Zen LM
Models

3D

Generate 3D assets from images and text prompts with Zen 3D.

Overview

Zen 3D transforms images and text prompts into high-quality 3D assets. Built on the Zen MoDE architecture, it provides a unified interface for converting 2D inputs into interactive 3D representations suitable for games, visualization, and design workflows.

Models

ModelParamsContextHFPaper
Zen 3Dweightspaper

Quick start

Using Transformers

from transformers import AutoImageProcessor, AutoModel
import torch
from PIL import Image

# Load model and processor
model_id = "zenlm/zen-3d"
processor = AutoImageProcessor.from_pretrained(model_id)
model = AutoModel.from_pretrained(model_id)

# Load an image
image = Image.open("path/to/image.jpg")

# Process and generate 3D representation
inputs = processor(images=image, return_tensors="pt")

with torch.no_grad():
    outputs = model(**inputs)

# outputs contain 3D mesh or point cloud representation

Using the Zen API

For production deployments, use the OpenAI-compatible Zen API:

import requests

headers = {
    "Authorization": f"Bearer {HANZO_API_KEY}",
    "Content-Type": "application/json"
}

response = requests.post(
    "https://api.hanzo.ai/v1/generations/3d",
    headers=headers,
    json={
        "model": "zen-3d",
        "image_url": "https://example.com/image.jpg",
        "prompt": "high-quality game asset"
    }
)

# Returns 3D mesh in standard format (glTF, USD, etc.)
result = response.json()

Get your free API key at console.hanzo.ai — $5 free credit on signup.

Key capabilities

  • Image-to-3D: Convert photorealistic images into 3D models
  • Text-to-3D: Generate 3D assets from natural language descriptions
  • Multi-format support: Export to glTF, USD, OBJ, and custom formats
  • Interactive meshes: Optimized geometry for real-time rendering
  • Batch processing: Efficiently handle multiple conversions

Use cases

  • Game asset generation and prototyping
  • 3D visualization and design tools
  • Product photography to interactive models
  • Avatar and character generation
  • Architectural visualization

On this page