Today's article topic introduces one of the latestAge TransformationFraming:SAM. This is a process framework that can demonstrate the process of changing the look and appearance of an individual character over time as the character transforms with age. The framework encodes real facial images directly into the latent space of a pre-trained unconditional GAN (e.g., StyleGAN), controlled by a given age offset. It also uses a pre-trainedAge Regression Network, explicitly instructing the encoder to generate potential codes corresponding to the desired age.SAMmethods treat successive age processes as regression tasks between the input age and the desired target age, providing fine-grained control over the generated images. Furthermore, unlike other methods that operate only in potential space and use a priori control of the age paths, theSAMmethod more decoupled and nonlinear paths. The end-to-end nature of the method, coupled with the rich semantic potential space of StyleGAN, allows for further editing of the generated images.
- github: https://github.com/yuval-alaluf/SAM
- Online experience: https://replicate.com/yuval-alaluf/sam
- Technical paper: https://arxiv.org/abs/2102.02754
Case Demonstration
The community already hascomfyui-faceagingofComfyUIPluginsSupported, the plugin can support -100 to +100 (-50 to +100 recommended) Age of the conversion straight out of the image. The plugin can be installed via git in the plugin manager.
- comfyui-facing plugin address:: https://github.com/leadbreak/comfyui-faceaging
- Environmental requirementsLinux or macOS (window needs to be tested). NVIDIA GPU + CUDA CuDNN (GPU support is required, the framework does not have built-in CPU support); Python3 environment.
- Dependent Environment Installation: Start the command line to install the corresponding dependency with the following command:
sudo apt-get install cmake
pip install dlib ninja
- (Optional) The CUDA environment should be set up in WSL2.: The command is shown below:
sudo apt install nvidia-cuda-toolkit
- Download the corresponding model: Need to download the modelsam_ffhq_aging.pt and shape_predictor_68_face_landmarks.datTo the catalogComfyUI/models/face_agingThe corresponding commands are shown below. The corresponding commands are shown below:sam_ffhq_aging.pt: https://drive.usercontent.google.com/download?id=1XyumF6_fdAxFmxpFcmPf-q84LU_22EMC&export=download&authuser=0 ;shape_predictor_68_face_landmarks.dat:: https://github.com/italojs/facial-landmarks-recognition/raw/master/shape_predictor_68_face_landmarks.dat
Flux Wensheng Diagram Workflow
Flux Vinyl Chart interested students can refer to LIBLIB online running workflow : FLUX [Continued] : 12B parameters 23G largest open source Vinyl Chart model , Dev version straight out of the stunning beauty to enjoy!
The ComfyUI workflow and models mentioned in this article can be downloaded on LIBLIBAI or run online for experience:
- FLUX.1 miles of cloth is available to run online - Dark Forest Studios:
https://www.liblib.art/modelinfo/488cd9d58cd4421b9e8000373d7da123
- F.1 - Cherry Dreams and Flowing Lights - Watery Mekong Geling Fragrance:
https://www.liblib.art/modelinfo/134c6dd95aef48e98a22b24e003e026b
- Workflow - Flux text | graphs to generate graphs + LORA + prompts backpropagation one-click switch workflow:
https://www.liblib.art/modelinfo/782aacd70f604da39e83368c696a02a8
Face Aging Workflow
Face AgingWorkflow uploadedLIBLIB PlatformAvailable for download at: https://www.liblib.art/modelinfo/33cdec2c32634338a4e530414e29b564?versionUuid=7d466ce3533b424880622b7b09484aac
Notice: There are some requirements for belonging to images here, not all images show the effect well (the license photo worked best in the test). Currently belongs to the entertainment image stream.
1.Flux Vincennes
pl, real photo, The image is a high-resolution photograph, A photograph of a beautiful Asian girl in school uniform,with a solid blue background,in the She has medium-length hair, a round face shape, and is wearing a white shirt, tie, and black blazer. Her makeup is simple, and she has a gentle,youthful smile with a natural expression. Her makeup is simple, and she has a gentle,youthful smile with a natural expression.,," she says.
2. Age conversion
3. More demos
The overall effect is a requirement for input images, and this workflow is currently in a position to be used only for entertainment purposes. Finally, facing the future: the only way to live up to this life is to cherish the present.