TripoSR Documentation
Learn how to turn a single image or a text prompt into a downloadable 3D model (GLB) with TripoSR, the browser-based AI 3D reconstruction platform.
TripoSR turns a single image or a short text prompt into a textured 3D model that you can download as a GLB file. Everything runs in your browser on cloud GPUs, so you do not need a local GPU or any install. These guides walk you through both tools, the credit cost of a generation, and how to use the exported model across the web, AR, game engines, and Blender.
What you can do
- Reconstruct a 3D model from one photo with Image-to-3D.
- Generate a 3D model from a written description with Text-to-3D.
- Download the result as a GLB file (a textured mesh) ready for the web, AR, Unity, Unreal, and Blender.
- Run it all in the browser on cloud GPUs, with no local setup.
An account is required
Generating a model requires a signed-in TripoSR account. There is no anonymous or no-signup generation path. Sign in first, then start a task. Each generation costs 12 credits.
Start here
Getting started
Make your first 3D model step by step, from sign-in to downloading the GLB.
Image-to-3D guide
Pick good input photos, know what to expect, and understand the credit cost.
Text-to-3D guide
Write prompts that describe a single, well-defined object for 3D.
Exporting the GLB
Use the downloaded GLB on the web, in AR, in Unity and Unreal, and in Blender.
How it works
TripoSR is built on the open-source TripoSR model (see the research paper). You provide an input (an image or a text prompt), the platform runs the reconstruction on a cloud GPU, and it returns a textured 3D mesh. You preview the model in the browser (drag to rotate, scroll to zoom) and download it as a GLB file. Both tools, Image-to-3D and Text-to-3D, produce the same GLB output and cost the same 12 credits per task.