search

ali-vilab / UniAnimate

Code for Paper "UniAnimate: Taming Unified Video Diffusion Models for Consistent Human Image Animation".
https://unianimate.github.io/
524 stars 29 forks source link

Is the output size must (512,768) or (768, 1216) #33

Open xiaohutongxue-sunny opened 5 days ago

xiaohutongxue-sunny commented 5 days ago

when i modify output size in UniAnimate_infer_long.yaml to (768, 1368) an error occured, Traceback (most recent call last): File "/home/UniAnimate/inference.py", line 18, in INFER_ENGINE.build(dict(type=cfg_update.TASK_TYPE), cfg_update=cfg_update.cfg_dict) File "/home/UniAnimate/utils/registry.py", line 107, in build return self.build_func(*args, kwargs, registry=self) File "/home/UniAnimate/utils/registry_class.py", line 7, in build_func return build_from_config(cfg, registry, kwargs) File "/home/UniAnimate/utils/registry.py", line 69, in build_from_config raise Exception(f"Failed to invoke function {req_type_entry}, with {e}") Exception: Failed to invoke function <function inference_unianimate_long_entrance at 0x7fb6cf08e0d0>, with Sizes of tensors must match except in dimension 1. Expected size 44 but got size 43 for tensor number 1 in the list.

wangxiang1230 commented 5 days ago

Hi, thanks for your attention. The input resolution should be divisible by 64 since vae and 3d-unet are used.