1-Click Installers and Upgraded APP for LivePortrait - Blazing Fast Static Image or Video to Video to Talking and Moving Animation

New post is here use this one : https://www.patreon.com/posts/119254105

Use new post : https://www.patreon.com/posts/119254105

This post no longer maintained

Use new one : https://www.patreon.com/posts/119254105

Patreon exclusive posts index to find our scripts easily, Patreon scripts updates history to see which updates arrived to which scripts and amazing Patreon special generative scripts list that you can use in any of your task.

Join discord to get help, chat, discuss and also tell me your discord username to get your special rank : SECourses Discord

Please also Star, Watch and Fork our Stable Diffusion & Generative AI  GitHub repository and join our Reddit subreddit and follow me on LinkedIn (my real profile)

=======

LivePortrait: Efficient Portrait Animation with Stitching and Retargeting Control

This is the newest open source project even better than the paid apps. Its speed is mind blowing. Like you can generate 20 seconds animation in lesser than 50 seconds with RTX 3090.

This is the official repository ( https://github.com/KwaiVGI/LivePortrait ) 

27 October 2024

• Updated to V11 installer : LP_V12.zip

• Upgraded to Torch 2.5 and xFormers 0.0.28.post2 and Triton 3.1 and ONNX GPU 1.19.0 both for Windows and Linux

• New Windows requirements are CUDA 12.4, cuDNN 8.9.7, C++ Tools, Git, FFMPEG

  • Full tutorial for this : https://youtu.be/DrhUHnYfwC0

• Tested on Windows, RunPod and Massed Compute

  • For RunPod new mandatory template is :

    • RunPod Pytorch 2.2.0

    • runpod/pytorch:2.2.0-py3.10-cuda12.1.1-devel-ubuntu22.04

• If you get cuDNN related error, try different onnx-gpu version like

  • pip install onnxruntime-gpu==1.17.1 or pip install onnxruntime-gpu==1.19.2 and such

  • This error happens due to cuDNN version mismatch

• You have to use a short installation path like C:\LP_V11 

  • Otherwise you will get very long path error

8 August 2024

• Animals Live animation added

22 July 2024

• Updated to V4

• Source maximum dimension option added

• So let's say you can modify 1920x1080p video and get same output resolution

• If you have installed V3 just run Windows_Update_Version.bat file. Zip file still same

21 July 2024

• Please make a fresh install since we changed the GitHub repository

• A lot of bugs fixed

• Video to video option added

• Working amazing just try it

• For video to video, don't forget to pick video 🎞️ Source Video in the step 1 when uploading source

10 July 2024 Update

• Full Windows tutorial with manually written captions / subtitles and video chapters published

• 🔗 https://youtu.be/FPtpNrmuwXk

• Full Cloud (Mac users follow this) Tutorial for Massed Compute, RunPod & Kaggle with manually written captions / subtitles and video chapters published

• 🔗 https://youtu.be/wG7oPp01COg

• If you also upvote this Reddit Post I appreciate that very much

8 July 2024 Update

• Amazing 59 demo_portrait_images.zip added to the post that you can test and use

• I have recorded myself as a text experimental driving video that you can use :

• example_driving_video_SECourses_73_seconds.mp4

• You can download 59 demo portrait fully rendered (animated) videos by clicking here (3.74 GB)

Our New Features

• Fixed FPS issue

• Audio is automatically saved in output videos

• Output video naming improved and overwriting previous files issue fixed

• Open folders tab added (works on Windows and desktop Linux systems)

• New feature Target Eye Lip Open Ratio added

• When this is enabled, you can change the opening rate of mouth and the eyelid. However when this is enabled, the movement of the head is currently lost and not maintained.

• 1-Click to install on Windows, RunPod, Massed Compute and Kaggle.

How to Use

• Extract attached zip file and install and use on your desired platform.

• The generated animations will be automatically saved inside animations folder.

Requirements for Windows

• Works with as low as 4 GB VRAM GPUs

• New Windows requirements are CUDA 12.4, cuDNN 8.9.7, C++ Tools, Git, FFMPEG

  • Full tutorial for this : https://youtu.be/DrhUHnYfwC0

• Windows full tutorial : https://youtu.be/FPtpNrmuwXk

Cloud Requirements

• Full Cloud (Mac users follow this) tutorial for Massed Compute, RunPod and Kaggle : https://youtu.be/wG7oPp01COg

• Windows tutorial : https://youtu.be/FPtpNrmuwXk

• For RunPod and Massed Compute follow instructions inside the zip file.

• You will see Massed_Compute_Instructions_READ.txt and Runpod_Instructions_READ.txt files.

• For Kaggle follow instructions on the Kaggle notebook.

• To be able to use Kaggle notebook get a free account and verify your phone number.

• Zip file also has demo material so you can quickly test. 

• Upload / download big files / models on cloud via Hugging Face tutorial : https://youtu.be/X5WVZ0NMaTg

• How to use permanent storage system of RunPod (storage network volume) : https://youtu.be/8Qf4x3-DFf4

• Massive RunPod tutorial (shows runpodctl) : https://youtu.be/QN1vdGhjcRc 

Usage Tips

• If you feel the animated video has some jitter, please adjust the driving_smooth_observation_variance parameter in src/config/argument_config.py. We recommend that you change it by a factor of 10. For example, if the current value is 3e-7 and you feel the animated result is a bit jittery, you can adjust it to 3e-6, and so on. If you are using app.py, the corresponding adjustment location is motion smooth strength (v2v).

• What does cropping do : https://github.com/KwaiVGI/LivePortrait/issues/187#issuecomment-2242107784

• In the driving video first frame must have mouth closed - not open, not smiling, not showing teeths etc

•