Trellis

Trellis 3D is an AI 3D model generator that creates downloadable 3D files from either an image or a natural-language prompt. Upload a JPG, PNG, or WebP (max 10MB) or switch to text mode, then generate a model and preview it directly in an interactive 3D viewer.

The whole workflow runs in the cloud and in your browser, so you don’t need to set up GPU hardware or install 3D software. Generation typically completes in a few minutes depending on the input and chosen settings.

Choose from multiple AI models based on the kind of geometry you need. Microsoft’s Trellis 2 is geared toward complex topology, open surfaces, and non-manifold shapes; Hunyuan 3D focuses on hard-surface precision; and Seed 3D uses multi-view reconstruction for fast results on organic and stylized forms. When you’re ready, export your model in standard formats such as GLB, OBJ, FBX, USDZ, or a rotating GIF preview.

What can I generate with Trellis 3D?

You can generate 3D models from a photo (image-to-3D) or from a text description (text-to-3D), then preview and download the result.

Do I need a GPU or to install anything?

No. Trellis 3D runs generation in the cloud and provides a browser preview, so there’s no local GPU setup and no software installation required.

What kinds of files can I upload for image-to-3D?

Upload JPG, PNG, or WebP images. The upload limit is 10MB per image, and you can choose output resolution settings during generation.

Which AI models are available in Trellis 3D?

You can select among Microsoft Trellis 2, Hunyuan 3D, and Seed 3D—each designed for different geometry strengths like complex/non-manifold topology, hard-surface precision, or fast organic generation.

What 3D formats can I export?

Trellis 3D supports exporting GLB, OBJ, FBX, and USDZ, plus an optional rotating GIF preview for quick sharing.

Is 3D modeling experience required to use Trellis 3D?

No. Upload an image or write a prompt, pick a model, and export—no topology, UV, or mesh cleanup knowledge is required.