3D Generation with PhotoGenius

PhotoGenius offers two ways to create 3D models: text-to-3D and image-to-3D conversion.

Available 3D Generation Methods

Text-to-3D Generation

Generate 3D models directly from text descriptions using the 3D Generation Studio.

Image-to-3D Conversion

Convert existing images into 3D models using the Image to 3D Converter.

Text-to-3D Workflow

Step 1: Optimize Your Prompt

3D generation works best with clear, single-object descriptions:

Good 3D Prompts:

"Red sports car, clear background, centered, full view"
"Wooden chair, simple design, isolated object"
"Golden crown, detailed ornament, single item"

Avoid:

• Complex scenes with multiple objects
• Abstract concepts
• Cluttered backgrounds

Step 2: Generation Settings

Key Settings:

• Generate Color: Keep enabled for textured models
• Guidance Strength: 7.5 (default) works well
• Seed: Use for reproducible results

Step 3: Processing Time

3D generation involves two stages:

1. 2D Reference Generation: 10-20 seconds
2. 3D Model Creation: 2-5 minutes

Total time: 3-6 minutes per model

Image-to-3D Workflow

Best Input Images

Ideal Image Characteristics:

• Single object focus
• Clear edges and contrast
• Minimal background clutter
• Good lighting
• Multiple angles (if available)

Supported Formats:

• PNG (recommended for transparency)
• JPG/JPEG
• WebP
• Maximum 10MB file size

Conversion Process

1. Upload Image: Click or drag to upload
2. Adjust Settings:
   • Color generation (on/off)
   • Guidance strength (0-10)
3. Wait for Processing: 2-5 minutes
4. Download GLB File: Universal 3D format

3D Model Output

File Format: GLB

PhotoGenius generates GLB (GL Transmission Format Binary) files:

• Universal compatibility: Works in all 3D software
• Embedded textures: Colors included in file
• Optimized size: Smaller than traditional formats
• Web-ready: Direct browser viewing

Model Characteristics

Typical Output:

• Polygon count: 10,000-50,000 faces
• Texture resolution: 1024x1024 to 2048x2048
• File size: 1-10MB
• Color/texture: Automatic UV mapping

Optimizing for 3D Generation

Prompt Engineering for 3D

Structure for Best Results:

[Object] + [Key features] + "clear background" + 
"centered" + "full view" + "simple lighting"

Example:

"Ancient Greek vase, decorated with patterns, 
clear background, centered, full view, simple lighting"

Common 3D Generation Issues

Issue: Incomplete Models

• Solution: Add "complete object" and "full view" to prompt

Issue: Poor Geometry

• Solution: Simplify prompt, focus on single object

Issue: Missing Details

• Solution: Increase guidance strength to 8-9

Issue: Wrong Perspective

• Solution: Specify viewing angle ("front view", "three-quarter view")

3D Viewer Controls

Navigation in PhotoGenius Viewer

• Rotate: Click and drag
• Zoom: Scroll wheel or pinch
• Pan: Right-click and drag
• Auto-rotate: Enabled by default
• Reset view: Double-click

Lighting Controls

Adjust viewing conditions:

• Brightness: 0-2x multiplier
• Contrast: 0-2x multiplier
• Environment: Studio lighting preset

Sharing 3D Models

Public Gallery

Models can be shared publicly:

1. Enable "Make Public" before generation
2. Share link after completion
3. View in Public Gallery

Direct Sharing

Share individual models:

1. Click "Share Creation"
2. Link copied to clipboard
3. Recipients can view in browser

Download Options

• 2D Reference: PNG image
• 3D Model: GLB file
• Both files include generation metadata

Use Cases for 3D Models

E-commerce

• Product visualization
• 360° product views
• AR try-on experiences

Game Development

• Asset prototyping
• Concept visualization
• Quick placeholder models

Design & Architecture

• Concept development
• Client presentations
• Spatial planning

Education

• Visual learning aids
• Interactive demonstrations
• Historical reconstructions

Quality Optimization Tips

For Objects

Best: "Single [object], isolated, white background"
Good: "[Object] alone, centered, clear view"
Avoid: "[Object] in complex scene with many elements"

For Characters/Creatures

Best: "[Character] in T-pose, neutral expression"
Good: "[Character] standing, front view"
Avoid: "[Character] in action pose with effects"

For Architecture

Best: "[Building] exterior, isometric view"
Good: "[Structure] complete view, simple"
Avoid: "[Building] with complex surroundings"

Technical Specifications

Processing Pipeline

1. Prompt/Image Analysis: AI interprets input
2. 2D Generation: Creates reference views
3. Depth Estimation: Calculates 3D structure
4. Mesh Generation: Builds polygon model
5. Texture Mapping: Applies colors/textures
6. Optimization: Reduces file size

Output Specifications

• Format: GLB 2.0
• Coordinate System: Y-up, right-handed
• Scale: Normalized to ~2 units
• Materials: PBR (Physically Based Rendering)

Troubleshooting 3D Generation

Generation Fails

Possible Causes:

• Prompt too complex
• Image quality too low
• Object not clearly defined

Solutions:

1. Simplify prompt to single object
2. Use higher quality input image
3. Ensure clear object boundaries

Model Looks Wrong

Common Issues:

• Incorrect proportions
• Missing parts
• Wrong orientation

Fixes:

1. Add dimensional hints ("tall", "wide", "thick")
2. Specify "complete object, all parts visible"
3. Include orientation ("upright", "horizontal")

Slow Generation

Factors:

• Server load
• Model complexity
• Network speed

Tips:

• Generate during off-peak hours
• Simplify prompts for faster processing
• Check internet connection

Best Practices

Do's

✓ Focus on single objects ✓ Use clear, descriptive prompts ✓ Include "clear background" ✓ Specify viewing angle ✓ Test with simple objects first

Don'ts

✗ Complex multi-object scenes ✗ Abstract or vague descriptions ✗ Extremely detailed prompts ✗ Moving or action elements ✗ Transparent or reflective materials

Integration Workflows

With Other PhotoGenius Tools

1. FLUX → 3D: Generate image, then convert to 3D
2. 3D → Nano Banana: Screenshot 3D, edit in 2D
3. Batch Workflow: Generate multiple angles for better 3D

External Software

GLB files work with:

• Blender: Free 3D software
• Unity/Unreal: Game engines
• Three.js: Web 3D library
• AR Quick Look: iOS AR viewer
• Model Viewer: Android 3D viewer

Credits and Pricing

Credit Usage

• Text-to-3D: 50 credits per generation
• Image-to-3D: 40 credits per conversion
• Public sharing: Free
• Downloads: Unlimited for your generations

Optimization Tips

1. Test prompts with 2D generation first (cheaper)
2. Use Image-to-3D when you have good reference
3. Batch similar objects for consistency
4. Save successful prompts for reuse

3D generation typically takes 3-6 minutes. Best results come from clear, single-object prompts with simple backgrounds.