Hunyuan AI Models - Advanced 3D, Video & Image Generation

Hunyuan 3D Rapid (0.20-0.25 credits) generates 3D models quickly with standard quality, ideal for prototyping and iteration. Hunyuan 3D Pro (0.40-0.45 credits) offers advanced quality with adjustable polygon counts from 40k to 1.5M faces, supports up to 8 camera views for image-to-3D, and delivers professional-grade results suitable for production use. Pro mode also provides finer control over mesh density and texture resolution.

Hunyuan Image v3 Instruct features internal reasoning capabilities that allow it to understand complex, nuanced prompts more intelligently than v2.1. The v3 model can interpret intent, handle multi-step instructions, and make intelligent decisions about composition and style. It also includes a dedicated editing mode (v3 Instruct Edit) for transforming existing images. Both cost 0.05-0.08 credits, but v3 delivers superior prompt adherence and creative interpretation.

Yes, Hunyuan 3D Pro Image-to-3D supports up to 8 views: front (required), plus optional back, left, right, top, bottom, left-front, and right-front angles. More views result in better reconstruction accuracy and detail. The Rapid version supports 4 views (front, back, left, right). Simply upload your reference images and the model will generate a complete 3D mesh with proper geometry all around.

Hunyuan Video models support multiple resolutions and can generate videos with professional quality. The standard text-to-video and image-to-video models offer pro mode options for enhanced quality. You can control motion dynamics, enable stability mode to reduce artifacts, and even use custom LoRA models for specific style control. Costs range from 0.22 credits for basic text-to-video to 0.60 credits for image-to-video with LoRA customization.

Hunyuan Motion converts natural language descriptions into 3D skeletal animations for human characters. Simply describe the motion you want (e.g., 'person walking slowly', 'jumping and waving') and the model generates realistic animation data. Choose Hunyuan Motion Fast (0.46B parameters, 0.08 credits) for quick iterations or the full Hunyuan Motion (1B parameters, 0.12 credits) for maximum quality and motion complexity.

Hunyuan 3D Smart Topology (0.10 credits) optimizes high-polygon 3D meshes into clean, production-ready geometry. It converts dense meshes into efficient topologies with either triangle or quad polygons, making them suitable for animation, game engines, or further editing. Use it when you have AIGC-generated models or sculpts that need retopology for rigging, UV mapping, or real-time rendering applications.

Yes, Hunyuan Custom (0.40 credits) is specifically designed for perfect subject consistency across frames using multi-modal inputs. This model ensures that characters, objects, or brand elements maintain their appearance throughout video sequences, making it ideal for storytelling, character-driven content, and branded video production where visual consistency is critical.

Hunyuan 3D models export to industry-standard formats including GLB, FBX, and OBJ. You can choose to include PBR materials (albedo, normal, roughness maps) or export geometry-only. The Part Splitter tool specifically works with FBX format and can segment models up to 100MB with ≤30k faces into separate components for easier manipulation in 3D software.

Hunyuan Video Foley (0.25 credits) analyzes your video content and generates contextually appropriate sound effects that match the on-screen action. It understands visual cues like footsteps, object interactions, environmental sounds, and movement dynamics to create realistic audio layers. This eliminates the need for manual sound design or stock audio libraries for many common video production scenarios.