3D Creation

3D Creation

Accelerate your 3D pipeline. WaveSpeed transforms 2D images or text descriptions into high-quality 3D models in seconds. Powered by state-of-the-art algorithms like TripoSR and SF3D, generate textured meshes ready for gaming, AR/VR, and e-commerce visualization. Export directly to GLB or OBJ compatible with Blender, Unity, and Unreal Engine.

The 3D Generation Workflow

Create assets in three simple steps.

1

Input: Image, Text, or 3D Scan

Start with a single photo, a text description, or even a rough 3D scan. Models like TripoSR and Wonder3D accept a 2D image and infer the complete 3D geometry in seconds. Text-to-3D models such as Shap-E and Point-E let you describe an object in plain language and receive a 3D mesh. For higher precision, upload a multi-view scan or a point cloud for refinement. Explore our best open-source image models to generate high-quality input images.
2

Generation: AI Mesh & Texture Synthesis

The AI reconstructs a watertight 3D mesh from your input, complete with UV-mapped textures. TripoSR produces a textured mesh in under 5 seconds; InstantMesh and LGM (Large Gaussian Model) create Gaussian splats for real-time rendering. Multi-view diffusion models like Zero123++ generate consistent views that are then fused into geometry. All models run on WaveSpeed's GPU cluster for fast inference.
3

Refinement & Export

Polish the generated asset with automated remeshing, texture upscaling, and PBR material estimation. Export to industry-standard formats: GLB, OBJ, FBX, or USDZ for AR Quick Look. Assets are game-ready and compatible with Blender, Unity, and Unreal Engine. Best for e-commerce 3D product viewers, AR try-on experiences, gaming assets, and architectural visualization.

Production-Ready 3D Assets

From rapid prototyping to final game assets, AI 3D creation fits seamlessly into modern workflows.

E-Commerce 3D Viewers

Convert product photos into interactive 3D models that customers can rotate and zoom. Proven to increase conversion rates by letting shoppers inspect products from every angle before purchasing.

Gaming & Virtual Worlds

Rapidly generate game-ready meshes for characters, props, and environments. Export to Unity or Unreal Engine with PBR materials for immediate use in real-time rendering pipelines.

AR Try-On & Visualization

Export USDZ files for Apple AR Quick Look, allowing customers to place furniture, accessories, or products in their physical space directly from a mobile browser.

Architectural Visualization

Transform concept sketches or floor plans into 3D walkthroughs. Combine with AI-generated textures for photorealistic material rendering of interiors and facades.

Q & A

How long does 3D creation take?
With fast inference models like TripoSR, a basic 3D model can be generated in under 10 seconds. Higher quality models with detailed textures (using SF3D or equivalent) may take 1-2 minutes.
Are the models rigged/animated?
Currently, the AI generates static meshes (geometry + texture). Rigging and animation must be done in external software like Blender or Mixamo, although auto-rigging features are in our roadmap.
What is the poly count?
The models typically range from Low Poly (5k-10k faces) to Mid Poly (20k-50k faces), making them suitable for real-time applications like games and AR filters.
Can I edit the generated model?
Yes. The exported files (GLB/OBJ) are standard 3D formats. You can import them into Blender, Maya, or Cinema 4D to modify the geometry, adjust UVs, or change textures manually.
Does it handle transparent objects?
AI 3D generation works best with opaque objects (wood, metal, plastic). Transparent or highly reflective materials (glass, water) are challenging and may require manual shader adjustments in your 3D engine after generation.