“Zeroscope stems from Modelscope, a multilevel text-to-video diffusion model with 1.7 billion parameters. It generates video content based on textual descriptions. It refines this concept, offering higher resolution, without the Shutterstock watermark, and closer to a 16:9 aspect ratio. Zeroscope features two components: Zeroscope_v2 567w, designed for rapid content creation in a resolution of 576×320 pixels to explore video concepts. Quality videos can then be upscaled to a “high definition” resolution of 1024×576 using zeroscope_v2 XL.”
Read more at the decoder.com: Zeroscope is a free text-to-video model that runs on modern graphics cards