ModelScope is a tool that allows you to generate realistic videos from text. You can type in a description of a scene and ModelScope will generate a short clip that matches your input. However, ModelScope uses stock footage from Shutterstock, a website that provides high-quality images and videos for various purposes. To use ModelScope for free, you have to accept that your generated clips will have a Shutterstock watermark on them. This means that you cannot claim the clips as your own or use them for commercial purposes
you can find the model in Huggingface or in their official website with the name: “ModelScope Text to Video Synthesis“
The ModelScope developers team has mentioned that they used data from various “public” sources but they didn’t mention websites with copyrights like Shutterstock
Despite how impressive being able to turn text into video is, beware to the fact that this model may output content that reinforces or exacerbates societal biases. The training data includes LAION5B, ImageNet, Webvid and other public datasets. The model was not trained to realistically represent people or events, so using it to generate such content is beyond the model’s capabilities.
