TripoSR Image to 3D

fal-ai/triposr
State of the art Image to 3D Object generation
Inference
Commercial use

Input

Additional Settings

Customize your input with more control.

Result

Idle

What would you like to do next?

Your request will cost $0.07 per generation.

Logs

TripoSR | [image-to-3d]

Stability AI's TripoSR delivers 3D mesh reconstruction from single images in under 0.5 seconds at $0.07 per generation. Trading multi-view requirements for speed, this transformer-based model produces GLB or OBJ outputs with configurable marching cubes resolution up to 1024. Built for game developers, industrial designers, and AR/VR creators who need rapid 3D asset generation without photogrammetry rigs.

Use Cases: Game Asset Prototyping | Industrial Design Visualization | E-commerce 3D Previews


Performance

TripoSR processes single images into production-ready 3D meshes in under 0.5 seconds, making it approximately 10-20x faster than traditional photogrammetry workflows while maintaining cost efficiency at $0.07 per generation.

MetricResultContext
Inference SpeedUnder 0.5 secondsFeed-forward transformer architecture on GPU
Cost per Generation$0.0714 generations per $1.00 on fal
Output FormatsGLB, OBJStandard 3D mesh formats with texture support
Resolution Control32-1024 marching cubesDefault 256, above 512 not recommended for performance
Background RemovalBuilt-in preprocessingAutomatic foreground isolation with 0.5-1.0 ratio control

Single-Image 3D Reconstruction Without Multi-View Capture

TripoSR uses a Large Reconstruction Model (LRM) architecture to generate complete 3D geometry from a single 2D image, eliminating the need for multiple camera angles or depth sensors that traditional photogrammetry requires.

What this means for you:

  • Instant asset generation: Convert concept art or product photos into 3D meshes in under 0.5 seconds without setting up multi-camera rigs or turntables

  • Configurable mesh density: Adjust marching cubes resolution from 32 to 1024 based on your quality-performance tradeoff, with 256 as the optimized default

  • Automatic background handling: Built-in foreground isolation removes manual masking work, with adjustable foreground ratio (0.5-1.0) for precise subject extraction

  • Production-ready output: Export directly to GLB for web/AR applications or OBJ for traditional 3D pipelines, compatible with Blender, Unity, and Unreal Engine


Technical Specifications

SpecDetails
ArchitectureTripoSR
Input FormatsSingle image (JPEG, PNG, WebP, GIF, AVIF) via URL
Output FormatsGLB (default), OBJ mesh with textures
Marching Cubes Resolution32-1024 (default: 256)
LicenseCommercial use permitted

API Documentation | Quickstart Guide | Enterprise Pricing


How It Stacks Up

Tripo3D Image to 3D ($0.06) – TripoSR prioritizes speed with its feed-forward transformer architecture for rapid prototyping workflows at $0.07 per generation. Tripo3D offers enhanced detail preservation and texture quality for production-ready assets where final output fidelity justifies the similar cost point and potentially longer processing time.