ali-vilab

/

modelscope-damo-text-to-video-synthesis

like 439

As per discussion in the paper page of Webvid, one of the datasets used for training, can be used for non-commercial purposes only, primarily for research. So maybe the intent of the dataset developer was to avoid the dataset being used to train commercial models.

Moreover, the model will likely generate images with watermarks from a company, and therefore, may confuse or deceive users of commercial applications using this model. As such, it’s my (personal) belief that this repository should have a non-commercial license, or, a restriction to not use the model without disclaiming the outputs are machine-generated and must not be used to confuse the user with regards to the commercial origin of the output.