Scenario Guide: Full 3D Character Pipeline

This guide demonstrates how to chain V2Fun's atomic APIs together to build an automated production pipeline — from a text description to an animated 3D model.

Core Workflow

1

Text / Concept Art Generation

→

2

3D Mesh Reconstruction

→

3

PBR Texture Mapping

→

4

Automated Rigging

→

5

Animation Retargeting

This pipeline gives developers a complete closed-loop workflow: input a description, output an animated 3D model.

---

Step-by-Step Guide

Task Chaining Logic

Throughout the pipeline, the task_uuid or file path (URI) from the result of each step is passed as the input_xxx parameter of the next step.

Step 1: Generate Character Concept Art (Text-to-Image)

First, use AIGC to generate a high-quality character reference image. It is recommended to include three-view or T-Pose descriptions in the prompt.

• API Endpoint: POST /images/generations
• Key Parameters:
  • prompt: "A futuristic cyber-ninja, full body, t-pose, front view, game asset style"
  • size: "1024x1024"
• Output: Retrieve the generated image path from result.

---

Step 2: Generate 3D Mesh (Image-to-Model)

Use the image from Step 1 to perform high-precision 3D reconstruction.

• API Endpoint: POST /3d_models/meshes
• Input: Pass the image path from the previous step as input_image.
• Tip: You can enable options.block: true to wait synchronously for the result.

---

Step 3: Generate PBR Textures (Retexturing)

Generate high-quality material textures for the reconstructed geometry.

• API Endpoint: POST /3d_models/textures
• Input:
  • input_model: The model path from the previous step.
  • prompt: Describe the material details, e.g. "carbon fiber armor with glowing blue lights".
• Core Logic: This step automatically generates BaseColor, Normal, and Roughness maps and returns a textured model.

---

Step 4: Skeleton Rigging

Give the static model the ability to move by providing joint coordinates for automatic skinning.

• API Endpoint: POST /motions/riggings
• Input:
  • input_model: The textured model path.
  • joints: 3D coordinates for 14 keypoints — head, shoulders, elbows, wrists, hips, knees, and ankles.
• Key Note: Joint coordinates must be normalized relative to the model's bounding box.

---

Step 5: Apply Animation (Retargeting)

Apply animations from the standard motion library to the newly rigged model.

• API Endpoint: POST /motions/animations
• Parameter Chaining:
  • input_model: The rigged model path from Step 4.
  • input_motion: A motion file path from the motion library (/motions/retrievals) or motion detection (/videos/motion_detections).
• Final Output: An exported .glb file ready to import into Unity or Unreal Engine.

---

Core API Reference

For a deeper understanding of any specific step, jump directly to the corresponding API reference page:

Module	Description	Docs
Image Generation	Multi-style character art, three-view and material reference image generation	Details
3D Model Generation	Automated geometry generation with support for multiple topology formats and LOD levels	Details
Motion & Animation	Intelligent skinning, skeleton format conversion, and multi-source animation retargeting	Details

---