Multi Image To 3d

Multi-Image to 3D Blender Skill

Converts 1-4 photos of the same object (from different angles) into a textured 3D model using Meshy AI's Multi-Image to 3D API, then imports it into the active Blender scene. More angles = more accurate geometry and textures.

When to Use

Trigger this skill when:

• User provides multiple photos of the same object and wants a 3D model
• User says "multi-angle 3D", "convert these images to 3D", "make a 3D model from these photos"
• User wants a more accurate 3D model than single-image can provide
• User has front/back/side shots of a product, character, or object

Prerequisites

• Meshy API Key: Already set in the environment variable MESHY_API_KEY. Never ask for it in chat, never put it on a command line or in a file, and never look for it in memory or notes
• Blender MCP: Must be connected (blender-mcp addon running on port 9876)
• Images: 1-4 local file paths (.jpg, .jpeg, .png) or publicly accessible URLs. They are uploaded to meshy.ai, a paid third-party service: say so before starting
• Same object: All images must depict the same object from different angles

Flow

Step 1: Prepare the Images

For each local file, base64 encode it into a data URI:

import base64, json

image_paths = ["front.png", "back.png", "side.png"]  # user-provided paths
image_urls = []

for path in image_paths:
    with open(path, 'rb') as f:
        b64 = base64.b64encode(f.read()).decode()
    ext = path.rsplit('.', 1)[-1].lower()
    mime = 'image/jpeg' if ext in ('jpg', 'jpeg') else 'image/png'
    image_urls.append(f"data:{mime};base64,{b64}")

Step 2: Create Multi-Image-to-3D Task

Write payload to a JSON file (base64 strings are too large for CLI args):

payload = {
    "image_urls": image_urls,  # array of 1-4 data URIs or public URLs
    "ai_model": "meshy-6",
    "enable_pbr": True,
    "should_texture": True,
    "target_formats": ["glb"]
}

with open("<JOB_DIR>/request.json", "w") as f:
    json.dump(payload, f)

Then POST:

curl -s https://api.meshy.ai/openapi/v1/multi-image-to-3d \
  -X POST \
  -H "Authorization: Bearer ${MESHY_API_KEY}" \
  -H 'Content-Type: application/json' \
  -d @<JOB_DIR>/request.json

Response: {"result": "<task_id>"}

Step 3: Poll for Completion

curl -s https://api.meshy.ai/openapi/v1/multi-image-to-3d/<task_id> \
  -H "Authorization: Bearer ${MESHY_API_KEY}"

Poll every 15 seconds. Check status:

• PENDING — queued (check preceding_tasks for queue position)
• IN_PROGRESS — generating (check progress for %)
• SUCCEEDED — done, download from model_urls.glb
• FAILED — check task_error.message

Step 4: Download the GLB

curl -L -o "<JOB_DIR>/<filename>.glb" "<model_urls.glb>"

Step 5: Import into Blender

Use the Blender MCP execute_blender_code tool:

import bpy

# Import GLB (into the scene as it is: nothing is cleared)
bpy.ops.import_scene.gltf(filepath="<path_to_glb>")

# Frame imported object
bpy.ops.object.select_all(action='SELECT')

# Add basic lighting
import math
from mathutils import Vector, Euler
bpy.ops.object.light_add(type='AREA', location=(2, -2, 3))
key = bpy.context.active_object
key.data.energy = 500
key.data.size = 3

# Set viewport to material preview
for area in bpy.context.screen.areas:
    if area.type == 'VIEW_3D':
        for space in area.spaces:
            if space.type == 'VIEW_3D':
                space.shading.type = 'MATERIAL'

Configuration Options

Tips for Best Results

• Different angles: Front, back, left side, right side work best
• Consistent lighting: Same lighting conditions across all photos
• Clean background: Plain/white backgrounds help
• Same object: All images MUST be the same object
• 2-4 images: More angles = better accuracy, but diminishing returns past 4

Error Handling

• 400: Invalid image count (must be 1-4), bad format, unreachable URL
• 402: Insufficient Meshy credits
• 429: Rate limited — wait and retry
• FAILED status: Report task_error.message to user