Hi authors,

Thanks for creating VBench, this is a very nice eval benchmark suite.

I find it useful to evaluate a custom prompt_list and corresponding videos. When running vbench2_beta_i2v with custom prompt_list, the prompts does not look to be correctly propagated to the eval. I think that is related to the missing of **kwargs in build_full_info_json:

• cur_full_info_path = self.build_full_info_json(videos_path, name, dimension_list, custom_image_folder=custom_image_folder, mode=mode)
• https://github.com/Vchitect/VBench/blob/master/vbench2_beta_i2v/__init__.py#L23