FaceChain Make Face portraits for Linkedin and Social media + I2VGen XL makes videos from images

Описание

Linkedin profile with AI social media portraits
+ image to video make small animations with just a picture
facechain + Modelscope I2VGen XL video
#linkedin #portraits #chatgpt #alpaca #ai #StableDiffusion

▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
SOCIAL MEDIA LINKS!
✨ Support my (*・‿・)ノ⌒*:･ﾟ✧

character available at https://ko-fi.com/impactframes/shop
SD related https://civitai.com/user/impactframes
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
VIDEO LINKS??️o(≧o≦)o?
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
facechain
https://github.com/camenduru/facechain-colab
https://github.com/modelscope/facechain
https://github.com/modelscope/modelscope
i2VGenXL
https://github.com/camenduru/I2VGen-XL-colab
https://modelscope.cn/models/damo/Video-to-Video/summary
https://github.com/modelscope/modelscope
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
?NOTES
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Easy
download pinokio and install the scripts
https://pinokio.computer/

Hard
Local install Linux https://pastebin.com/tDUu0WYB

(IF NUMPY OR ACCELERATE PROBLEMS)
pip install --upgrade accelerate numpy

PYTORCH AND TENSORFLOW
https://pytorch.org/get-started/locally/
https://www.tensorflow.org/install/pip
export your CUDA variable b4 running
export CUDA_HOME=/usr/local/cuda-11.8
echo $CUDA_HOME

FACECHAIN (requires 19GB GPU vram)

GIT_LFS_SKIP_SMUDGE=1 git clone https://github.com/modelscope/facechain.git --depth 1
cd facechain
micromamba create -n facechain -c conda-forge -c defaults python=3.10
micromamba activate facechain
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip3 install -r requirements.txt
pip3 install -U openmim
mim install mmcv-full==1.7.0 or maybe faster (pip3 install https://modelscope.oss-cn-beijing.aliyuncs.com/packages/mmcv/mmcv_full-1.7.0-cp310-cp310-linux_x86_64.whl)

export CUDA_HOME=/usr/local/cuda-11.8
echo $CUDA_HOME
python3 app.py


MODELSCOPE + I2VGEN (requires min 24GB Vram)

micromamba create -n modelscope -c conda-forge -c defaults python=3.10
micromamba activate modelscope
pip install -r modelscope/requirements.txt
pip install -e modelscope
pip install gradio huggingface_hub
pip install jupyter notebook
pip install -q xformers==0.0.20 triton==2.0.0 gradio torchsde open_clip_torch einops rotary-embedding-torch fairscale
pip install -q pytorch-lightning torchmetrics -U
mim install mmcv-full==1.7.0 or maybe faster (pip3 install https://modelscope.oss-cn-beijing.aliyuncs.com/packages/mmcv/mmcv_full-1.7.0-cp310-cp310-linux_x86_64.whl)

continue from the sudo apt install aria section here https://pastebin.com/tDUu0WYB


download camenduru or the oficial notebook to the folder to use

EXPORT YOUR CUDA ENV VARIABLES B4 LAUNCH
export CUDA_HOME=/usr/local/cuda-11.8
echo $CUDA_HOME