Show 1

Show 1 generates video from text prompts on your own hardware. It's aimed at researchers and developers who want to experiment with video generation while keeping the depicted scene and motion close to the prompt.

• Text-to-video generation: Describe a scene or action in words and generate a video using the published base model weights.
• Frame interpolation: A separate interpolation model adds frames between generated frames.
• Super-resolution: Dedicated models increase video resolution, using a hybrid approach designed to preserve prompt details while reducing GPU memory use.
• Local output: The generation script saves videos from each stage locally as GIF files by default.

The project reports roughly 15 GB of GPU memory use during inference for its hybrid approach. That figure is a reported result, rather than a guarantee for every workload. Show 1 uses PyTorch, and its code builds on diffusers, DeepFloyd IF, ModelScope and Zeroscope.

Show Lab at the National University of Singapore publishes the model weights on Hugging Face. The generation script downloads the weights it needs, so fetching them requires internet access. Colab and Replicate demos provide hosted ways to try the model; those runs take place outside your local machine. Commercial use requires approval from the project.