HunyuanVideo-1.5

HunyuanVideo-1.5 (Tencent)

Large-scale text-to-video generation focused on cinematic quality and complex motion.

• High-capacity video model. Trained at scale for text-to-video generation, with an emphasis on long-range temporal coherence, rich motion, and visually dense scenes.

• Strong scene understanding. Handles multi-object scenes, camera movement, and environmental motion better than smaller video models, with fewer temporal artifacts.

• Quality over speed. Requires substantial compute and VRAM, trading fast iteration for higher visual fidelity and more stable videos.

• General-purpose text control. Responds well to descriptive prompts without needing reference images, making it suitable for open-ended video generation.

• Research-grade release. Intended for evaluation, benchmarking, and exploration rather than lightweight or real-time use.

Why pick it for Norman AI?

HunyuanVideo-1.5 is a solid choice when you want to explore high-end text-to-video quality. It’s useful for benchmarking large video models, testing complex prompts, and comparing cinematic motion and scene consistency.

response = await norman.invoke(
    {
        "model_name": "hunyuanVideo-1.5",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": "A cat playing with a ball on mars"
            }
        ]
    }
)