Deep-Live-Cam
Deep-Live-Cam copied to clipboard
Very Low FPS (1 FPS) CPU 60% Usage, GPU barely 5% (RTX 2060 Super + i9-10900X)
Hi everyone, I'm currently facing a performance issue with Deep Live Cam. GPU: NVIDIA RTX 2060 Super 8GB CPU: Intel i9-10900X RAM: 16GB OS: Windows 10
When running the project, I only get around 1 FPS. While monitoring my system, I noticed: CPU usage is at around 60% GPU usage stays extremely low (under 5%)
It seems like the GPU is not being properly utilized for the heavy lifting (inference, frame processing, etc.). I'm running the project with the following command:
python run.py --execution-provider cuda But still, no improvement. Has anyone managed to get smooth performance on Windows with Deep Live Cam? Is there something I'm missing in the setup or code to ensure GPU is fully used? Any suggestions, tips, or potential fixes are greatly appreciated!
I found two similar discussions that might be relevant to your issue:
-
The workflow is published as a tool and the output variable files cannot fetch the file - This discussion is still open and might provide insights into handling file outputs in workflows [1].
-
How can I output a file in
Workflow- This discussion is also open and addresses issues related to file outputs in workflows [2].
Regarding your issue, when deploying a workflow as a web app, Dify processes file outputs by checking if they contain files with a specific identity (__dify__file__). These files are then displayed in a FileList component, allowing users to download or preview them. If the files are not appearing, it might be due to a misconfiguration or an issue with how the files are being processed in the deployed environment [3].
Additionally, ensure that your self-hosted setup is correctly configured for file export. You need to set the FILES_URL parameter in the docker/.env file to match your Dify access address, and verify that the volumes section in your docker-compose-template.yaml file includes a mount for the storage directory to the container [4][5].
To continue talking to Dosu, mention @dosu.
Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other