> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryflowy.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# 3D Node

> Generate 3D models and inspect them in the built-in viewer.

export const DemoVideo = ({src, poster, caption, aspectRatio = "16 / 10", className = ""}) => {
  return <figure className={`not-prose my-6 w-full ${className}`}>
      <div className="overflow-hidden rounded-2xl border border-gray-950/10 bg-gray-100 dark:border-white/10 dark:bg-white/5" style={{
    aspectRatio
  }}>
        <video src={src} poster={poster} loop autoPlay muted playsInline className="h-full w-full object-cover" />
      </div>
      {caption && <figcaption className="mt-2 text-center text-sm text-gray-500 dark:text-gray-400">
          {caption}
        </figcaption>}
    </figure>;
};

The 3D node generates a textured 3D model from reference images, a text prompt, or both — and lets you inspect the result right on the canvas.

<DemoVideo src="/videos/canvas/node-3d.mp4" poster="/images/posters/canvas/node-3d.jpg" caption="A 3D node wired to a reference image — engine, quality, material, and export format set in the panel" />

## Inputs

Connect up to 4 Image nodes — treated as front, back, left, and right views of one object — and/or write a text prompt describing the model you want. A single clean product shot is enough to start; extra views tighten the geometry on the sides the model can't see.

## Engines

| Engine           | Inputs                     | Strengths                                                                |
| ---------------- | -------------------------- | ------------------------------------------------------------------------ |
| **Rodin**        | Text and/or images         | Premium quality, text-to-3D, material control (`PBR`, `Shaded`, or both) |
| **Hunyuan3D v3** | Images only (at least one) | Cheaper, multi-view, polygon and face-count control                      |

Pick Rodin when you have a text prompt or want material and quality control; pick Hunyuan3D v3 when you're working from images alone and want an economical result with mesh-level control. With Rodin and multiple images, the condition mode decides whether they're read as views of **one** object or combined as **different** objects.

## The built-in viewer

A finished model loads into an interactive viewer inside the node — drag to orbit around it and inspect the geometry and textures from every angle, no external tools needed. When it holds up, use **Download Model** to export it.

Output formats: `GLB`, `USDZ`, `FBX`, `OBJ`, `STL`.

## Where outputs go

3D models are an endpoint on the graph: they don't feed other generation nodes. The output is the downloadable file — drop it into a game engine, an AR pipeline, or a DCC tool of your choice. To get more references *into* a 3D node, generate views with the [Change Angle tool](/canvas/image-tools/change-angle) first.