Using Asset Tap

Complete guide to the Asset Tap GUI -- generating 3D models, using the viewer, managing your library, and configuring settings.

This guide walks through the Asset Tap GUI from first launch to exporting your 3D models.

First Launch

When you open Asset Tap for the first time, the welcome screen asks for your API key. Enter a key from any supported provider -- a fal.ai API key, a Meshy API key, or a Tripo API key -- and click Save. You only need one to run the full pipeline. You can add the others later (or swap between them) in Settings.

Main Window

The main window has two areas: the sidebar on the left for inputs and controls, and the viewer on the right for previewing results.

The sidebar is where you configure and launch generations:

  • Prompt -- Type a text description of the 3D model you want (e.g., "a cowboy ninja with a leather duster, bandana mask, and dual katanas on the back")
  • Template -- Optionally select a prompt template to structure your input
  • Provider -- Select the AI provider to use
  • Image Model -- Choose which model generates the image from your text
  • Image Model Settings -- Per-model parameters declared by the provider (aspect ratio, resolution, seed, and so on), shown as sliders, checkboxes, and dropdowns. Values persist per provider and model.
  • 3D Model -- Choose which model converts the image to 3D
  • 3D Model Settings -- The same, for the image-to-3D model (topology, polycount, PBR, texture resolution, and so on)
  • Image only (skip 3D) -- Stop after the image; no GLB is written
  • Generate -- Start the generation pipeline

Using an Existing Image

If you already have an image, you can skip the text-to-image step. Use the image input in the sidebar to load a file directly -- Asset Tap will send it straight to the image-to-3D model.

Generating a Model

Click Generate to start the pipeline. Asset Tap runs in stages:

  1. Image Generation -- Your text prompt is sent to the AI provider, which returns an image
  2. Image Approval -- Review the generated image before proceeding to 3D conversion
  3. 3D Generation -- The image is converted to a 3D model (GLB format)

Image Approval

After the image is generated, Asset Tap shows you a preview so you can decide whether to proceed. If the image doesn't match what you had in mind, you can go back and adjust your prompt.

3D Viewer

Once generation completes, the 3D model loads in the built-in viewer. You can interact with the model directly:

Controls follow Blender's conventions.

Mouse:

  • Rotate -- Left-drag, or middle-click drag, to orbit around the model
  • Pan -- Shift + left-drag, or Shift + middle-click drag
  • Zoom -- Scroll wheel (no modifier)

Trackpad:

  • Rotate -- Two-finger scroll to orbit
  • Pan -- Shift + two-finger scroll
  • Zoom -- Pinch, or Ctrl/Cmd + two-finger scroll (a mouse wheel zooms with no modifier)

The 3D tab is an inspector (Grid, Axes, Reset View). A Reset View button in the viewer toolbar restores the default camera.

Animation (experimental)

Animate opens an optional panel beside the viewer: Rig, then Bind, then the clip list. This path is experimental. Fingers are not weighted, and the panel may change.

Download Universal Animation Libraries (Help, or Animate → Add pack...) fetches the Standard libraries from the latest release (hash-verified). Already-installed packs are left alone, so a Source upgrade is not overwritten.

On an unbound mesh the first step is Rig: pose the shipped skeleton on a frozen mesh. Dragging a joint never deforms the model. Auto-fit is a button, not something that runs on open. Bind skins the mesh; a joint off the body is refused by name. After a bind, click a clip to play it (preview never writes) and Bake the ticked set into model.glb.

The full loop, CLI and MCP included: Animation (experimental).

The viewer supports models from all providers and handles vertex colors, textures, and node transforms automatically.

Library

The Library view lets you browse all previously generated models. Each entry shows the prompt, timestamp, and a quick preview. Click any entry to load it in the 3D viewer.

Bundles are loaded from your configured output directory. See Bundle Structure for details on the output format.

Importing Bundles

Bring assets into your library from anywhere -- a CLI run that used a custom output directory, a bundle someone shared, an exported archive, or a lone .glb / image:

  • Drag & drop (zones, not the whole window):
    • Sidebar “Drop image here” — pipeline input (skip text-to-image).
    • Bundle Info — always a new library bundle (.glb, image, zip, folder, bundle.json). Drop a still and a mesh together to pair them.
    • Empty Image or 3D tab on an open bundle — attach the missing image.png / model.glb. The Add … to this bundle picker does the same. File → Import always creates a new bundle.
    • Animation panel — clip packs only. A UAL zip dropped on the preview is not wrapped as model.glb.
    • Dropped somewhere that isn't a zone — the menu bar, the progress pane — and nothing is imported; a toast says where it belongs.
    • On Linux, zones are unavailable: the window system reports no cursor position while files are being dragged, so a drop is routed by what it is instead of where it landed — a still becomes the pipeline input, a pack installs, and anything else becomes a new bundle. Use the File menu and the Add … to this bundle pickers to reach the other targets.
  • File → Import Bundle... -- pick a .zip, a .glb, an image, or navigate into a bundle folder and double-click its bundle.json.
  • File → Import Bundle Folder... -- single-click the folder, then Open (double-clicking navigates into it -- that's the OS file picker, not us).
  • File → Install Animation Pack... -- a UAL .zip / .glb, or Install Animation Pack Folder... for an extracted download. Same as Animate → Add pack, or dropping the file on the Animation panel.

A loose .glb or image is copied into a new timestamped bundle with the standard filenames (model.glb, image.png) and a bundle.json. Imports leave the source untouched. Zips made with macOS's built-in Compress work fine.

Settings

Open Settings from the gear icon to configure:

  • API Keys -- Add or update provider API keys
  • Output Directory -- Choose where generated models are saved

Templates

Asset Tap includes prompt templates that help structure your text input for better results. Select a template from the sidebar dropdown, fill in the variables, and the template generates an optimized prompt.

You can browse available templates with the template selector in the sidebar.

What's Next