search

Tencent / MimicMotion

High-Quality Human Motion Video Generation with Confidence-aware Pose Guidance
https://tencent.github.io/MimicMotion/
Other
1.95k stars 166 forks source link

Extended Inference Time #7

Closed yachty66 closed 5 months ago

yachty66 commented 5 months ago

I am having a problem where, when I run the model for inference, it shows me that the inference time is about 40 minutes. I am following all the mentioned steps, with the only difference being the GPU (A40) instead of V100.

vrgamegirl19 commented 5 months ago

Hey, how did u get it working? I literally spent 5 hours and still couldn't get it to work. I'm thinking I'm missing files or my structure is wrong. Can u take a screen grab of your folders structure? I posted my issue in the issue list.

xuanlongORZ commented 5 months ago

I also come across this issue. I figure out that the the number of frames is too big for the given example. I put sample_stride in test.ymal as 8 and it's now able to generate a short video in around 15min using Nvidia L20 (it is better to set a lower fps for this moment).

sweetgrass commented 5 months ago

I also come across this issue. I figure out that the the number of frames is too big for the given example. I put sample_stride in test.ymal as 8 and it's now able to generate a short video in around 15min using Nvidia L20 (it is better to set a lower fps for this moment).

老哥，这个跟之前的那个Muse Pose比，效果怎么样？

yachty66 commented 5 months ago

It seems to be a normal inference time. With A100, it currently takes me 20 minutes to run the demo on inference. Closing this issue.