extremely slow infering

[Fqlox]

I wanted to render infer with the model on I2V following the example in the readme and got infinte slow time. Whereas on Comfyui I ran a video generation in around 10 minutes (without lora). As seen below:

I ran on windows 10 with a 3090 a on conda env, python 3.12.7, torch 2.5.1, diffusers 0.31.

[wenqsun]

HI, it may be caused by the lack of GPU resources. Now we have released the online huggingface demo: https://huggingface.co/spaces/fffiloni/DimensionX

You can try our model online! It would be much faster!

[Fqlox]

HI, it may be caused by the lack of GPU resources. Now we have released the online huggingface demo: https://huggingface.co/spaces/fffiloni/DimensionX

You can try our model online! It would be much faster!

As mentioned, I run the inference on a 3090 RTX and the model works well into comfyui

[dtaddis]

I'm also experiencing very slow inferring - 120s/it, resulting in about a 90-minute total time.

Windows 11, RTX 4090, Python 3.12, Torch 2.5.1.