
last_device = "CUDA"
|
||||
elif torch.backends.mps.is_available() or hasattr(torch,'dml'):
|
||||
elif torch.backends.mps.is_available() or hasattr(torch,'dml') or hasattr(torch,'privateuseone'):
|
||||
ort = "onnxruntime"
|
||||
# to prevent errors when ORT-GPU is installed but we want ORT instead:
|
||||
if first_run:
|
||||
pip_uninstall("onnxruntime", "onnxruntime-gpu")
|
||||
# just in case:
|
||||
if last_device == "CUDA":
|
||||
if last_device == "CUDA" or last_device is None:
|
||||
last_device = "CPU"
|
||||
else:
|
||||
if last_device == "CUDA":
|
||||
if last_device == "CUDA" or last_device is None:
|
||||
last_device = "CPU"
|
||||
with open(os.path.join(BASE_PATH, "last_device.txt"), "w") as txt:
|
||||
txt.write(last_device)
|
||||
if not is_installed(ort,"1.16.1",False):
|
||||
if cuda_version is not None and float(cuda_version)>=12: # CU12
|
||||
if not is_installed(ort,"1.17.1",False):
|
||||
install_count += 1
|
||||
pip_uninstall("onnxruntime", "onnxruntime-gpu")
|
||||
pip_install(ort,"--extra-index-url","https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/")
|
||||
elif not is_installed(ort,"1.16.1",False):
|
||||
install_count += 1
|
||||
pip_install(ort, "-U")
|
||||
except Exception as e:
|
||||
print(e)
|
||||
print(f"\nERROR: Failed to install {ort} - ReActor won't start")
|
||||
raise e
|
||||
print(f"Device: {last_device}")
|
||||
# print(f"Device: {last_device}")
|
||||
strict = True
|
||||
for package in file:
|
||||
package_version = None
|
||||
|
||||
176
reactor_modules/reactor_mask.py
Normal file
176
reactor_modules/reactor_mask.py
Normal file
@ -0,0 +1,176 @@
|
||||
import cv2
|
||||
import numpy as np
|
||||
from PIL import Image, ImageDraw
|
||||
|
||||
from torchvision.transforms.functional import to_pil_image
|
||||
|
||||
from scripts.reactor_logger import logger
|
||||
from scripts.reactor_inferencers.bisenet_mask_generator import BiSeNetMaskGenerator
|
||||
from scripts.reactor_entities.face import FaceArea
|
||||
from scripts.reactor_entities.rect import Rect
|
||||
|
||||
|
||||
colors = [
|
||||
(255, 0, 0),
|
||||
(0, 255, 0),
|
||||
(0, 0, 255),
|
||||
(255, 255, 0),
|
||||
(255, 0, 255),
|
||||
(0, 255, 255),
|
||||
(255, 255, 255),
|
||||
(128, 0, 0),
|
||||
(0, 128, 0),
|
||||
(128, 128, 0),
|
||||
(0, 0, 128),
|
||||
(0, 128, 128),
|
||||
]
|
||||
|
||||
def color_generator(colors):
|
||||
while True:
|
||||
for color in colors:
|
||||
yield color
|
||||
|
||||
|
||||
def process_face_image(
|
||||
face: FaceArea,
|
||||
**kwargs,
|
||||
) -> Image:
|
||||
image = np.array(face.image)
|
||||
overlay = image.copy()
|
||||
color_iter = color_generator(colors)
|
||||
cv2.rectangle(overlay, (0, 0), (image.shape[1], image.shape[0]), next(color_iter), -1)
|
||||
l, t, r, b = face.face_area_on_image
|
||||
cv2.rectangle(overlay, (l, t), (r, b), (0, 0, 0), 10)
|
||||
if face.landmarks_on_image is not None:
|
||||
for landmark in face.landmarks_on_image:
|
||||
cv2.circle(overlay, (int(landmark.x), int(landmark.y)), 6, (0, 0, 0), 10)
|
||||
alpha = 0.3
|
||||
output = cv2.addWeighted(image, 1 - alpha, overlay, alpha, 0)
|
||||
|
||||
return Image.fromarray(output)
|
||||
|
||||
|
||||
def apply_face_mask(swapped_image:np.ndarray,target_image:np.ndarray,target_face,entire_mask_image:np.array)->np.ndarray:
|
||||
logger.status("Correcting Face Mask")
|
||||
mask_generator = BiSeNetMaskGenerator()
|
||||
face = FaceArea(target_image,Rect.from_ndarray(np.array(target_face.bbox)),1.6,512,"")
|
||||
face_image = np.array(face.image)
|
||||
process_face_image(face)
|
||||
face_area_on_image = face.face_area_on_image
|
||||
mask = mask_generator.generate_mask(
|
||||
face_image,
|
||||
face_area_on_image=face_area_on_image,
|
||||
affected_areas=["Face"],
|
||||
mask_size=0,
|
||||
use_minimal_area=True
|
||||
)
|
||||
mask = cv2.blur(mask, (12, 12))
|
||||
# """entire_mask_image = np.zeros_like(target_image)"""
|
||||
larger_mask = cv2.resize(mask, dsize=(face.width, face.height))
|
||||
entire_mask_image[
|
||||
face.top : face.bottom,
|
||||
face.left : face.right,
|
||||
] = larger_mask
|
||||
|
||||
result = Image.composite(Image.fromarray(swapped_image),Image.fromarray(target_image), Image.fromarray(entire_mask_image).convert("L"))
|
||||
return np.array(result)
|
||||
|
||||
|
||||
def rotate_array(image: np.ndarray, angle: float) -> np.ndarray:
|
||||
if angle == 0:
|
||||
return image
|
||||
|
||||
h, w = image.shape[:2]
|
||||
center = (w // 2, h // 2)
|
||||
|
||||
M = cv2.getRotationMatrix2D(center, angle, 1.0)
|
||||
return cv2.warpAffine(image, M, (w, h))
|
||||
|
||||
|
||||
def rotate_image(image: Image, angle: float) -> Image:
|
||||
if angle == 0:
|
||||
return image
|
||||
return Image.fromarray(rotate_array(np.array(image), angle))
|
||||
|
||||
|
||||
def correct_face_tilt(angle: float) -> bool:
|
||||
angle = abs(angle)
|
||||
if angle > 180:
|
||||
angle = 360 - angle
|
||||
return angle > 40
|
||||
|
||||
|
||||
def _dilate(arr: np.ndarray, value: int) -> np.ndarray:
|
||||
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (value, value))
|
||||
return cv2.dilate(arr, kernel, iterations=1)
|
||||
|
||||
|
||||
def _erode(arr: np.ndarray, value: int) -> np.ndarray:
|
||||
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (value, value))
|
||||
return cv2.erode(arr, kernel, iterations=1)
|
||||
|
||||
|
||||
def dilate_erode(img: Image.Image, value: int) -> Image.Image:
|
||||
"""
|
||||
The dilate_erode function takes an image and a value.
|
||||
If the value is positive, it dilates the image by that amount.
|
||||
If the value is negative, it erodes the image by that amount.
|
||||
|
||||
Parameters
|
||||
----------
|
||||
img: PIL.Image.Image
|
||||
the image to be processed
|
||||
value: int
|
||||
kernel size of dilation or erosion
|
||||
|
||||
Returns
|
||||
-------
|
||||
PIL.Image.Image
|
||||
The image that has been dilated or eroded
|
||||
"""
|
||||
if value == 0:
|
||||
return img
|
||||
|
||||
arr = np.array(img)
|
||||
arr = _dilate(arr, value) if value > 0 else _erode(arr, -value)
|
||||
|
||||
return Image.fromarray(arr)
|
||||
|
||||
def mask_to_pil(masks, shape: tuple[int, int]) -> list[Image.Image]:
|
||||
"""
|
||||
Parameters
|
||||
----------
|
||||
masks: torch.Tensor, dtype=torch.float32, shape=(N, H, W).
|
||||
The device can be CUDA, but `to_pil_image` takes care of that.
|
||||
|
||||
shape: tuple[int, int]
|
||||
(width, height) of the original image
|
||||
"""
|
||||
n = masks.shape[0]
|
||||
return [to_pil_image(masks[i], mode="L").resize(shape) for i in range(n)]
|
||||
|
||||
def create_mask_from_bbox(
|
||||
bboxes: list[list[float]], shape: tuple[int, int]
|
||||
) -> list[Image.Image]:
|
||||
"""
|
||||
Parameters
|
||||
----------
|
||||
bboxes: list[list[float]]
|
||||
list of [x1, y1, x2, y2]
|
||||
bounding boxes
|
||||
shape: tuple[int, int]
|
||||
shape of the image (width, height)
|
||||
|
||||
Returns
|
||||
-------
|
||||
masks: list[Image.Image]
|
||||
A list of masks
|
||||
|
||||
"""
|
||||
masks = []
|
||||
for bbox in bboxes:
|
||||
mask = Image.new("L", shape, 0)
|
||||
mask_draw = ImageDraw.Draw(mask)
|
||||
mask_draw.rectangle(bbox, fill=255)
|
||||
masks.append(mask)
|
||||
return masks
|
||||
5
reactor_ui/__init__.py
Normal file
5
reactor_ui/__init__.py
Normal file
@ -0,0 +1,5 @@
|
||||
import reactor_ui.reactor_upscale_ui as ui_upscale
|
||||
import reactor_ui.reactor_tools_ui as ui_tools
|
||||
import reactor_ui.reactor_settings_ui as ui_settings
|
||||
import reactor_ui.reactor_main_ui as ui_main
|
||||
import reactor_ui.reactor_detection_ui as ui_detection
|
||||
54
reactor_ui/reactor_detection_ui.py
Normal file
54
reactor_ui/reactor_detection_ui.py
Normal file
@ -0,0 +1,54 @@
|
||||
import gradio as gr
|
||||
from scripts.reactor_swapper import (
|
||||
clear_faces,
|
||||
clear_faces_list,
|
||||
clear_faces_target,
|
||||
clear_faces_all
|
||||
)
|
||||
|
||||
# TAB DETECTION
|
||||
def show(show_br: bool = True):
|
||||
with gr.Tab("Detection"):
|
||||
with gr.Row():
|
||||
det_thresh = gr.Slider(
|
||||
minimum=0.1,
|
||||
maximum=1.0,
|
||||
value=0.5,
|
||||
step=0.01,
|
||||
label="Threshold",
|
||||
info="The higher the value, the more sensitive the detection is to what is considered a face (0.5 by default)",
|
||||
scale=2
|
||||
)
|
||||
det_maxnum = gr.Slider(
|
||||
minimum=0,
|
||||
maximum=20,
|
||||
value=0,
|
||||
step=1,
|
||||
label="Max Faces",
|
||||
info="Maximum number of faces to detect (0 is unlimited)",
|
||||
scale=1
|
||||
)
|
||||
# gr.Markdown("<br>", visible=show_br)
|
||||
gr.Markdown("Hashed images get processed with previously set detection parameters (the face is hashed with all available parameters to bypass the analyzer and speed up the process). Please clear the hash if you want to apply new detection settings.", visible=show_br)
|
||||
with gr.Row():
|
||||
imgs_hash_clear_single = gr.Button(
|
||||
value="Clear Source Images Hash (Single)",
|
||||
scale=1
|
||||
)
|
||||
imgs_hash_clear_multiple = gr.Button(
|
||||
value="Clear Source Images Hash (Multiple)",
|
||||
scale=1
|
||||
)
|
||||
imgs_hash_clear_target = gr.Button(
|
||||
value="Clear Target Image Hash",
|
||||
scale=1
|
||||
)
|
||||
imgs_hash_clear_all = gr.Button(
|
||||
value="Clear All Hash"
|
||||
)
|
||||
progressbar_area = gr.Markdown("")
|
||||
imgs_hash_clear_single.click(clear_faces,None,[progressbar_area])
|
||||
imgs_hash_clear_multiple.click(clear_faces_list,None,[progressbar_area])
|
||||
imgs_hash_clear_target.click(clear_faces_target,None,[progressbar_area])
|
||||
imgs_hash_clear_all.click(clear_faces_all,None,[progressbar_area])
|
||||
return det_thresh, det_maxnum
|
||||
192
reactor_ui/reactor_main_ui.py
Normal file
192
reactor_ui/reactor_main_ui.py
Normal file
@ -0,0 +1,192 @@
|
||||
import gradio as gr
|
||||
from scripts.reactor_helpers import (
|
||||
get_model_names,
|
||||
get_facemodels
|
||||
)
|
||||
from scripts.reactor_swapper import (
|
||||
clear_faces_list,
|
||||
)
|
||||
from modules import shared
|
||||
|
||||
# SAVE_ORIGINAL: bool = False
|
||||
|
||||
def update_fm_list(selected: str):
|
||||
return gr.Dropdown.update(
|
||||
value=selected, choices=get_model_names(get_facemodels)
|
||||
)
|
||||
|
||||
# TAB MAIN
|
||||
def show(is_img2img: bool, show_br: bool = True, **msgs):
|
||||
|
||||
# def on_select_source(selected: bool, evt: gr.SelectData):
|
||||
def on_select_source(evt: gr.SelectData):
|
||||
# global SAVE_ORIGINAL
|
||||
if evt.index == 2:
|
||||
# if SAVE_ORIGINAL != selected:
|
||||
# SAVE_ORIGINAL = selected
|
||||
return {
|
||||
control_col_1: gr.Column.update(visible=False),
|
||||
control_col_2: gr.Column.update(visible=False),
|
||||
control_col_3: gr.Column.update(visible=True),
|
||||
# save_original: gr.Checkbox.update(value=False,visible=False),
|
||||
imgs_hash_clear: gr.Button.update(visible=True)
|
||||
}
|
||||
if evt.index == 0:
|
||||
return {
|
||||
control_col_1: gr.Column.update(visible=True),
|
||||
control_col_2: gr.Column.update(visible=False),
|
||||
control_col_3: gr.Column.update(visible=False),
|
||||
# save_original: gr.Checkbox.update(value=SAVE_ORIGINAL,visible=show_br),
|
||||
imgs_hash_clear: gr.Button.update(visible=False)
|
||||
}
|
||||
if evt.index == 1:
|
||||
return {
|
||||
control_col_1: gr.Column.update(visible=False),
|
||||
control_col_2: gr.Column.update(visible=True),
|
||||
control_col_3: gr.Column.update(visible=False),
|
||||
# save_original: gr.Checkbox.update(value=SAVE_ORIGINAL,visible=show_br),
|
||||
imgs_hash_clear: gr.Button.update(visible=False)
|
||||
}
|
||||
|
||||
progressbar_area = gr.Markdown("")
|
||||
with gr.Tab("Main"):
|
||||
with gr.Column():
|
||||
with gr.Row():
|
||||
select_source = gr.Radio(
|
||||
["Image(s)","Face Model","Folder"],
|
||||
value="Image(s)",
|
||||
label="Select Source",
|
||||
type="index",
|
||||
scale=1,
|
||||
)
|
||||
with gr.Column(visible=False) as control_col_2:
|
||||
with gr.Row():
|
||||
face_models = get_model_names(get_facemodels)
|
||||
face_model = gr.Dropdown(
|
||||
choices=face_models,
|
||||
label="Choose Face Model",
|
||||
value="None",
|
||||
scale=1,
|
||||
)
|
||||
fm_update = gr.Button(
|
||||
value="🔄",
|
||||
variant="tool",
|
||||
)
|
||||
fm_update.click(
|
||||
update_fm_list,
|
||||
inputs=[face_model],
|
||||
outputs=[face_model],
|
||||
)
|
||||
imgs_hash_clear = gr.Button(
|
||||
value="Clear Source Images Hash",
|
||||
scale=1,
|
||||
visible=False,
|
||||
)
|
||||
imgs_hash_clear.click(clear_faces_list,None,[progressbar_area])
|
||||
gr.Markdown("<br>", visible=show_br)
|
||||
with gr.Column(visible=True) as control_col_1:
|
||||
gr.Markdown("<center>🔽🔽🔽 Single Image has priority when both Areas in use 🔽🔽🔽</center>")
|
||||
with gr.Row():
|
||||
img = gr.Image(
|
||||
type="pil",
|
||||
label="Single Source Image",
|
||||
)
|
||||
imgs = gr.Files(
|
||||
label=f"Multiple Source Images{msgs['extra_multiple_source']}",
|
||||
file_types=["image"],
|
||||
)
|
||||
with gr.Column(visible=False) as control_col_3:
|
||||
gr.Markdown("<span style='display:block;text-align:right;padding-right:3px;margin: -15px 0;font-size:1.1em'><sup>Clear Hash if you see the previous face was swapped instead of the new one</sup></span>")
|
||||
with gr.Row():
|
||||
source_folder = gr.Textbox(
|
||||
value="",
|
||||
placeholder="Paste here the path to the folder containing source faces images",
|
||||
label=f"Source Folder{msgs['extra_multiple_source']}",
|
||||
scale=2,
|
||||
)
|
||||
random_image = gr.Checkbox(
|
||||
False,
|
||||
label="Random Image",
|
||||
info="Randomly select an image from the path",
|
||||
scale=1,
|
||||
)
|
||||
setattr(face_model, "do_not_save_to_config", True)
|
||||
if is_img2img:
|
||||
save_original = gr.Checkbox(
|
||||
False,
|
||||
label="Save Original (Swap in generated only)",
|
||||
info="Save the original image(s) made before swapping"
|
||||
)
|
||||
else:
|
||||
save_original = gr.Checkbox(
|
||||
False,
|
||||
label="Save Original",
|
||||
info="Save the original image(s) made before swapping",
|
||||
visible=show_br
|
||||
)
|
||||
# imgs.upload(on_files_upload_uncheck_so,[save_original],[save_original],show_progress=False)
|
||||
# imgs.clear(on_files_clear,None,[save_original],show_progress=False)
|
||||
imgs.clear(clear_faces_list,None,None,show_progress=False)
|
||||
mask_face = gr.Checkbox(
|
||||
False,
|
||||
label="Face Mask Correction",
|
||||
info="Apply this option if you see some pixelation around face contours"
|
||||
)
|
||||
gr.Markdown("<br>", visible=show_br)
|
||||
gr.Markdown("Source Image (above):")
|
||||
with gr.Row():
|
||||
source_faces_index = gr.Textbox(
|
||||
value="0",
|
||||
placeholder="Which face(s) to use as Source (comma separated)",
|
||||
label="Comma separated face number(s); Example: 0,2,1",
|
||||
)
|
||||
gender_source = gr.Radio(
|
||||
["No", "Female Only", "Male Only"],
|
||||
value="No",
|
||||
label="Gender Detection (Source)",
|
||||
type="index",
|
||||
)
|
||||
gr.Markdown("<br>", visible=show_br)
|
||||
gr.Markdown("Target Image (result):")
|
||||
with gr.Row():
|
||||
faces_index = gr.Textbox(
|
||||
value="0",
|
||||
placeholder="Which face(s) to Swap into Target (comma separated)",
|
||||
label="Comma separated face number(s); Example: 1,0,2",
|
||||
)
|
||||
gender_target = gr.Radio(
|
||||
["No", "Female Only", "Male Only"],
|
||||
value="No",
|
||||
label="Gender Detection (Target)",
|
||||
type="index",
|
||||
)
|
||||
gr.Markdown("<br>", visible=show_br)
|
||||
with gr.Row():
|
||||
face_restorer_name = gr.Radio(
|
||||
label="Restore Face",
|
||||
choices=["None"] + [x.name() for x in shared.face_restorers],
|
||||
value=shared.face_restorers[0].name(),
|
||||
type="value",
|
||||
)
|
||||
with gr.Column():
|
||||
face_restorer_visibility = gr.Slider(
|
||||
0, 1, 1, step=0.1, label="Restore Face Visibility"
|
||||
)
|
||||
codeformer_weight = gr.Slider(
|
||||
0, 1, 0.5, step=0.1, label="CodeFormer Weight (Fidelity)", info="0 = far from original (max restoration), 1 = close to original (min restoration)"
|
||||
)
|
||||
gr.Markdown("<br>", visible=show_br)
|
||||
swap_in_source = gr.Checkbox(
|
||||
False,
|
||||
label="Swap in source image",
|
||||
visible=is_img2img,
|
||||
)
|
||||
swap_in_generated = gr.Checkbox(
|
||||
True,
|
||||
label="Swap in generated image",
|
||||
visible=is_img2img,
|
||||
)
|
||||
# select_source.select(on_select_source,[save_original],[control_col_1,control_col_2,control_col_3,save_original,imgs_hash_clear],show_progress=False)
|
||||
select_source.select(on_select_source,None,[control_col_1,control_col_2,control_col_3,imgs_hash_clear],show_progress=False)
|
||||
|
||||
return img, imgs, select_source, face_model, source_folder, save_original, mask_face, source_faces_index, gender_source, faces_index, gender_target, face_restorer_name, face_restorer_visibility, codeformer_weight, swap_in_source, swap_in_generated, random_image
|
||||
77
reactor_ui/reactor_settings_ui.py
Normal file
77
reactor_ui/reactor_settings_ui.py
Normal file
@ -0,0 +1,77 @@
|
||||
import gradio as gr
|
||||
from scripts.reactor_logger import logger
|
||||
from scripts.reactor_helpers import get_models, set_Device
|
||||
from scripts.reactor_globals import DEVICE, DEVICE_LIST
|
||||
try:
|
||||
import torch.cuda as cuda
|
||||
EP_is_visible = True if cuda.is_available() else False
|
||||
except:
|
||||
EP_is_visible = False
|
||||
|
||||
def update_models_list(selected: str):
|
||||
return gr.Dropdown.update(
|
||||
value=selected, choices=get_models()
|
||||
)
|
||||
|
||||
def show(hash_check_block: bool = True):
|
||||
# TAB SETTINGS
|
||||
with gr.Tab("Settings"):
|
||||
models = get_models()
|
||||
with gr.Row(visible=EP_is_visible):
|
||||
device = gr.Radio(
|
||||
label="Execution Provider",
|
||||
choices=DEVICE_LIST,
|
||||
value=DEVICE,
|
||||
type="value",
|
||||
info="Click 'Save' to apply. If you already run 'Generate' - RESTART is required: (A1111) Extensions Tab -> 'Apply and restart UI' or (SD.Next) close the Server and start it again",
|
||||
scale=2,
|
||||
)
|
||||
save_device_btn = gr.Button("Save", scale=0)
|
||||
save = gr.Markdown("", visible=EP_is_visible)
|
||||
setattr(device, "do_not_save_to_config", True)
|
||||
save_device_btn.click(
|
||||
set_Device,
|
||||
inputs=[device],
|
||||
outputs=[save],
|
||||
)
|
||||
with gr.Row():
|
||||
if len(models) == 0:
|
||||
logger.warning(
|
||||
"You should at least have one model in models directory, please read the doc here: https://github.com/Gourieff/sd-webui-reactor/"
|
||||
)
|
||||
model = gr.Dropdown(
|
||||
choices=models,
|
||||
label="Model not found, please download one and refresh the list"
|
||||
)
|
||||
else:
|
||||
model = gr.Dropdown(
|
||||
choices=models, label="Model", value=models[0]
|
||||
)
|
||||
models_update = gr.Button(
|
||||
value="🔄",
|
||||
variant="tool",
|
||||
)
|
||||
models_update.click(
|
||||
update_models_list,
|
||||
inputs=[model],
|
||||
outputs=[model],
|
||||
)
|
||||
console_logging_level = gr.Radio(
|
||||
["No log", "Minimum", "Default"],
|
||||
value="Minimum",
|
||||
label="Console Log Level",
|
||||
type="index"
|
||||
)
|
||||
gr.Markdown("<br>", visible=hash_check_block)
|
||||
with gr.Row(visible=hash_check_block):
|
||||
source_hash_check = gr.Checkbox(
|
||||
True,
|
||||
label="Source Image Hash Check",
|
||||
info="Recommended to keep it ON. Processing is faster when Source Image is the same."
|
||||
)
|
||||
target_hash_check = gr.Checkbox(
|
||||
False,
|
||||
label="Target Image Hash Check",
|
||||
info="Affects if you use Extras tab or img2img with only 'Swap in source image' on."
|
||||
)
|
||||
return model, device, console_logging_level, source_hash_check, target_hash_check
|
||||
61
reactor_ui/reactor_tools_ui.py
Normal file
61
reactor_ui/reactor_tools_ui.py
Normal file
@ -0,0 +1,61 @@
|
||||
import gradio as gr
|
||||
from scripts.reactor_swapper import build_face_model, blend_faces
|
||||
|
||||
# TAB TOOLS
|
||||
def show():
|
||||
with gr.Tab("Tools"):
|
||||
with gr.Tab("Face Models"):
|
||||
|
||||
with gr.Tab("Single"):
|
||||
gr.Markdown("Load an image containing one person, name it and click 'Build and Save'")
|
||||
img_fm = gr.Image(
|
||||
type="pil",
|
||||
label="Load an Image to build -Face Model-",
|
||||
)
|
||||
with gr.Row(equal_height=True):
|
||||
fm_name = gr.Textbox(
|
||||
value="",
|
||||
placeholder="Please type any name (e.g. Elena)",
|
||||
label="Face Model Name",
|
||||
)
|
||||
save_fm_btn = gr.Button("Build and Save")
|
||||
save_fm = gr.Markdown("You can find saved models in 'models/reactor/faces'")
|
||||
save_fm_btn.click(
|
||||
build_face_model,
|
||||
inputs=[img_fm, fm_name],
|
||||
outputs=[save_fm],
|
||||
)
|
||||
|
||||
with gr.Tab("Blend"):
|
||||
gr.Markdown("Load a set of images containing any person, name it and click 'Build and Save'")
|
||||
with gr.Row():
|
||||
imgs_fm = gr.Files(
|
||||
label=f"Load Images to build -Blended Face Model-",
|
||||
file_types=["image"]
|
||||
)
|
||||
with gr.Column():
|
||||
compute_method = gr.Radio(
|
||||
["Mean", "Median", "Mode"],
|
||||
value="Mean",
|
||||
label="Compute Method",
|
||||
type="index",
|
||||
info="Mean (recommended) - Average value (best result 👍); Median* - Mid-point value (may be funny 😅); Mode - Most common value (may be scary 😨); *Mean and Median will be simillar if you load two images"
|
||||
)
|
||||
shape_check = gr.Checkbox(
|
||||
False,
|
||||
label="Check -Embedding Shape- on Similarity",
|
||||
info="(Experimental) Turn it ON if you want to skip the faces which are too much different from the first one in the list to prevent some probable 'shape mismatches'"
|
||||
)
|
||||
with gr.Row(equal_height=True):
|
||||
fm_name = gr.Textbox(
|
||||
value="",
|
||||
placeholder="Please type any name (e.g. Elena)",
|
||||
label="Face Model Name",
|
||||
)
|
||||
save_fm_btn = gr.Button("Build and Save")
|
||||
save_fm = gr.Markdown("You can find saved models in 'models/reactor/faces'")
|
||||
save_fm_btn.click(
|
||||
blend_faces,
|
||||
inputs=[imgs_fm, fm_name, compute_method, shape_check],
|
||||
outputs=[save_fm],
|
||||
)
|
||||
47
reactor_ui/reactor_upscale_ui.py
Normal file
47
reactor_ui/reactor_upscale_ui.py
Normal file
@ -0,0 +1,47 @@
|
||||
import gradio as gr
|
||||
from modules import shared
|
||||
|
||||
def update_upscalers_list(selected: str):
|
||||
return gr.Dropdown.update(
|
||||
value=selected, choices=[upscaler.name for upscaler in shared.sd_upscalers]
|
||||
)
|
||||
|
||||
# TAB UPSCALE
|
||||
def show(show_br: bool = True):
|
||||
with gr.Tab("Upscale"):
|
||||
with gr.Row():
|
||||
restore_first = gr.Checkbox(
|
||||
True,
|
||||
label="1. Restore Face -> 2. Upscale (-Uncheck- if you want vice versa)",
|
||||
info="Postprocessing Order",
|
||||
scale=2
|
||||
)
|
||||
upscale_force = gr.Checkbox(
|
||||
False,
|
||||
label="Force Upscale",
|
||||
info="Upscale anyway - even if no face found",
|
||||
scale=1
|
||||
)
|
||||
with gr.Row():
|
||||
upscaler_name = gr.Dropdown(
|
||||
choices=[upscaler.name for upscaler in shared.sd_upscalers],
|
||||
label="Upscaler",
|
||||
value="None",
|
||||
info="Won't scale if you choose -Swap in Source- via img2img, only 1x-postprocessing will affect (texturing, denoising, restyling etc.)"
|
||||
)
|
||||
upscalers_update = gr.Button(
|
||||
value="🔄",
|
||||
variant="tool",
|
||||
)
|
||||
upscalers_update.click(
|
||||
update_upscalers_list,
|
||||
inputs=[upscaler_name],
|
||||
outputs=[upscaler_name],
|
||||
)
|
||||
gr.Markdown("<br>", visible=show_br)
|
||||
with gr.Row():
|
||||
upscaler_scale = gr.Slider(1, 8, 1, step=0.1, label="Scale by")
|
||||
upscaler_visibility = gr.Slider(
|
||||
0, 1, 1, step=0.1, label="Upscaler Visibility (if scale = 1)"
|
||||
)
|
||||
return restore_first, upscaler_name, upscaler_scale, upscaler_visibility, upscale_force
|
||||
@ -1,12 +1,18 @@
|
||||
'''
|
||||
Thanks SpenserCai for the original version of the roop api script
|
||||
-----------------------------------
|
||||
--- ReActor External API v1.0.1 ---
|
||||
--- ReActor External API v1.0.7 ---
|
||||
-----------------------------------
|
||||
'''
|
||||
import os, glob
|
||||
from datetime import datetime, date
|
||||
from fastapi import FastAPI, Body
|
||||
# from fastapi.exceptions import HTTPException
|
||||
# from io import BytesIO
|
||||
# from PIL import Image
|
||||
# import base64
|
||||
# import numpy as np
|
||||
# import cv2
|
||||
|
||||
# from modules.api.models import *
|
||||
from modules import scripts, shared
|
||||
@ -14,8 +20,18 @@ from modules.api import api
|
||||
|
||||
import gradio as gr
|
||||
|
||||
from scripts.reactor_swapper import EnhancementOptions, swap_face
|
||||
from scripts.reactor_swapper import EnhancementOptions, swap_face, DetectionOptions
|
||||
from scripts.reactor_logger import logger
|
||||
from scripts.reactor_helpers import get_facemodels
|
||||
|
||||
# XYZ init:
|
||||
from scripts.reactor_xyz import run
|
||||
try:
|
||||
import modules.script_callbacks as script_callbacks
|
||||
script_callbacks.on_before_ui(run)
|
||||
# script_callbacks.on_app_started(reactor_api)
|
||||
except:
|
||||
pass
|
||||
|
||||
|
||||
def default_file_path():
|
||||
@ -52,6 +68,19 @@ def get_full_model(model_name):
|
||||
return model
|
||||
return None
|
||||
|
||||
# def decode_base64_to_image_rgba(encoding):
|
||||
# if encoding.startswith("data:image/"):
|
||||
# encoding = encoding.split(";")[1].split(",")[1]
|
||||
# try:
|
||||
# im_bytes = base64.b64decode(encoding)
|
||||
# im_arr = np.frombuffer(im_bytes, dtype=np.uint8) # im_arr is one-dim Numpy array
|
||||
# img = cv2.imdecode(im_arr, flags=cv2.IMREAD_UNCHANGED)
|
||||
# img = cv2.cvtColor(img, cv2.COLOR_BGRA2RGBA)
|
||||
# image = Image.fromarray(img, mode="RGBA")
|
||||
# return image
|
||||
# except Exception as e:
|
||||
# raise HTTPException(status_code=500, detail="Invalid encoded image") from e
|
||||
|
||||
def reactor_api(_: gr.Blocks, app: FastAPI):
|
||||
@app.post("/reactor/image")
|
||||
async def reactor_image(
|
||||
@ -60,7 +89,7 @@ def reactor_api(_: gr.Blocks, app: FastAPI):
|
||||
source_faces_index: list[int] = Body([0],title="Comma separated face number(s) from swap-source image"),
|
||||
face_index: list[int] = Body([0],title="Comma separated face number(s) for target image (result)"),
|
||||
upscaler: str = Body("None",title="Upscaler"),
|
||||
scale: int = Body(1,title="Scale by"),
|
||||
scale: float = Body(1,title="Scale by"),
|
||||
upscale_visibility: float = Body(1,title="Upscaler visibility (if scale = 1)"),
|
||||
face_restorer: str = Body("None",title="Restore Face: 0 - None; 1 - CodeFormer; 2 - GFPGA"),
|
||||
restorer_visibility: float = Body(1,title="Restore visibility value"),
|
||||
@ -71,29 +100,53 @@ def reactor_api(_: gr.Blocks, app: FastAPI):
|
||||
gender_target: int = Body(0,title="Gender Detection (Target) (0 - No, 1 - Female Only, 2 - Male Only)"),
|
||||
save_to_file: int = Body(0,title="Save Result to file, 0 - No, 1 - Yes"),
|
||||
result_file_path: str = Body("",title="(if 'save_to_file = 1') Result file path"),
|
||||
device: str = Body("CPU",title="CPU or CUDA (if you have it)")
|
||||
device: str = Body("CPU",title="CPU or CUDA (if you have it)"),
|
||||
mask_face: int = Body(0,title="Face Mask Correction, 1 - True, 0 - False"),
|
||||
select_source: int = Body(0,title="Select Source, 0 - Image, 1 - Face Model, 2 - Source Folder"),
|
||||
face_model: str = Body("None",title="Filename of the face model (from 'models/reactor/faces'), e.g. elena.safetensors"),
|
||||
source_folder: str = Body("",title="The path to the folder containing source faces images"),
|
||||
random_image: int = Body(0,title="Randomly select an image from the path"),
|
||||
upscale_force: int = Body(0,title="Force Upscale even if no face found"),
|
||||
det_thresh: float = Body(0.5,title="Face Detection Threshold"),
|
||||
det_maxnum: int = Body(0,title="Maximum number of faces to detect (0 is unlimited)"),
|
||||
):
|
||||
s_image = api.decode_base64_to_image(source_image)
|
||||
s_image = api.decode_base64_to_image(source_image) if select_source == 0 else None
|
||||
t_image = api.decode_base64_to_image(target_image)
|
||||
|
||||
if t_image.mode == 'RGBA':
|
||||
_, _, _, alpha = t_image.split()
|
||||
else:
|
||||
alpha = None
|
||||
|
||||
sf_index = source_faces_index
|
||||
f_index = face_index
|
||||
gender_s = gender_source
|
||||
gender_t = gender_target
|
||||
restore_first_bool = True if restore_first == 1 else False
|
||||
up_options = EnhancementOptions(do_restore_first=restore_first_bool, scale=scale, upscaler=get_upscaler(upscaler), upscale_visibility=upscale_visibility,face_restorer=get_face_restorer(face_restorer),restorer_visibility=restorer_visibility,codeformer_weight=codeformer_weight)
|
||||
mask_face = True if mask_face == 1 else False
|
||||
random_image = False if random_image == 0 else True
|
||||
upscale_force = False if upscale_force == 0 else True
|
||||
up_options = EnhancementOptions(do_restore_first=restore_first_bool, scale=scale, upscaler=get_upscaler(upscaler), upscale_visibility=upscale_visibility,face_restorer=get_face_restorer(face_restorer),restorer_visibility=restorer_visibility,codeformer_weight=codeformer_weight,upscale_force=upscale_force)
|
||||
det_options = DetectionOptions(det_thresh=det_thresh, det_maxnum=det_maxnum)
|
||||
use_model = get_full_model(model)
|
||||
if use_model is None:
|
||||
Exception("Model not found")
|
||||
result = swap_face(s_image, t_image, use_model, sf_index, f_index, up_options, gender_s, gender_t, True, True, device)
|
||||
result = swap_face(s_image, t_image, use_model, sf_index, f_index, up_options, gender_s, gender_t, True, True, device, mask_face, select_source, face_model, source_folder, None, random_image,det_options)
|
||||
result_img = result[0]
|
||||
|
||||
if alpha is not None:
|
||||
result_img = result_img.convert("RGBA")
|
||||
result_img.putalpha(alpha)
|
||||
|
||||
if save_to_file == 1:
|
||||
if result_file_path == "":
|
||||
result_file_path = default_file_path()
|
||||
try:
|
||||
result[0].save(result_file_path, format='PNG')
|
||||
result_img.save(result_file_path, format='PNG')
|
||||
logger.status("Result has been saved to: %s", result_file_path)
|
||||
except Exception as e:
|
||||
logger.error("Error while saving result: %s",e)
|
||||
return {"image": api.encode_pil_to_base64(result[0])}
|
||||
return {"image": api.encode_pil_to_base64(result_img)}
|
||||
|
||||
@app.get("/reactor/models")
|
||||
async def reactor_models():
|
||||
@ -104,6 +157,11 @@ def reactor_api(_: gr.Blocks, app: FastAPI):
|
||||
async def reactor_upscalers():
|
||||
names = [upscaler.name for upscaler in shared.sd_upscalers]
|
||||
return {"upscalers": names}
|
||||
|
||||
@app.get("/reactor/facemodels")
|
||||
async def reactor_facemodels():
|
||||
facemodels = [os.path.split(model)[1].split(".")[0] for model in get_facemodels()]
|
||||
return {"facemodels": facemodels}
|
||||
|
||||
try:
|
||||
import modules.script_callbacks as script_callbacks
|
||||
|
||||
@ -6,10 +6,10 @@ from modules import images
|
||||
from PIL import Image
|
||||
|
||||
|
||||
from scripts.entities.rect import Point, Rect
|
||||
from scripts.reactor_entities.rect import Point, Rect
|
||||
|
||||
|
||||
class Face:
|
||||
class FaceArea:
|
||||
def __init__(self, entire_image: np.ndarray, face_area: Rect, face_margin: float, face_size: int, upscaler: str):
|
||||
self.face_area = face_area
|
||||
self.center = face_area.center
|
||||
File diff suppressed because it is too large
Load Diff
@ -1,18 +1,42 @@
|
||||
import os
|
||||
from pathlib import Path
|
||||
|
||||
try:
|
||||
from modules.paths_internal import models_path
|
||||
except:
|
||||
try:
|
||||
from modules.paths import models_path
|
||||
except:
|
||||
models_path = os.path.abspath("models")
|
||||
|
||||
IS_RUN: bool = False
|
||||
BASE_PATH = os.path.join(Path(__file__).parents[1])
|
||||
DEVICE_LIST: list = ["CPU", "CUDA"]
|
||||
|
||||
MODELS_PATH = models_path
|
||||
SWAPPER_MODELS_PATH = os.path.join(MODELS_PATH, "insightface")
|
||||
REACTOR_MODELS_PATH = os.path.join(MODELS_PATH, "reactor")
|
||||
FACE_MODELS_PATH = os.path.join(REACTOR_MODELS_PATH, "faces")
|
||||
|
||||
IS_SDNEXT = False
|
||||
|
||||
if not os.path.exists(REACTOR_MODELS_PATH):
|
||||
os.makedirs(REACTOR_MODELS_PATH)
|
||||
if not os.path.exists(FACE_MODELS_PATH):
|
||||
os.makedirs(FACE_MODELS_PATH)
|
||||
|
||||
def updateDevice():
|
||||
try:
|
||||
LAST_DEVICE_PATH = os.path.join(BASE_PATH, "last_device.txt")
|
||||
with open(LAST_DEVICE_PATH) as f:
|
||||
for el in f:
|
||||
device = el.strip()
|
||||
except:
|
||||
device = "CPU"
|
||||
device = f.readline().strip()
|
||||
if device not in DEVICE_LIST:
|
||||
print(f"Error: Device {device} is not in DEVICE_LIST")
|
||||
device = DEVICE_LIST[0]
|
||||
print(f"Execution Provider has been set to {device}")
|
||||
except Exception as e:
|
||||
device = DEVICE_LIST[0]
|
||||
print(f"Error: {e}\nExecution Provider has been set to {device}")
|
||||
return device
|
||||
|
||||
DEVICE = updateDevice()
|
||||
|
||||
@ -1,14 +1,27 @@
|
||||
import os
|
||||
import os, glob, random
|
||||
from collections import Counter
|
||||
from PIL import Image
|
||||
from math import isqrt, ceil
|
||||
from typing import List
|
||||
import logging
|
||||
import hashlib
|
||||
import torch
|
||||
from safetensors.torch import save_file, safe_open
|
||||
from insightface.app.common import Face
|
||||
|
||||
from modules.images import FilenameGenerator, get_next_sequence_number
|
||||
from modules import shared, script_callbacks
|
||||
from scripts.reactor_globals import DEVICE, BASE_PATH
|
||||
from scripts.reactor_globals import DEVICE, BASE_PATH, FACE_MODELS_PATH, IS_SDNEXT
|
||||
|
||||
try:
|
||||
from modules.paths_internal import models_path
|
||||
except:
|
||||
try:
|
||||
from modules.paths import models_path
|
||||
except:
|
||||
model_path = os.path.abspath("models")
|
||||
|
||||
MODELS_PATH = None
|
||||
|
||||
def set_Device(value):
|
||||
global DEVICE
|
||||
@ -20,6 +33,14 @@ def get_Device():
|
||||
global DEVICE
|
||||
return DEVICE
|
||||
|
||||
def set_SDNEXT():
|
||||
global IS_SDNEXT
|
||||
IS_SDNEXT = True
|
||||
|
||||
def get_SDNEXT():
|
||||
global IS_SDNEXT
|
||||
return IS_SDNEXT
|
||||
|
||||
def make_grid(image_list: List):
|
||||
|
||||
# Count the occurrences of each image size in the image_list
|
||||
@ -133,3 +154,82 @@ def addLoggingLevel(levelName, levelNum, methodName=None):
|
||||
def get_image_md5hash(image: Image.Image):
|
||||
md5hash = hashlib.md5(image.tobytes())
|
||||
return md5hash.hexdigest()
|
||||
|
||||
def save_face_model(face: Face, filename: str) -> None:
|
||||
try:
|
||||
tensors = {
|
||||
"bbox": torch.tensor(face["bbox"]),
|
||||
"kps": torch.tensor(face["kps"]),
|
||||
"det_score": torch.tensor(face["det_score"]),
|
||||
"landmark_3d_68": torch.tensor(face["landmark_3d_68"]),
|
||||
"pose": torch.tensor(face["pose"]),
|
||||
"landmark_2d_106": torch.tensor(face["landmark_2d_106"]),
|
||||
"embedding": torch.tensor(face["embedding"]),
|
||||
"gender": torch.tensor(face["gender"]),
|
||||
"age": torch.tensor(face["age"]),
|
||||
}
|
||||
save_file(tensors, filename)
|
||||
# print(f"Face model has been saved to '{filename}'")
|
||||
except Exception as e:
|
||||
print(f"Error: {e}")
|
||||
|
||||
def get_models():
|
||||
global MODELS_PATH
|
||||
models_path_init = os.path.join(models_path, "insightface/*")
|
||||
models = glob.glob(models_path_init)
|
||||
models = [x for x in models if x.endswith(".onnx") or x.endswith(".pth")]
|
||||
models_names = []
|
||||
for model in models:
|
||||
model_path = os.path.split(model)
|
||||
if MODELS_PATH is None:
|
||||
MODELS_PATH = model_path[0]
|
||||
model_name = model_path[1]
|
||||
models_names.append(model_name)
|
||||
return models_names
|
||||
|
||||
def load_face_model(filename: str):
|
||||
face = {}
|
||||
model_path = os.path.join(FACE_MODELS_PATH, filename)
|
||||
with safe_open(model_path, framework="pt") as f:
|
||||
for k in f.keys():
|
||||
face[k] = f.get_tensor(k).numpy()
|
||||
return Face(face)
|
||||
|
||||
def get_facemodels():
|
||||
models_path = os.path.join(FACE_MODELS_PATH, "*")
|
||||
models = glob.glob(models_path)
|
||||
models = [x for x in models if x.endswith(".safetensors")]
|
||||
return models
|
||||
|
||||
def get_model_names(get_models):
|
||||
models = get_models()
|
||||
names = ["None"]
|
||||
for x in models:
|
||||
names.append(os.path.basename(x))
|
||||
return names
|
||||
|
||||
def get_images_from_folder(path: str):
|
||||
files_path = os.path.join(path, "*")
|
||||
files = glob.glob(files_path)
|
||||
images = []
|
||||
images_names = []
|
||||
for x in files:
|
||||
if x.endswith(('jpg', 'png', 'jpeg', 'webp', 'bmp')):
|
||||
images.append(Image.open(x))
|
||||
images_names.append(os.path.basename(x))
|
||||
return images,images_names
|
||||
# return [Image.open(x) for x in images if x.endswith(('jpg', 'png', 'jpeg', 'webp', 'bmp'))],[os.path.basename(x) for x in images if x.endswith(('jpg', 'png', 'jpeg', 'webp', 'bmp'))]
|
||||
|
||||
def get_random_image_from_folder(path: str):
|
||||
images,names = get_images_from_folder(path)
|
||||
random_image_index = random.randint(0, len(images) - 1)
|
||||
return [images[random_image_index]],[names[random_image_index]]
|
||||
|
||||
def get_images_from_list(imgs: List):
|
||||
images = []
|
||||
images_names = []
|
||||
for x in imgs:
|
||||
images.append(Image.open(os.path.abspath(x.name)))
|
||||
images_names.append(os.path.basename(x.name))
|
||||
return images,images_names
|
||||
# return [Image.open(os.path.abspath(x.name)) for x in imgs],[os.path.basename(x.name) for x in imgs]
|
||||
|
||||
@ -9,9 +9,8 @@ from facexlib.parsing import init_parsing_model
|
||||
from facexlib.utils.misc import img2tensor
|
||||
from torchvision.transforms.functional import normalize
|
||||
from PIL import Image
|
||||
from scripts.inferencers.vignette_mask_generator import VignetteMaskGenerator
|
||||
from scripts.inferencers.mask_generator import MaskGenerator
|
||||
|
||||
from scripts.reactor_inferencers.vignette_mask_generator import VignetteMaskGenerator
|
||||
from scripts.reactor_inferencers.mask_generator import MaskGenerator
|
||||
|
||||
class BiSeNetMaskGenerator(MaskGenerator):
|
||||
def __init__(self) -> None:
|
||||
@ -31,7 +30,7 @@ class BiSeNetMaskGenerator(MaskGenerator):
|
||||
fallback_ratio: float = 0.10,
|
||||
**kwargs,
|
||||
) -> np.ndarray:
|
||||
original_face_image = face_image
|
||||
# original_face_image = face_image
|
||||
face_image = face_image.copy()
|
||||
face_image = face_image[:, :, ::-1]
|
||||
|
||||
@ -67,6 +66,11 @@ class BiSeNetMaskGenerator(MaskGenerator):
|
||||
mask = self.fallback_mask_generator.generate_mask(
|
||||
original_face_image, face_area_on_image, use_minimal_area=True
|
||||
)
|
||||
# """if MaskGenerator.calculate_mask_coverage(mask) < fallback_ratio:
|
||||
# logger.info("Use fallback mask generator")
|
||||
# mask = self.fallback_mask_generator.generate_mask(
|
||||
# original_face_image, face_area_on_image, use_minimal_area=True
|
||||
# )"""
|
||||
|
||||
return mask
|
||||
|
||||
@ -4,7 +4,6 @@ from typing import Tuple
|
||||
import cv2
|
||||
import numpy as np
|
||||
|
||||
|
||||
class MaskGenerator(ABC):
|
||||
@abstractmethod
|
||||
def name(self) -> str:
|
||||
@ -3,7 +3,7 @@ from typing import Tuple
|
||||
import cv2
|
||||
import numpy as np
|
||||
|
||||
from scripts.inferencers.mask_generator import MaskGenerator
|
||||
from scripts.reactor_inferencers.mask_generator import MaskGenerator
|
||||
|
||||
|
||||
class VignetteMaskGenerator(MaskGenerator):
|
||||
@ -5,22 +5,35 @@ from typing import List, Tuple, Union
|
||||
|
||||
import cv2
|
||||
import numpy as np
|
||||
from numpy import uint8
|
||||
from PIL import Image, ImageDraw
|
||||
from scripts.inferencers.bisenet_mask_generator import BiSeNetMaskGenerator
|
||||
from scripts.entities.face import Face
|
||||
from scripts.entities.rect import Rect
|
||||
from PIL import Image
|
||||
from scipy import stats
|
||||
|
||||
import insightface
|
||||
from torchvision.transforms.functional import to_pil_image
|
||||
from scripts.reactor_helpers import get_image_md5hash, get_Device
|
||||
from insightface.app.common import Face
|
||||
|
||||
from scripts.reactor_globals import FACE_MODELS_PATH
|
||||
from scripts.reactor_helpers import (
|
||||
get_image_md5hash,
|
||||
get_Device,
|
||||
save_face_model,
|
||||
load_face_model,
|
||||
get_images_from_folder,
|
||||
get_random_image_from_folder,
|
||||
get_images_from_list,
|
||||
set_SDNEXT
|
||||
)
|
||||
from scripts.console_log_patch import apply_logging_patch
|
||||
|
||||
from modules.face_restoration import FaceRestoration
|
||||
try: # A1111
|
||||
from modules import codeformer_model
|
||||
from modules import codeformer_model, gfpgan_model
|
||||
except: # SD.Next
|
||||
from modules.postprocess import codeformer_model
|
||||
from modules.postprocess import codeformer_model, gfpgan_model
|
||||
set_SDNEXT()
|
||||
from modules.upscaler import UpscalerData
|
||||
from modules.shared import state
|
||||
from scripts.reactor_logger import logger
|
||||
from reactor_modules.reactor_mask import apply_face_mask
|
||||
|
||||
try:
|
||||
from modules.paths_internal import models_path
|
||||
@ -28,7 +41,7 @@ except:
|
||||
try:
|
||||
from modules.paths import models_path
|
||||
except:
|
||||
model_path = os.path.abspath("models")
|
||||
models_path = os.path.abspath("models")
|
||||
|
||||
import warnings
|
||||
|
||||
@ -57,7 +70,12 @@ class EnhancementOptions:
|
||||
face_restorer: FaceRestoration = None
|
||||
restorer_visibility: float = 0.5
|
||||
codeformer_weight: float = 0.5
|
||||
|
||||
upscale_force: bool = False
|
||||
|
||||
@dataclass
|
||||
class DetectionOptions:
|
||||
det_thresh: float = 0.5
|
||||
det_maxnum: int = 0
|
||||
|
||||
@dataclass
|
||||
class MaskOptions:
|
||||
@ -94,16 +112,46 @@ def check_process_halt(msgforced: bool = False):
|
||||
|
||||
|
||||
FS_MODEL = None
|
||||
ANALYSIS_MODEL = None
|
||||
MASK_MODEL = None
|
||||
|
||||
CURRENT_FS_MODEL_PATH = None
|
||||
CURRENT_MASK_MODEL_PATH = None
|
||||
ANALYSIS_MODEL = None
|
||||
|
||||
SOURCE_FACES = None
|
||||
SOURCE_IMAGE_HASH = None
|
||||
TARGET_FACES = None
|
||||
TARGET_IMAGE_HASH = None
|
||||
SOURCE_FACES_LIST = []
|
||||
SOURCE_IMAGE_LIST_HASH = []
|
||||
|
||||
def clear_faces():
|
||||
global SOURCE_FACES, SOURCE_IMAGE_HASH
|
||||
SOURCE_FACES = None
|
||||
SOURCE_IMAGE_HASH = None
|
||||
logger.status("Source Images Hash has been reset (for Single Source or Face Model)")
|
||||
|
||||
def clear_faces_list():
|
||||
global SOURCE_FACES_LIST, SOURCE_IMAGE_LIST_HASH
|
||||
SOURCE_FACES_LIST = []
|
||||
SOURCE_IMAGE_LIST_HASH = []
|
||||
logger.status("Source Images Hash has been reset (for Multiple or Folder Source)")
|
||||
|
||||
def clear_faces_target():
|
||||
global TARGET_FACES, TARGET_IMAGE_HASH
|
||||
TARGET_FACES = None
|
||||
TARGET_IMAGE_HASH = None
|
||||
logger.status("Target Images Hash has been reset")
|
||||
|
||||
def clear_faces_all():
|
||||
global SOURCE_FACES, SOURCE_IMAGE_HASH, SOURCE_FACES_LIST, SOURCE_IMAGE_LIST_HASH, TARGET_FACES, TARGET_IMAGE_HASH
|
||||
SOURCE_FACES = None
|
||||
SOURCE_IMAGE_HASH = None
|
||||
TARGET_FACES = None
|
||||
TARGET_IMAGE_HASH = None
|
||||
SOURCE_FACES_LIST = []
|
||||
SOURCE_IMAGE_LIST_HASH = []
|
||||
logger.status("All Images Hash has been reset")
|
||||
|
||||
def getAnalysisModel():
|
||||
global ANALYSIS_MODEL
|
||||
@ -124,8 +172,6 @@ def getFaceSwapModel(model_path: str):
|
||||
return FS_MODEL
|
||||
|
||||
|
||||
|
||||
|
||||
def restore_face(image: Image, enhancement_options: EnhancementOptions):
|
||||
result_image = image
|
||||
|
||||
@ -134,14 +180,16 @@ def restore_face(image: Image, enhancement_options: EnhancementOptions):
|
||||
|
||||
if enhancement_options.face_restorer is not None:
|
||||
original_image = result_image.copy()
|
||||
logger.status("Restoring the face with %s", enhancement_options.face_restorer.name())
|
||||
numpy_image = np.array(result_image)
|
||||
if enhancement_options.face_restorer.name() == "CodeFormer":
|
||||
logger.status("Restoring the face with %s (weight: %s)", enhancement_options.face_restorer.name(), enhancement_options.codeformer_weight)
|
||||
numpy_image = codeformer_model.codeformer.restore(
|
||||
numpy_image, w=enhancement_options.codeformer_weight
|
||||
)
|
||||
else:
|
||||
numpy_image = enhancement_options.face_restorer.restore(numpy_image)
|
||||
else: # GFPGAN:
|
||||
logger.status("Restoring the face with %s", enhancement_options.face_restorer.name())
|
||||
numpy_image = gfpgan_model.gfpgan_fix_faces(numpy_image)
|
||||
# numpy_image = enhancement_options.face_restorer.restore(numpy_image)
|
||||
restored_image = Image.fromarray(numpy_image)
|
||||
result_image = Image.blend(
|
||||
original_image, restored_image, enhancement_options.restorer_visibility
|
||||
@ -269,13 +317,13 @@ def half_det_size(det_size):
|
||||
logger.status("Trying to halve 'det_size' parameter")
|
||||
return (det_size[0] // 2, det_size[1] // 2)
|
||||
|
||||
def analyze_faces(img_data: np.ndarray, det_size=(640, 640)):
|
||||
def analyze_faces(img_data: np.ndarray, det_size=(640, 640), det_thresh=0.5, det_maxnum=0):
|
||||
logger.info("Applied Execution Provider: %s", PROVIDERS[0])
|
||||
face_analyser = copy.deepcopy(getAnalysisModel())
|
||||
face_analyser.prepare(ctx_id=0, det_size=det_size)
|
||||
return face_analyser.get(img_data)
|
||||
face_analyser.prepare(ctx_id=0, det_thresh=det_thresh, det_size=det_size)
|
||||
return face_analyser.get(img_data, max_num=det_maxnum)
|
||||
|
||||
def get_face_single(img_data: np.ndarray, face, face_index=0, det_size=(640, 640), gender_source=0, gender_target=0):
|
||||
def get_face_single(img_data: np.ndarray, face, face_index=0, det_size=(640, 640), gender_source=0, gender_target=0, det_thresh=0.5, det_maxnum=0):
|
||||
|
||||
buffalo_path = os.path.join(models_path, "insightface/models/buffalo_l.zip")
|
||||
if os.path.exists(buffalo_path):
|
||||
@ -298,20 +346,20 @@ def get_face_single(img_data: np.ndarray, face, face_index=0, det_size=(640, 640
|
||||
if gender_source != 0:
|
||||
if len(face) == 0 and det_size[0] > 320 and det_size[1] > 320:
|
||||
det_size_half = half_det_size(det_size)
|
||||
return get_face_single(img_data, analyze_faces(img_data, det_size_half), face_index, det_size_half, gender_source, gender_target)
|
||||
return get_face_single(img_data, analyze_faces(img_data, det_size_half, det_thresh, det_maxnum), face_index, det_size_half, gender_source, gender_target, det_thresh, det_maxnum)
|
||||
faces, wrong_gender = get_face_gender(face,face_index,gender_source,"Source",gender_detected)
|
||||
return faces, wrong_gender, face_age, face_gender
|
||||
|
||||
if gender_target != 0:
|
||||
if len(face) == 0 and det_size[0] > 320 and det_size[1] > 320:
|
||||
det_size_half = half_det_size(det_size)
|
||||
return get_face_single(img_data, analyze_faces(img_data, det_size_half), face_index, det_size_half, gender_source, gender_target)
|
||||
return get_face_single(img_data, analyze_faces(img_data, det_size_half, det_thresh, det_maxnum), face_index, det_size_half, gender_source, gender_target, det_thresh, det_maxnum)
|
||||
faces, wrong_gender = get_face_gender(face,face_index,gender_target,"Target",gender_detected)
|
||||
return faces, wrong_gender, face_age, face_gender
|
||||
|
||||
if len(face) == 0 and det_size[0] > 320 and det_size[1] > 320:
|
||||
det_size_half = half_det_size(det_size)
|
||||
return get_face_single(img_data, analyze_faces(img_data, det_size_half), face_index, det_size_half, gender_source, gender_target)
|
||||
return get_face_single(img_data, analyze_faces(img_data, det_size_half, det_thresh, det_maxnum), face_index, det_size_half, gender_source, gender_target, det_thresh, det_maxnum)
|
||||
|
||||
try:
|
||||
return sorted(face, key=lambda x: x.bbox[0])[face_index], 0, face_age, face_gender
|
||||
@ -331,10 +379,17 @@ def swap_face(
|
||||
source_hash_check: bool = True,
|
||||
target_hash_check: bool = False,
|
||||
device: str = "CPU",
|
||||
mask_face:bool = False,
|
||||
mask_options:Union[MaskOptions, None]= None
|
||||
mask_face: bool = False,
|
||||
select_source: int = 0,
|
||||
face_model: str = "None",
|
||||
source_folder: str = "",
|
||||
source_imgs: Union[List, None] = None,
|
||||
random_image: bool = False,
|
||||
detection_options: Union[DetectionOptions, None] = None,
|
||||
mask_options:Union[MaskOptions, None]= None,
|
||||
):
|
||||
global SOURCE_FACES, SOURCE_IMAGE_HASH, TARGET_FACES, TARGET_IMAGE_HASH, PROVIDERS
|
||||
global SOURCE_FACES, SOURCE_IMAGE_HASH, TARGET_FACES, TARGET_IMAGE_HASH, PROVIDERS, SOURCE_FACES_LIST, SOURCE_IMAGE_LIST_HASH
|
||||
|
||||
result_image = target_img
|
||||
masked_faces = None
|
||||
PROVIDERS = ["CUDAExecutionProvider"] if device == "CUDA" else ["CPUExecutionProvider"]
|
||||
@ -359,186 +414,457 @@ def swap_face(
|
||||
|
||||
source_img = Image.open(io.BytesIO(img_bytes))
|
||||
|
||||
source_img = cv2.cvtColor(np.array(source_img), cv2.COLOR_RGB2BGR)
|
||||
target_img = cv2.cvtColor(np.array(target_img), cv2.COLOR_RGB2BGR)
|
||||
|
||||
target_img_orig = cv2.cvtColor(np.array(target_img), cv2.COLOR_RGB2BGR)
|
||||
entire_mask_image = np.zeros_like(np.array(target_img))
|
||||
|
||||
output: List = []
|
||||
output_info: str = ""
|
||||
swapped = 0
|
||||
|
||||
# *****************
|
||||
# SWAP from FOLDER or MULTIPLE images:
|
||||
|
||||
if source_hash_check:
|
||||
if (select_source == 0 and source_imgs is not None) or (select_source == 2 and (source_folder is not None and source_folder != "")):
|
||||
|
||||
source_image_md5hash = get_image_md5hash(source_img)
|
||||
result = []
|
||||
|
||||
if SOURCE_IMAGE_HASH is None:
|
||||
SOURCE_IMAGE_HASH = source_image_md5hash
|
||||
source_image_same = False
|
||||
if random_image and select_source == 2:
|
||||
source_images,source_images_names = get_random_image_from_folder(source_folder)
|
||||
logger.status(f"Processing with Random Image from the folder: {source_images_names[0]}")
|
||||
else:
|
||||
source_image_same = True if SOURCE_IMAGE_HASH == source_image_md5hash else False
|
||||
if not source_image_same:
|
||||
SOURCE_IMAGE_HASH = source_image_md5hash
|
||||
source_images,source_images_names = get_images_from_folder(source_folder) if select_source == 2 else get_images_from_list(source_imgs)
|
||||
|
||||
logger.info("Source Image MD5 Hash = %s", SOURCE_IMAGE_HASH)
|
||||
logger.info("Source Image the Same? %s", source_image_same)
|
||||
if len(source_images) > 0:
|
||||
source_img_ff = []
|
||||
source_faces_ff = []
|
||||
for i, source_image in enumerate(source_images):
|
||||
|
||||
if SOURCE_FACES is None or not source_image_same:
|
||||
logger.status("Analyzing Source Image...")
|
||||
source_faces = analyze_faces(source_img)
|
||||
SOURCE_FACES = source_faces
|
||||
elif source_image_same:
|
||||
logger.status("Using Ready Source Face(s) Model...")
|
||||
source_faces = SOURCE_FACES
|
||||
source_image = cv2.cvtColor(np.array(source_image), cv2.COLOR_RGB2BGR)
|
||||
source_img_ff.append(source_image)
|
||||
|
||||
if source_hash_check:
|
||||
|
||||
source_image_md5hash = get_image_md5hash(source_image)
|
||||
|
||||
if len(SOURCE_IMAGE_LIST_HASH) == 0:
|
||||
SOURCE_IMAGE_LIST_HASH = [source_image_md5hash]
|
||||
source_image_same = False
|
||||
elif len(SOURCE_IMAGE_LIST_HASH) == i:
|
||||
SOURCE_IMAGE_LIST_HASH.append(source_image_md5hash)
|
||||
source_image_same = False
|
||||
else:
|
||||
source_image_same = True if SOURCE_IMAGE_LIST_HASH[i] == source_image_md5hash else False
|
||||
if not source_image_same:
|
||||
SOURCE_IMAGE_LIST_HASH[i] = source_image_md5hash
|
||||
|
||||
logger.info("(Image %s) Source Image MD5 Hash = %s", i, SOURCE_IMAGE_LIST_HASH[i])
|
||||
logger.info("(Image %s) Source Image the Same? %s", i, source_image_same)
|
||||
|
||||
if len(SOURCE_FACES_LIST) == 0:
|
||||
logger.status(f"Analyzing Source Image {i}: {source_images_names[i]}...")
|
||||
source_faces = analyze_faces(source_image, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
SOURCE_FACES_LIST = [source_faces]
|
||||
elif len(SOURCE_FACES_LIST) == i and not source_image_same:
|
||||
logger.status(f"Analyzing Source Image {i}: {source_images_names[i]}...")
|
||||
source_faces = analyze_faces(source_image, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
SOURCE_FACES_LIST.append(source_faces)
|
||||
elif len(SOURCE_FACES_LIST) != i and not source_image_same:
|
||||
logger.status(f"Analyzing Source Image {i}: {source_images_names[i]}...")
|
||||
source_faces = analyze_faces(source_image, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
SOURCE_FACES_LIST[i] = source_faces
|
||||
elif source_image_same:
|
||||
logger.status("(Image %s) Using Hashed Source Face(s) Model...", i)
|
||||
source_faces = SOURCE_FACES_LIST[i]
|
||||
|
||||
else:
|
||||
logger.status(f"Analyzing Source Image {i}...")
|
||||
source_faces = analyze_faces(source_image, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
|
||||
if source_faces is not None:
|
||||
source_faces_ff.append(source_faces)
|
||||
|
||||
if len(source_faces_ff) > 0:
|
||||
|
||||
if target_hash_check:
|
||||
|
||||
target_image_md5hash = get_image_md5hash(target_img)
|
||||
|
||||
if TARGET_IMAGE_HASH is None:
|
||||
TARGET_IMAGE_HASH = target_image_md5hash
|
||||
target_image_same = False
|
||||
else:
|
||||
target_image_same = True if TARGET_IMAGE_HASH == target_image_md5hash else False
|
||||
if not target_image_same:
|
||||
TARGET_IMAGE_HASH = target_image_md5hash
|
||||
|
||||
logger.info("Target Image MD5 Hash = %s", TARGET_IMAGE_HASH)
|
||||
logger.info("Target Image the Same? %s", target_image_same)
|
||||
|
||||
if TARGET_FACES is None or not target_image_same:
|
||||
logger.status("Analyzing Target Image...")
|
||||
target_faces = analyze_faces(target_img, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
TARGET_FACES = target_faces
|
||||
elif target_image_same:
|
||||
logger.status("Using Hashed Target Face(s) Model...")
|
||||
target_faces = TARGET_FACES
|
||||
|
||||
else:
|
||||
logger.status("Analyzing Target Image...")
|
||||
target_faces = analyze_faces(target_img, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
|
||||
for i,source_faces in enumerate(source_faces_ff):
|
||||
|
||||
logger.status("(Image %s) Detecting Source Face, Index = %s", i, source_faces_index[0])
|
||||
source_face, wrong_gender, source_age, source_gender = get_face_single(source_img_ff[i], source_faces, face_index=source_faces_index[0], gender_source=gender_source, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
|
||||
if source_age != "None" or source_gender != "None":
|
||||
logger.status("(Image %s) Detected: -%s- y.o. %s", i, source_age, source_gender)
|
||||
|
||||
if len(source_faces_index) != 0 and len(source_faces_index) != 1 and len(source_faces_index) != len(faces_index):
|
||||
logger.status("Source Faces must have no entries (default=0), one entry, or same number of entries as target faces.")
|
||||
|
||||
elif source_face is not None:
|
||||
|
||||
result_image, output, swapped = operate(source_img_ff[i],target_img,target_img_orig,model,source_faces_index,faces_index,source_faces,target_faces,gender_source,gender_target,source_face,wrong_gender,source_age,source_gender,output,swapped,mask_face,entire_mask_image,enhancement_options,detection_options)
|
||||
|
||||
result.append(result_image)
|
||||
|
||||
result = [result_image] if len(result) == 0 else result
|
||||
|
||||
return result, output, swapped
|
||||
|
||||
# END
|
||||
# *****************
|
||||
|
||||
# ***********************
|
||||
# SWAP from IMG or MODEL:
|
||||
|
||||
else:
|
||||
logger.status("Analyzing Source Image...")
|
||||
source_faces = analyze_faces(source_img)
|
||||
|
||||
if source_faces is not None:
|
||||
|
||||
if target_hash_check:
|
||||
|
||||
target_image_md5hash = get_image_md5hash(target_img)
|
||||
|
||||
if TARGET_IMAGE_HASH is None:
|
||||
TARGET_IMAGE_HASH = target_image_md5hash
|
||||
target_image_same = False
|
||||
else:
|
||||
target_image_same = True if TARGET_IMAGE_HASH == target_image_md5hash else False
|
||||
if not target_image_same:
|
||||
TARGET_IMAGE_HASH = target_image_md5hash
|
||||
|
||||
logger.info("Target Image MD5 Hash = %s", TARGET_IMAGE_HASH)
|
||||
logger.info("Target Image the Same? %s", target_image_same)
|
||||
|
||||
if select_source == 0 and source_img is not None:
|
||||
|
||||
if TARGET_FACES is None or not target_image_same:
|
||||
logger.status("Analyzing Target Image...")
|
||||
target_faces = analyze_faces(target_img)
|
||||
TARGET_FACES = target_faces
|
||||
elif target_image_same:
|
||||
logger.status("Using Ready Target Face(s) Model...")
|
||||
target_faces = TARGET_FACES
|
||||
source_img = cv2.cvtColor(np.array(source_img), cv2.COLOR_RGB2BGR)
|
||||
|
||||
if source_hash_check:
|
||||
|
||||
source_image_md5hash = get_image_md5hash(source_img)
|
||||
|
||||
if SOURCE_IMAGE_HASH is None:
|
||||
SOURCE_IMAGE_HASH = source_image_md5hash
|
||||
source_image_same = False
|
||||
else:
|
||||
source_image_same = True if SOURCE_IMAGE_HASH == source_image_md5hash else False
|
||||
if not source_image_same:
|
||||
SOURCE_IMAGE_HASH = source_image_md5hash
|
||||
|
||||
logger.info("Source Image MD5 Hash = %s", SOURCE_IMAGE_HASH)
|
||||
logger.info("Source Image the Same? %s", source_image_same)
|
||||
|
||||
if SOURCE_FACES is None or not source_image_same:
|
||||
logger.status("Analyzing Source Image...")
|
||||
source_faces = analyze_faces(source_img, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
SOURCE_FACES = source_faces
|
||||
elif source_image_same:
|
||||
logger.status("Using Hashed Source Face(s) Model...")
|
||||
source_faces = SOURCE_FACES
|
||||
|
||||
else:
|
||||
logger.status("Analyzing Source Image...")
|
||||
source_faces = analyze_faces(source_img, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
|
||||
elif select_source == 1 and (face_model is not None and face_model != "None"):
|
||||
source_face_model = [load_face_model(face_model)]
|
||||
if source_face_model is not None:
|
||||
source_faces_index = [0]
|
||||
source_faces = source_face_model
|
||||
logger.status(f"Using Loaded Source Face Model: {face_model}")
|
||||
else:
|
||||
logger.error(f"Cannot load Face Model File: {face_model}")
|
||||
|
||||
else:
|
||||
logger.status("Analyzing Target Image...")
|
||||
target_faces = analyze_faces(target_img)
|
||||
logger.error("Cannot detect any Source")
|
||||
return result_image, [], 0
|
||||
|
||||
logger.status("Detecting Source Face, Index = %s", source_faces_index[0])
|
||||
source_face, wrong_gender, source_age, source_gender = get_face_single(source_img, source_faces, face_index=source_faces_index[0], gender_source=gender_source)
|
||||
if source_faces is not None:
|
||||
|
||||
if target_hash_check:
|
||||
|
||||
target_image_md5hash = get_image_md5hash(target_img)
|
||||
|
||||
if TARGET_IMAGE_HASH is None:
|
||||
TARGET_IMAGE_HASH = target_image_md5hash
|
||||
target_image_same = False
|
||||
else:
|
||||
target_image_same = True if TARGET_IMAGE_HASH == target_image_md5hash else False
|
||||
if not target_image_same:
|
||||
TARGET_IMAGE_HASH = target_image_md5hash
|
||||
|
||||
logger.info("Target Image MD5 Hash = %s", TARGET_IMAGE_HASH)
|
||||
logger.info("Target Image the Same? %s", target_image_same)
|
||||
|
||||
if TARGET_FACES is None or not target_image_same:
|
||||
logger.status("Analyzing Target Image...")
|
||||
target_faces = analyze_faces(target_img, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
TARGET_FACES = target_faces
|
||||
elif target_image_same:
|
||||
logger.status("Using Hashed Target Face(s) Model...")
|
||||
target_faces = TARGET_FACES
|
||||
|
||||
else:
|
||||
logger.status("Analyzing Target Image...")
|
||||
target_faces = analyze_faces(target_img, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
|
||||
logger.status("Detecting Source Face, Index = %s", source_faces_index[0])
|
||||
if select_source == 0 and source_img is not None:
|
||||
source_face, wrong_gender, source_age, source_gender = get_face_single(source_img, source_faces, face_index=source_faces_index[0], gender_source=gender_source, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
else:
|
||||
source_face = sorted(source_faces, key=lambda x: x.bbox[0])[source_faces_index[0]]
|
||||
wrong_gender = 0
|
||||
source_age = source_face["age"]
|
||||
source_gender = "Female" if source_face["gender"] == 0 else "Male"
|
||||
|
||||
if source_age != "None" or source_gender != "None":
|
||||
logger.status("Detected: -%s- y.o. %s", source_age, source_gender)
|
||||
|
||||
output_info = f"SourceFaceIndex={source_faces_index[0]};Age={source_age};Gender={source_gender}\n"
|
||||
output.append(output_info)
|
||||
|
||||
if len(source_faces_index) != 0 and len(source_faces_index) != 1 and len(source_faces_index) != len(faces_index):
|
||||
logger.status("Source Faces must have no entries (default=0), one entry, or same number of entries as target faces.")
|
||||
|
||||
elif source_face is not None:
|
||||
|
||||
result_image, output, swapped = operate(source_img,target_img,target_img_orig,model,source_faces_index,faces_index,source_faces,target_faces,gender_source,gender_target,source_face,wrong_gender,source_age,source_gender,output,swapped,mask_face,entire_mask_image,enhancement_options,detection_options)
|
||||
|
||||
else:
|
||||
logger.status("No source face(s) in the provided Index")
|
||||
else:
|
||||
logger.status("No source face(s) found")
|
||||
|
||||
return result_image, output, swapped
|
||||
|
||||
# END
|
||||
# **********************
|
||||
|
||||
return result_image, [], 0
|
||||
|
||||
def build_face_model(image: Image.Image, name: str, save_model: bool = True, det_size=(640, 640)):
|
||||
if image is None:
|
||||
error_msg = "Please load an Image"
|
||||
logger.error(error_msg)
|
||||
return error_msg
|
||||
if name is None:
|
||||
error_msg = "Please filled out the 'Face Model Name' field"
|
||||
logger.error(error_msg)
|
||||
return error_msg
|
||||
apply_logging_patch(1)
|
||||
image = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
|
||||
if save_model:
|
||||
logger.status("Building Face Model...")
|
||||
face_model = analyze_faces(image, det_size)
|
||||
|
||||
if len(face_model) == 0:
|
||||
det_size_half = half_det_size(det_size)
|
||||
face_model = analyze_faces(image, det_size_half)
|
||||
|
||||
if face_model is not None and len(face_model) > 0:
|
||||
if save_model:
|
||||
face_model_path = os.path.join(FACE_MODELS_PATH, name + ".safetensors")
|
||||
save_face_model(face_model[0],face_model_path)
|
||||
logger.status("--Done!--")
|
||||
done_msg = f"Face model has been saved to '{face_model_path}'"
|
||||
logger.status(done_msg)
|
||||
return done_msg
|
||||
else:
|
||||
return face_model[0]
|
||||
else:
|
||||
no_face_msg = "No face found, please try another image"
|
||||
logger.error(no_face_msg)
|
||||
return no_face_msg
|
||||
|
||||
def blend_faces(images_list: List, name: str, compute_method: int = 0, shape_check: bool = False):
|
||||
faces = []
|
||||
embeddings = []
|
||||
images: List[Image.Image] = []
|
||||
images, images_names = get_images_from_list(images_list)
|
||||
for i,image in enumerate(images):
|
||||
logger.status(f"Building Face Model for {images_names[i]}...")
|
||||
face = build_face_model(image,str(i),save_model=False)
|
||||
if isinstance(face, str):
|
||||
# logger.error(f"No faces found in {images_names[i]}, skipping")
|
||||
continue
|
||||
if shape_check:
|
||||
if i == 0:
|
||||
embedding_shape = face.embedding.shape
|
||||
elif face.embedding.shape != embedding_shape:
|
||||
logger.error(f"Embedding Shape Mismatch for {images_names[i]}, skipping")
|
||||
continue
|
||||
faces.append(face)
|
||||
embeddings.append(face.embedding)
|
||||
if len(faces) > 0:
|
||||
# if shape_check:
|
||||
# embedding_shape = embeddings[0].shape
|
||||
# for embedding in embeddings:
|
||||
# if embedding.shape != embedding_shape:
|
||||
# logger.error("Embedding Shape Mismatch")
|
||||
# break
|
||||
compute_method_name = "Mean" if compute_method == 0 else "Median" if compute_method == 1 else "Mode"
|
||||
logger.status(f"Blending with Compute Method {compute_method_name}...")
|
||||
blended_embedding = np.mean(embeddings, axis=0) if compute_method == 0 else np.median(embeddings, axis=0) if compute_method == 1 else stats.mode(embeddings, axis=0)[0].astype(np.float32)
|
||||
blended_face = Face(
|
||||
bbox=faces[0].bbox,
|
||||
kps=faces[0].kps,
|
||||
det_score=faces[0].det_score,
|
||||
landmark_3d_68=faces[0].landmark_3d_68,
|
||||
pose=faces[0].pose,
|
||||
landmark_2d_106=faces[0].landmark_2d_106,
|
||||
embedding=blended_embedding,
|
||||
gender=faces[0].gender,
|
||||
age=faces[0].age
|
||||
)
|
||||
if blended_face is not None:
|
||||
face_model_path = os.path.join(FACE_MODELS_PATH, name + ".safetensors")
|
||||
save_face_model(blended_face,face_model_path)
|
||||
logger.status("--Done!--")
|
||||
done_msg = f"Face model has been saved to '{face_model_path}'"
|
||||
logger.status(done_msg)
|
||||
return done_msg
|
||||
else:
|
||||
no_face_msg = "Something went wrong, please try another set of images"
|
||||
logger.error(no_face_msg)
|
||||
return no_face_msg
|
||||
return "No faces found"
|
||||
|
||||
|
||||
def operate(
|
||||
source_img,
|
||||
target_img,
|
||||
target_img_orig,
|
||||
model,
|
||||
source_faces_index,
|
||||
faces_index,
|
||||
source_faces,
|
||||
target_faces,
|
||||
gender_source,
|
||||
gender_target,
|
||||
source_face,
|
||||
wrong_gender,
|
||||
source_age,
|
||||
source_gender,
|
||||
output,
|
||||
swapped,
|
||||
mask_face,
|
||||
entire_mask_image,
|
||||
enhancement_options,
|
||||
detection_options,
|
||||
):
|
||||
result = target_img
|
||||
face_swapper = getFaceSwapModel(model)
|
||||
|
||||
source_face_idx = 0
|
||||
|
||||
for face_num in faces_index:
|
||||
if check_process_halt():
|
||||
return result_image, [], 0
|
||||
if len(source_faces_index) > 1 and source_face_idx > 0:
|
||||
logger.status("Detecting Source Face, Index = %s", source_faces_index[source_face_idx])
|
||||
source_face, wrong_gender, source_age, source_gender = get_face_single(source_img, source_faces, face_index=source_faces_index[source_face_idx], gender_source=gender_source, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
if source_age != "None" or source_gender != "None":
|
||||
logger.status("Detected: -%s- y.o. %s", source_age, source_gender)
|
||||
|
||||
output_info = f"SourceFaceIndex={source_faces_index[0]};Age={source_age};Gender={source_gender}\n"
|
||||
output_info = f"SourceFaceIndex={source_faces_index[source_face_idx]};Age={source_age};Gender={source_gender}\n"
|
||||
output.append(output_info)
|
||||
|
||||
if len(source_faces_index) != 0 and len(source_faces_index) != 1 and len(source_faces_index) != len(faces_index):
|
||||
logger.status("Source Faces must have no entries (default=0), one entry, or same number of entries as target faces.")
|
||||
elif source_face is not None:
|
||||
|
||||
result = target_img
|
||||
face_swapper = getFaceSwapModel(model)
|
||||
source_face_idx += 1
|
||||
|
||||
source_face_idx = 0
|
||||
if source_face is not None and wrong_gender == 0:
|
||||
logger.status("Detecting Target Face, Index = %s", face_num)
|
||||
target_face, wrong_gender, target_age, target_gender = get_face_single(target_img, target_faces, face_index=face_num, gender_target=gender_target, det_thresh=detection_options.det_thresh, det_maxnum=detection_options.det_maxnum)
|
||||
if target_age != "None" or target_gender != "None":
|
||||
logger.status("Detected: -%s- y.o. %s", target_age, target_gender)
|
||||
|
||||
for face_num in faces_index:
|
||||
if check_process_halt():
|
||||
return result_image, [], 0
|
||||
if len(source_faces_index) > 1 and source_face_idx > 0:
|
||||
logger.status("Detecting Source Face, Index = %s", source_faces_index[source_face_idx])
|
||||
source_face, wrong_gender, source_age, source_gender = get_face_single(source_img, source_faces, face_index=source_faces_index[source_face_idx], gender_source=gender_source)
|
||||
if source_age != "None" or source_gender != "None":
|
||||
logger.status("Detected: -%s- y.o. %s", source_age, source_gender)
|
||||
|
||||
output_info = f"SourceFaceIndex={source_faces_index[source_face_idx]};Age={source_age};Gender={source_gender}\n"
|
||||
output.append(output_info)
|
||||
|
||||
source_face_idx += 1
|
||||
|
||||
if source_face is not None and wrong_gender == 0:
|
||||
logger.status("Detecting Target Face, Index = %s", face_num)
|
||||
target_face, wrong_gender, target_age, target_gender = get_face_single(target_img, target_faces, face_index=face_num, gender_target=gender_target)
|
||||
if target_age != "None" or target_gender != "None":
|
||||
logger.status("Detected: -%s- y.o. %s", target_age, target_gender)
|
||||
|
||||
output_info = f"TargetFaceIndex={face_num};Age={target_age};Gender={target_gender}\n"
|
||||
output.append(output_info)
|
||||
|
||||
if target_face is not None and wrong_gender == 0:
|
||||
logger.status("Swapping Source into Target")
|
||||
swapped_image = face_swapper.get(result, target_face, source_face)
|
||||
|
||||
if mask_face:
|
||||
result = apply_face_mask(swapped_image=swapped_image,target_image=result,target_face=target_face,entire_mask_image=entire_mask_image,mask_options=mask_options)
|
||||
else:
|
||||
result = swapped_image
|
||||
swapped += 1
|
||||
|
||||
elif wrong_gender == 1:
|
||||
wrong_gender = 0
|
||||
|
||||
if source_face_idx == len(source_faces_index):
|
||||
result_image = Image.fromarray(cv2.cvtColor(result, cv2.COLOR_BGR2RGB))
|
||||
|
||||
if enhancement_options is not None and len(source_faces_index) > 1:
|
||||
result_image = enhance_image(result_image, enhancement_options)
|
||||
|
||||
return result_image, output, swapped
|
||||
|
||||
else:
|
||||
logger.status(f"No target face found for {face_num}")
|
||||
|
||||
elif wrong_gender == 1:
|
||||
wrong_gender = 0
|
||||
|
||||
if source_face_idx == len(source_faces_index):
|
||||
result_image = Image.fromarray(cv2.cvtColor(result, cv2.COLOR_BGR2RGB))
|
||||
|
||||
if enhancement_options is not None and len(source_faces_index) > 1:
|
||||
result_image = enhance_image(result_image, enhancement_options)
|
||||
|
||||
return result_image, output, swapped
|
||||
|
||||
output_info = f"TargetFaceIndex={face_num};Age={target_age};Gender={target_gender}\n"
|
||||
output.append(output_info)
|
||||
|
||||
if target_face is not None and wrong_gender == 0:
|
||||
logger.status("Swapping Source into Target")
|
||||
swapped_image = face_swapper.get(result, target_face, source_face)
|
||||
|
||||
if mask_face:
|
||||
result = apply_face_mask(swapped_image=swapped_image,target_image=result,target_face=target_face,entire_mask_image=entire_mask_image)
|
||||
else:
|
||||
logger.status(f"No source face found for face number {source_face_idx}.")
|
||||
result = swapped_image
|
||||
swapped += 1
|
||||
|
||||
elif wrong_gender == 1:
|
||||
wrong_gender = 0
|
||||
|
||||
if source_face_idx == len(source_faces_index):
|
||||
result_image = Image.fromarray(cv2.cvtColor(result, cv2.COLOR_BGR2RGB))
|
||||
|
||||
if enhancement_options is not None and len(source_faces_index) > 1:
|
||||
result_image = enhance_image(result_image, enhancement_options)
|
||||
|
||||
return result_image, output, swapped
|
||||
|
||||
else:
|
||||
logger.status(f"No target face found for {face_num}")
|
||||
|
||||
elif wrong_gender == 1:
|
||||
wrong_gender = 0
|
||||
|
||||
if source_face_idx == len(source_faces_index):
|
||||
result_image = Image.fromarray(cv2.cvtColor(result, cv2.COLOR_BGR2RGB))
|
||||
|
||||
if enhancement_options is not None and len(source_faces_index) > 1:
|
||||
result_image = enhance_image(result_image, enhancement_options)
|
||||
|
||||
return result_image, output, swapped
|
||||
|
||||
else:
|
||||
logger.status(f"No source face found for face number {source_face_idx}.")
|
||||
|
||||
result_image = Image.fromarray(cv2.cvtColor(result, cv2.COLOR_BGR2RGB))
|
||||
|
||||
if enhancement_options is not None and swapped > 0:
|
||||
if mask_face and entire_mask_image is not None:
|
||||
result_image, masked_faces = enhance_image_and_mask(result_image, enhancement_options,Image.fromarray(target_img_orig),Image.fromarray(entire_mask_image).convert("L"))
|
||||
if mask_face and entire_mask_image is not None:
|
||||
result_image = enhance_image_and_mask(result_image, enhancement_options,Image.fromarray(target_img_orig),Image.fromarray(entire_mask_image).convert("L"))
|
||||
else:
|
||||
result_image = enhance_image(result_image, enhancement_options)
|
||||
elif mask_face and entire_mask_image is not None and swapped > 0:
|
||||
result_image = Image.composite(result_image,Image.fromarray(target_img_orig),Image.fromarray(entire_mask_image).convert("L"))
|
||||
|
||||
else:
|
||||
logger.status("No source face(s) in the provided Index")
|
||||
else:
|
||||
logger.status("No source face(s) found")
|
||||
logger.status("No source face(s) in the provided Index")
|
||||
else:
|
||||
logger.status("No source face(s) found")
|
||||
|
||||
return result_image, output, swapped,masked_faces
|
||||
return result_image, output, swapped
|
||||
|
||||
|
||||
|
||||
def apply_face_mask(swapped_image:np.ndarray,target_image:np.ndarray,target_face,entire_mask_image:np.array,mask_options:Union[MaskOptions,None] = None)->np.ndarray:
|
||||
logger.status("Masking Face")
|
||||
mask_generator = BiSeNetMaskGenerator()
|
||||
face = Face(target_image,Rect.from_ndarray(np.array(target_face.bbox)),1.6,mask_options.face_size,"")
|
||||
face_image = np.array(face.image)
|
||||
face_area_on_image = face.face_area_on_image
|
||||
|
||||
mask = mask_generator.generate_mask(face_image,face_area_on_image=face_area_on_image,affected_areas=mask_options.mask_areas,mask_size=0,use_minimal_area=mask_options.use_minimal_area)
|
||||
mask = cv2.blur(mask, (mask_options.mask_blur, mask_options.mask_blur))
|
||||
larger_mask = cv2.resize(mask, dsize=(face.width, face.height))
|
||||
entire_mask_image[
|
||||
face.top : face.bottom,
|
||||
face.left : face.right,
|
||||
] = larger_mask
|
||||
# def apply_face_mask(swapped_image:np.ndarray,target_image:np.ndarray,target_face,entire_mask_image:np.array)->np.ndarray:
|
||||
# logger.status("Masking Face")
|
||||
# mask_generator = BiSeNetMaskGenerator()
|
||||
# face = Face(target_image,Rect.from_ndarray(np.array(target_face.bbox)),1.6,512,"")
|
||||
# face_image = np.array(face.image)
|
||||
# process_face_image(face)
|
||||
# face_area_on_image = face.face_area_on_image
|
||||
# mask = mask_generator.generate_mask(face_image,face_area_on_image=face_area_on_image,affected_areas=["Face"],mask_size=0,use_minimal_area=True)
|
||||
# mask = cv2.blur(mask, (12, 12))
|
||||
# """entire_mask_image = np.zeros_like(target_image)"""
|
||||
# larger_mask = cv2.resize(mask, dsize=(face.width, face.height))
|
||||
# entire_mask_image[
|
||||
# face.top : face.bottom,
|
||||
# face.left : face.right,
|
||||
# ] = larger_mask
|
||||
|
||||
|
||||
result = Image.composite(Image.fromarray(swapped_image),Image.fromarray(target_image), Image.fromarray(entire_mask_image).convert("L"))
|
||||
return np.array(result)
|
||||
|
||||
# result = Image.composite(Image.fromarray(swapped_image),Image.fromarray(target_image), Image.fromarray(entire_mask_image).convert("L"))
|
||||
# return np.array(result)
|
||||
|
||||
|
||||
def correct_face_tilt(angle: float) -> bool:
|
||||
|
||||
@ -577,7 +903,22 @@ def color_generator(colors):
|
||||
|
||||
|
||||
color_iter = color_generator(colors)
|
||||
|
||||
def process_face_image(
|
||||
face: Face,
|
||||
**kwargs,
|
||||
) -> Image:
|
||||
image = np.array(face.image)
|
||||
overlay = image.copy()
|
||||
cv2.rectangle(overlay, (0, 0), (image.shape[1], image.shape[0]), next(color_iter), -1)
|
||||
l, t, r, b = face.face_area_on_image
|
||||
cv2.rectangle(overlay, (l, t), (r, b), (0, 0, 0), 10)
|
||||
if face.landmarks_on_image is not None:
|
||||
for landmark in face.landmarks_on_image:
|
||||
cv2.circle(overlay, (int(landmark.x), int(landmark.y)), 6, (0, 0, 0), 10)
|
||||
alpha = 0.3
|
||||
output = cv2.addWeighted(image, 1 - alpha, overlay, alpha, 0)
|
||||
|
||||
return Image.fromarray(output)
|
||||
def dilate_erode(img: Image.Image, value: int) -> Image.Image:
|
||||
"""
|
||||
The dilate_erode function takes an image and a value.
|
||||
|
||||
@ -1,10 +1,11 @@
|
||||
app_title = "ReActor"
|
||||
version_flag = "v0.5.0"
|
||||
version_flag = "v0.7.0-b6"
|
||||
|
||||
from scripts.reactor_logger import logger, get_Run, set_Run
|
||||
from scripts.reactor_globals import DEVICE
|
||||
|
||||
is_run = get_Run()
|
||||
|
||||
if not is_run:
|
||||
logger.status(f"Running {version_flag}")
|
||||
logger.status(f"Running {version_flag} on Device: {DEVICE}")
|
||||
set_Run(True)
|
||||
|
||||
86
scripts/reactor_xyz.py
Normal file
86
scripts/reactor_xyz.py
Normal file
@ -0,0 +1,86 @@
|
||||
'''
|
||||
Thanks @ledahu for contributing
|
||||
'''
|
||||
|
||||
from modules import scripts
|
||||
from modules.shared import opts
|
||||
|
||||
from scripts.reactor_helpers import (
|
||||
get_model_names,
|
||||
get_facemodels
|
||||
)
|
||||
|
||||
# xyz_grid = [x for x in scripts.scripts_data if x.script_class.__module__ == "xyz_grid.py"][0].module
|
||||
|
||||
def find_module(module_names):
|
||||
if isinstance(module_names, str):
|
||||
module_names = [s.strip() for s in module_names.split(",")]
|
||||
for data in scripts.scripts_data:
|
||||
if data.script_class.__module__ in module_names and hasattr(data, "module"):
|
||||
return data.module
|
||||
return None
|
||||
|
||||
def bool_(string):
|
||||
string = str(string)
|
||||
if string in ["None", ""]:
|
||||
return None
|
||||
elif string.lower() in ["true", "1"]:
|
||||
return True
|
||||
elif string.lower() in ["false", "0"]:
|
||||
return False
|
||||
else:
|
||||
raise ValueError(f"Could not convert string to boolean: {string}")
|
||||
|
||||
def choices_bool():
|
||||
return ["False", "True"]
|
||||
|
||||
def choices_face_models():
|
||||
return get_model_names(get_facemodels)
|
||||
|
||||
def float_applier(value_name:str, min_range:float = 0, max_range:float = 1):
|
||||
"""
|
||||
Returns a function that applies the given value to the given value_name in opts.data.
|
||||
"""
|
||||
def validate(value_name:str, value:str):
|
||||
value = float(value)
|
||||
# validate value
|
||||
if not min_range == 0:
|
||||
assert value >= min_range, f"Value {value} for {value_name} must be greater than or equal to {min_range}"
|
||||
if not max_range == 1:
|
||||
assert value <= max_range, f"Value {value} for {value_name} must be less than or equal to {max_range}"
|
||||
def apply_float(p, x, xs):
|
||||
validate(value_name, x)
|
||||
opts.data[value_name] = float(x)
|
||||
return apply_float
|
||||
|
||||
def bool_applier(value_name:str):
|
||||
def apply_bool(p, x, xs):
|
||||
x_normed = bool_(x)
|
||||
opts.data[value_name] = x_normed
|
||||
# print(f'normed = {x_normed}')
|
||||
return apply_bool
|
||||
|
||||
def str_applier(value_name:str):
|
||||
def apply_str(p, x, xs):
|
||||
opts.data[value_name] = x
|
||||
return apply_str
|
||||
|
||||
|
||||
def add_axis_options(xyz_grid):
|
||||
extra_axis_options = [
|
||||
xyz_grid.AxisOption("[ReActor] CodeFormer Weight", float, float_applier("codeformer_weight", 0, 1)),
|
||||
xyz_grid.AxisOption("[ReActor] Restorer Visibility", float, float_applier("restorer_visibility", 0, 1)),
|
||||
xyz_grid.AxisOption("[ReActor] Face Mask Correction", str, bool_applier("mask_face"), choices=choices_bool),
|
||||
xyz_grid.AxisOption("[ReActor] Face Models", str, str_applier("face_model"), choices=choices_face_models),
|
||||
]
|
||||
set_a = {opt.label for opt in xyz_grid.axis_options}
|
||||
set_b = {opt.label for opt in extra_axis_options}
|
||||
if set_a.intersection(set_b):
|
||||
return
|
||||
|
||||
xyz_grid.axis_options.extend(extra_axis_options)
|
||||
|
||||
def run():
|
||||
xyz_grid = find_module("xyz_grid.py, xy_grid.py")
|
||||
if xyz_grid:
|
||||
add_axis_options(xyz_grid)
|
||||
Loading…
x
Reference in New Issue
Block a user