Very long generation time on 3090 24GB

[gt732]

Hi,

Thank you all for the hard work and open sourcing this. I went through a quick demo on my windows 10 machine. With my own custom 768x768 image and generation takes about 10mins for a 2sec clip.

3090 24GB
80GB RAM

Any idea what could be the issue or is this expected? It takes around 10GB VRAM when generating the video once it hits 100% it maxes out my GPU.

[AA-Developer]

hi
Your problem will be that you are using the cpu instead of the gpu.
I am working on an advanced interface that fixes these errors. I will publish it tonight or tomorrow.
Currently there are several bugs in the FancyVideo code
It seems that they are just for testing the processing but I am developing a system that supports all changes in a complete interface without the need for a configuration file.
I will publish it soon.

[gt732]

@AA-Developer Sounds good ty! It seems like it uses about 10GB of VRAM during the generation part, once it hits 100% it goes haywire and eats up a ton of resources then takes 10minutes to finally save the video.

[AA-Developer]

@AA-Developer Sounds good ty! It seems like it uses about 10GB of VRAM during the generation part, once it hits 100% it goes haywire and eats up a ton of resources then takes 10minutes to finally save the video.

use this
https://github.com/AA-Developer/FancyVideo/tree/main

If you have the original installation :
Download the file
app.py
Put it in the root
/
Download the two files
i2v_ui.yaml
t2v_ui.yaml
From the path
configs/inference
https://github.com/AA-Developer/FancyVideo/tree/main/configs/inference
Put it in the same path

then run : python app.py
go to : 127.0.0.1:7860

[MaAo]

@AA-Developer Sounds good ty! It seems like it uses about 10GB of VRAM during the generation part, once it hits 100% it goes haywire and eats up a ton of resources then takes 10minutes to finally save the video.

use this https://github.com/AA-Developer/FancyVideo/tree/main

If you have the original installation : Download the file app.py Put it in the root / Download the two files i2v_ui.yaml t2v_ui.yaml From the path configs/inference https://github.com/AA-Developer/FancyVideo/tree/main/configs/inference Put it in the same path

then run : python app.py go to : 127.0.0.1:7860

Thank you for a job really well done!

[gt732]

@AA-Developer Sounds good ty! It seems like it uses about 10GB of VRAM during the generation part, once it hits 100% it goes haywire and eats up a ton of resources then takes 10minutes to finally save the video.

use this https://github.com/AA-Developer/FancyVideo/tree/main

If you have the original installation : Download the file app.py Put it in the root / Download the two files i2v_ui.yaml t2v_ui.yaml From the path configs/inference https://github.com/AA-Developer/FancyVideo/tree/main/configs/inference Put it in the same path

then run : python app.py go to : 127.0.0.1:7860

Thank you for working on this, I will give it a go when I get a chance!

[nitinmukesh]

@AA-Developer

Thank you for the changes. Please could you also include fp18 for low VRAM in your fork.