search

Inferencer / LipSick

🤢 LipSick: Fast, High Quality, Low Resource Lipsync Tool 🤮
The Unlicense
187 stars 29 forks source link

Lipsick Never worked not even a single time for me #26

Open Ex3cutionerX opened 5 months ago

Ex3cutionerX commented 5 months ago

I've tried it multiple times with different videos at 25fps and different resolutions, it never works and give me the error raise ValueError("No faces found in the frame.") or the error 13 Permission denied, i have full permission of the folders and even changed then. I've installed all the requirements, what am i doing wrong?

Inferencer commented 5 months ago

Later today ill get to the bottom of it, i'm sorry about your experience, I'll give it my full attention as this is a common issue that I have been avoiding although I have been doing some test with users I suspect there could either be a bug that is generating an extra frame in extraction or an issue with the face tracking code but I'll find it and fix it as in 6 hours time I will have 2 days availability to be alone with my PC.

The permission 'denied error' part of the message just appears in addition to whenever there is an error that I haven't coded to give a custom response so we can ignore that part.

Inferencer commented 5 months ago

I found the issue was with dlib, not sure if we are using hog or cnn dlib models either way I needed to upgrade the face tracker to sfd as per the roadmap which I have confirmed works on all user vids that had issues with face tracking, dlib was slow too as the wheels were not built with cuda so that saves me a job, I'm just bug testing and implementing the new tracker so hopefully have that later today working at a reasonable speed, I will need to make some changes to the env and will probs switch to only using cuda 11.8 on windows due to nightly deprecation of 11.6

Kafka588 commented 5 months ago

I found the issue was with dlib, not sure if we are using hog or cnn dlib models either way I needed to upgrade the face tracker to sfd as per the roadmap which I have confirmed works on all user vids that had issues with face tracking, dlib was slow too as the wheels were not built with cuda so that saves me a job, I'm just bug testing and implementing the new tracker so hopefully have that later today working at a reasonable speed, I will need to make some changes to the env and will probs switch to only using cuda 11.8 on windows due to nightly deprecation of 11.6

I found the issue was with dlib, not sure if we are using hog or cnn dlib models either way I needed to upgrade the face tracker to sfd as per the roadmap which I have confirmed works on all user vids that had issues with face tracking, dlib was slow too as the wheels were not built with cuda so that saves me a job, I'm just bug testing and implementing the new tracker so hopefully have that later today working at a reasonable speed, I will need to make some changes to the env and will probs switch to only using cuda 11.8 on windows due to nightly deprecation of 11.6

im having hard time trying to work on my gpu with environment u have provided it was working on my cpu and there was something like model not found tensorflow then i tried to install different versions of cuda and tensorflow still not working could you pls provide new environment???

giandre commented 3 months ago

This is how claude.ai helped me resolve this issue..

  1. Run Command prompt as ADMIN - this is key.
  2. Go to your lipsick folder
  3. conda activate lipsick
  4. set KMP_DUPLICATE_LIB_OK=TRUE 5.type python app.py

Now the problem I have is with double chin and the gradio interface not allowing me to select anything in the advanced options.