PiAPI Trellis Image to 3D Model icon

PiAPI Trellis Image to 3D Model

Generate 3D models from images using Microsoft Trellis via PiAPI

Overview

This node generates 3D models from input images using the Microsoft Trellis service via PiAPI. It accepts an image in one of three formats—URL, Base64 string, or binary data—and sends it to the API for conversion into a 3D model representation. The node can either return immediately with a task ID or wait until the 3D model generation completes before returning the result.

Common scenarios:

  • Creating 3D assets from 2D images for use in games, VR/AR applications, or digital art.
  • Automating 3D model creation workflows by integrating image-to-3D conversion into larger automation pipelines.
  • Experimenting with AI-driven 3D reconstruction from photos or sketches.

Practical example:

  • Input an image URL of a product photo and receive a 3D model that can be used for interactive web display.
  • Use binary image data from a previous node (e.g., file read or image processing) to generate a 3D model without needing to upload the image separately.

Properties

Name Meaning
Image Input Method How the image is provided: URL (image accessible via URL), Base64 (base64 encoded image data), or Binary Data (binary data from a previous node).
Image URL URL of the image to convert to 3D. Required if "Image Input Method" is URL. Max size 1024x1024.
Base64 Image Base64 encoded image data. Required if "Image Input Method" is Base64. Max size 1024x1024.
Binary Property Name of the binary property containing the image to convert. Required if "Image Input Method" is Binary Data.
Seed Random seed number for reproducible results. Use 0 for random seed.
Advanced Options Collection of advanced parameters controlling quality and speed:
• Surface Sampling Steps (10-50): Number of sampling steps for surface creation.
• SLAT Sampling Steps (10-50): Number of sampling steps for SLAT process.
• Surface Guidance Strength (0.1-10): Guidance strength for surface creation.
• SLAT Guidance Strength (0.1-10): Guidance strength for SLAT process.
Wait for Completion Boolean flag indicating whether to wait for the 3D model generation to complete before returning. If false, returns immediately with task info; if true, waits for completion and returns final result.
Note Informational notice that 3D model generation can take several minutes and suggests using a Task Status node to check progress if not waiting for completion.

Output

The node outputs JSON data containing information about the 3D model generation task:

  • When not waiting for completion, the output includes a task_id and initial task data returned by the API.
  • When waiting for completion, the output contains the completed task data including the generated 3D model details.

The exact structure depends on the API response but generally includes:

  • task_id: Identifier for the generation task.
  • Status and metadata about the 3D model generation.
  • Result data when available (e.g., links or references to the generated 3D model).

No binary data is output directly by this node; all data is returned as JSON.

Dependencies

  • Requires an API key credential for accessing the PiAPI service that provides Microsoft Trellis 3D model generation.
  • The node makes HTTP requests to the PiAPI endpoint /api/v1/task.
  • For image input via URL, the node downloads the image itself, so the URL must be directly accessible without authentication or blocking.
  • Uses internal helper functions for HTTP requests and binary data handling.

Troubleshooting

  • Failed to download image from URL: This error occurs if the node cannot access the image URL directly. Some websites block direct downloads. Solution: Download the image manually or use the Binary Data input method instead.
  • No binary data found in property: Happens if the specified binary property does not exist or is empty in the input data. Verify the binary property name and ensure the previous node outputs the expected binary data.
  • API Error: Returned if the PiAPI service responds with an error code. Check your API credentials, network connectivity, and input parameters.
  • Timeouts or long delays: 3D model generation can take several minutes. If not waiting for completion, use a separate Task Status node to poll for results.

Links and References

Discussion