rdawgemfl-rdawg_3D_pack/docs/USAGE.md

Usage Guide

Getting Started

Basic Workflow

The fundamental RDAWG 3D workflow consists of three main steps:

1. Load 3D Model - Import your 3D file
2. Transform - Apply modifications (scale, rotate, position)
3. Render - Convert to 2D image

Load 3D Model → Transform 3D Mesh → 3D to Image → Save Image

Node Reference

Core Nodes

🔷 Load 3D Model (RDAWG+Open3D)

Purpose: Load 3D models from various file formats

Inputs:

• file_path (STRING): Path to 3D file (OBJ, STL, PLY, GLTF)
• load_texture (BOOLEAN): Load texture coordinates if available
• normalize (BOOLEAN): Normalize model to unit sphere
• device (OPTIONS): Processing device (auto/cpu/cuda)
• use_open3d (BOOLEAN): Prioritize Open3D over Trimesh

Outputs:

• mesh (MESH): Loaded 3D mesh data
• info (STRING): Loading information and statistics

Tips:

• Use "auto" device for best performance
• Enable normalization for consistent scaling
• Open3D provides better loading for complex formats

Example Usage:

file_path: "C:/models/car.obj"
load_texture: true
normalize: true
device: "auto"
use_open3d: true

🔷 Create 3D Mesh (RDAWG+Open3D)

Purpose: Create 3D mesh from vertex and face tensors

Inputs:

• vertices (TENSOR): Vertex coordinates [N, 3]
• faces (TENSOR): Face indices [M, 3]
• device (OPTIONS): Processing device (auto/cpu/cuda)
• vertex_colors (TENSOR, optional): Vertex colors [N, 3/4]
• texture_uv (TENSOR, optional): UV coordinates [N, 2]
• create_open3d (BOOLEAN): Create Open3D mesh object

Outputs:

• mesh (MESH): Created 3D mesh
• info (STRING): Creation information

Tips:

• Vertices should be in XYZ format
• Faces should reference vertex indices
• Vertex colors support RGB or RGBA

🔷 Transform 3D Mesh (RDAWG)

Purpose: Apply geometric transformations to 3D meshes

Inputs:

• mesh (MESH): Input mesh to transform
• scale (FLOAT): Uniform scale factor (1.0 = no change)
• rotation_x (FLOAT): Rotation around X axis in degrees
• rotation_y (FLOAT): Rotation around Y axis in degrees
• rotation_z (FLOAT): Rotation around Z axis in degrees
• translate_x (FLOAT): Translation along X axis
• translate_y (FLOAT): Translation along Y axis
• translate_z (FLOAT): Translation along Z axis

Outputs:

• transformed_mesh (MESH): Transformed mesh

Tips:

• Transformations are applied in order: scale → rotate → translate
• Use small rotation values for precise control
• Translation units match model units

Common Transformations:

Scale up 2x: scale = 2.0
Rotate 45° around Y: rotation_y = 45.0
Move up 1 unit: translate_y = 1.0

🔷 3D to Image (RDAWG+Open3D)

Purpose: Render 3D mesh to 2D image

Inputs:

• mesh (MESH): Mesh to render
• width (INT): Output image width (64-2048)
• height (INT): Output image height (64-2048)
• background_color (STRING): Hex color for background (#RRGGBB)
• mesh_color (STRING): Hex color for mesh (#RRGGBB)
• use_open3d_render (BOOLEAN): Use Open3D advanced rendering
• camera_distance (FLOAT): Camera distance (0.5-10.0)
• elevation (FLOAT): Camera elevation angle (-90° to 90°)
• azimuth (FLOAT): Camera azimuth angle (-180° to 180°)

Outputs:

• rendered_image (IMAGE): 2D rendered image

Tips:

• Higher resolution = better quality but slower
• Open3D rendering provides better lighting and shading
• Camera controls work like a virtual camera orbit

Camera Settings:

Front view: elevation = 0, azimuth = 0
Top view: elevation = 90, azimuth = 0
Side view: elevation = 0, azimuth = 90

Workflow Examples

Basic Model Viewer

Create a simple 3D model viewer:

1. Load 3D Model

   • File path: your model file
   • Enable normalization and textures

2. Transform 3D Mesh

   • Set rotation_y to 45 for angled view
   • Adjust scale if needed

3. 3D to Image

   • Set resolution to 1024x1024
   • Use white background, black mesh
   • Set camera distance to 2.0

4. Save Image

   • Connect rendered image to SaveImage node

Product Photography Setup

Render product images from multiple angles:

1. Load 3D Model → Load your product model
2. Transform 3D Mesh → Apply product positioning
3. 3D to Image → Render with studio lighting
4. Save Image → Export high-quality images

Pro Tips:

• Use neutral background (#FFFFFF or #F0F0F0)
• Set mesh_color to "#808080" for professional look
• Try different camera angles for best presentation

Architectural Visualization

Render architectural models:

1. Load 3D Model → Load building model
2. Transform 3D Mesh → Position for best view
3. 3D to Image → Render with appropriate lighting
4. Save Image → Export presentation images

Settings for Architecture:

• Camera distance: 3.0-5.0 (show more context)
• Elevation: 15-30° (natural eye level)
• Background: "#87CEEB" (sky blue)

Advanced Techniques

Batch Processing

Process multiple models or views:

1. Create multiple Transform nodes with different rotations
2. Use multiple 3D to Image nodes
3. Combine outputs for comparison or animation

Material Variations

Create different material looks:

1. Load single model
2. Create multiple Transform nodes (same settings)
3. Use different mesh_color values in 3D to Image nodes
4. Compare material variations

Size Comparison

Scale models for size comparison:

1. Load multiple models
2. Use Transform nodes to normalize sizes
3. Render with consistent camera settings
4. Create comparison grid

Performance Optimization

Memory Management

• Large Models: Reduce mesh complexity before loading
• High Resolution: Start with lower resolution, increase as needed
• Batch Processing: Process items sequentially to avoid memory overflow

Rendering Speed

• Open3D vs Wireframe: Use wireframe for faster previews
• Resolution: Use 512x512 for testing, 1024x1024+ for final
• Device: Use CUDA for GPU acceleration when available

Quality vs Speed

Setting	Fast	High Quality
Resolution	512x512	2048x2048
Rendering	Wireframe	Open3D
Anti-aliasing	Off	On

Troubleshooting

Common Issues

Model doesn't load:

• Check file path and format
• Ensure file is not corrupted
• Try different format (OBJ, STL, PLY)

Black screen on render:

• Check mesh_color and background_color
• Verify camera distance (not too close/far)
• Ensure mesh has vertices and faces

Memory errors:

• Reduce resolution
• Simplify mesh geometry
• Use CPU instead of CUDA

Poor image quality:

• Increase resolution
• Enable Open3D rendering
• Adjust camera position

Performance Tips

Faster Loading:

• Use simple file formats (OBJ, STL)
• Disable texture loading if not needed
• Use CPU for small models

Better Rendering:

• Use Open3D for realistic lighting
• Increase resolution for final output
• Experiment with camera angles

Memory Efficiency:

• Clear unused nodes
• Use appropriate resolution
• Monitor VRAM usage

File Format Support

Supported Formats

• OBJ: Wavefront OBJ (recommended)
• STL: Stereolithography (mesh only)
• PLY: Stanford PLY (with colors)
• GLTF: GL Transmission Format

Format Recommendations

• General Use: OBJ (good compatibility)
• 3D Printing: STL (simple mesh)
• High Quality: PLY (supports colors)
• Web Ready: GLTF (compressed)

Conversion Tips

# Convert STL to OBJ if needed
import trimesh
mesh = trimesh.load('model.stl')
mesh.export('model.obj')

Integration with Other Nodes

Image Processing

• Connect to Image Resize for different resolutions
• Use Image Blend to add backgrounds
• Apply Image Filter for post-processing

Animation

• Use KSampler with varying rotation values
• Create frame sequences with Image Batch
• Export as GIF or video

AI Integration

• Connect to Upscale nodes for enhanced quality
• Use ControlNet for guided generation
• Apply Style Transfer for artistic effects

Best Practices

Workflow Organization

• Group related nodes
• Use descriptive names
• Keep workflows modular

File Management

• Use organized folder structure
• Keep original files separate
• Document custom settings

Performance Monitoring

• Watch VRAM usage with complex models
• Monitor render times
• Optimize based on your hardware

Backup and Versioning

• Save workflow versions
• Backup custom models
• Document successful settings