HunyuanVideo

HunyuanVideo is an AI video generation model for creators and developers who want to generate clips on their own GPUs. Downloadable weights and inference code support generation from text prompts or an input image.

• Text and image inputs: Generate video from a written description or use the image-to-video model to start with an existing image.
• Workflow integration: Use the model through ComfyUI or Hugging Face Diffusers, with a Gradio web demo also available.
• GPU options: Choose FP8 quantized weights as an alternative to full weights, or use multiple GPUs to speed up inference.
• Prompt rewriting: Use the separate Hunyuan-Large rewrite model to adapt prompts or expand descriptions of composition, lighting and camera movement.

The prompt rewrite model has downloadable weights for local deployment. Its visually focused mode can lose some details of the original request.

Tencent provides PyTorch model definitions, pretrained weights and sampling code for developers who want to work directly with the model. Local inference uses your GPU hardware.