当前位置：首页 > news >正文

2023-12-04 AIGC-Stable Diffusion和SadTalker-搭建及使用

news 来源：原创 2025/5/11 5:12:32

摘要:

2023-12-04 AIGC-SadTalker-搭建及使用

代码仓库:

https://github.com/Stability-AI/stablediffusion

https://github.com/camenduru/stable-diffusion-webui-colab

https://github.com/OpenTalker/SadTalker

https://github.com/adofsauron/SadTalker-Video-Lip-Sync

文档:

Stable Diffusion一键安装包 Windows版 - Stable Diffusion中文网

类似D-ID的免费开源虚拟数字人制作工具SadTalker搭建教程及效果演示 - 哔哩哔哩

SadTalker调参实验 - 知乎

https://github.com/OpenTalker/SadTalker/blob/main/docs/best_practice.md

SadTalker参数:

(sadtalker) PS D:\sd\SadTalker>  python inference.py  --help
usage: inference.py [-h] [--driven_audio DRIVEN_AUDIO] [--source_image SOURCE_IMAGE] [--ref_eyeblink REF_EYEBLINK] [--ref_pose REF_POSE] [--checkpoint_dir CHECKPOINT_DIR]
                    [--result_dir RESULT_DIR] [--pose_style POSE_STYLE] [--batch_size BATCH_SIZE] [--size SIZE] [--expression_scale EXPRESSION_SCALE]
                    [--input_yaw INPUT_YAW [INPUT_YAW ...]] [--input_pitch INPUT_PITCH [INPUT_PITCH ...]] [--input_roll INPUT_ROLL [INPUT_ROLL ...]] [--enhancer ENHANCER]
                    [--background_enhancer BACKGROUND_ENHANCER] [--cpu] [--face3dvis] [--still] [--preprocess {crop,extcrop,resize,full,extfull}] [--verbose] [--old_version]
                    [--net_recon {resnet18,resnet34,resnet50}] [--init_path INIT_PATH] [--use_last_fc USE_LAST_FC] [--bfm_folder BFM_FOLDER] [--bfm_model BFM_MODEL]
                    [--focal FOCAL] [--center CENTER] [--camera_d CAMERA_D] [--z_near Z_NEAR] [--z_far Z_FAR]

optional arguments:
  -h, --help            show this help message and exit
  --driven_audio DRIVEN_AUDIO
                        path to driven audio
  --source_image SOURCE_IMAGE
                        path to source image
  --ref_eyeblink REF_EYEBLINK
                        path to reference video providing eye blinking
  --ref_pose REF_POSE   path to reference video providing pose
  --checkpoint_dir CHECKPOINT_DIR
                        path to output
  --result_dir RESULT_DIR
                        path to output
  --pose_style POSE_STYLE
                        input pose style from [0, 46)
  --batch_size BATCH_SIZE
                        the batch size of facerender
  --size SIZE           the image size of the facerender
  --expression_scale EXPRESSION_SCALE
                        the batch size of facerender
  --input_yaw INPUT_YAW [INPUT_YAW ...]
                        the input yaw degree of the user
  --input_pitch INPUT_PITCH [INPUT_PITCH ...]
                        the input pitch degree of the user
  --input_roll INPUT_ROLL [INPUT_ROLL ...]
                        the input roll degree of the user
  --enhancer ENHANCER   Face enhancer, [gfpgan, RestoreFormer]
  --background_enhancer BACKGROUND_ENHANCER
                        background enhancer, [realesrgan]
  --cpu
  --face3dvis           generate 3d face and 3d landmarks
  --still               can crop back to the original videos for the full body aniamtion
  --preprocess {crop,extcrop,resize,full,extfull}
                        how to preprocess the images
  --verbose             saving the intermedia output or not
  --old_version         use the pth other than safetensor version
  --net_recon {resnet18,resnet34,resnet50}
                        useless
  --init_path INIT_PATH
                        Useless
  --use_last_fc USE_LAST_FC
                        zero initialize the last fc
  --bfm_folder BFM_FOLDER
  --bfm_model BFM_MODEL
                        bfm model
  --focal FOCAL
  --center CENTER
  --camera_d CAMERA_D
  --z_near Z_NEAR
  --z_far Z_FAR