metadata
title: CompI — Final Dashboard
emoji: 🎨
colorFrom: indigo
colorTo: purple
sdk: streamlit
app_file: src/ui/compi_phase3_final_dashboard.py
pinned: false
CompI - Compositional Intelligence Project
A multi-modal AI system that generates creative content by combining text, images, audio, and emotional context.
Note: All documentation has been consolidated under docs/. See docs/README.md for an index of guides.
🚀 Project Overview
CompI (Compositional Intelligence) is designed to create rich, contextually-aware content by:
- Processing text prompts with emotional analysis
- Generating images using Stable Diffusion
- Creating audio compositions
- Combining multiple modalities for enhanced creative output
📁 Project Structure
Project CompI/
├── src/ # Source code
│ ├── generators/ # Image generation modules
│ ├── models/ # Model implementations
│ ├── utils/ # Utility functions
│ ├── data/ # Data processing
│ ├── ui/ # User interface components
│ └── setup_env.py # Environment setup script
├── notebooks/ # Jupyter notebooks for experimentation
├── data/ # Dataset storage
├── outputs/ # Generated content
├── tests/ # Unit tests
├── run_*.py # Convenience scripts for generators
├── requirements.txt # Python dependencies
└── README.md # This file
🛠️ Setup Instructions
1. Create Virtual Environment
# Using conda (recommended for ML projects)
conda create -n compi-env python=3.10 -y
conda activate compi-env
# OR using venv
python -m venv compi-env
# Windows
compi-env\Scripts\activate
# Linux/Mac
source compi-env/bin/activate
2. Install Dependencies
For GPU users (recommended for faster generation):
# First, check your CUDA version
nvidia-smi
# Install PyTorch with CUDA support first (replace cu121 with your CUDA version)
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
# Then install remaining requirements
pip install -r requirements.txt
For CPU-only users:
pip install -r requirements.txt
3. Test Installation
python src/test_setup.py
🚀 Quick Start
Phase 1: Text-to-Image Generation
# Basic text-to-image generation
python run_basic_generation.py "A magical forest, digital art"
# Advanced generation with style conditioning
python run_advanced_styling.py "dragon in a crystal cave" --style "oil painting" --mood "dramatic"
# Interactive style selection
python run_styled_generation.py
# Quality evaluation and analysis
python run_evaluation.py
# Personal style training with LoRA
python run_lora_training.py --dataset-dir datasets/my_style
# Generate with personal style
python run_style_generation.py --lora-path lora_models/my_style/checkpoint-1000 "artwork in my_style"
Phase 2.A: Audio-to-Image Generation 🎵
# Install audio processing dependencies
pip install openai-whisper
# Streamlit UI (Recommended)
streamlit run src/ui/compi_phase2a_streamlit_ui.py
# Command line generation
python run_phase2a_audio_to_image.py --prompt "mystical forest" --audio "music.mp3"
# Interactive mode
python run_phase2a_audio_to_image.py --interactive
# Test installation
python src/test_phase2a.py
# Run examples
python examples/phase2a_audio_examples.py --example all
Phase 2.B: Data/Logic-to-Image Generation 📊
# Streamlit UI (Recommended)
streamlit run src/ui/compi_phase2b_streamlit_ui.py
# Command line generation with CSV data
python run_phase2b_data_to_image.py --prompt "data visualization" --csv "data.csv"
# Mathematical formula generation
python run_phase2b_data_to_image.py --prompt "mathematical harmony" --formula "np.sin(np.linspace(0, 4*np.pi, 100))"
# Batch processing
python run_phase2b_data_to_image.py --batch-csv "data_folder/" --prompt "scientific patterns"
# Interactive mode
python run_phase2b_data_to_image.py --interactive
Phase 2.C: Emotional/Contextual Input to Image Generation 🌀
# Streamlit UI (Recommended)
streamlit run src/ui/compi_phase2c_streamlit_ui.py
# Command line generation with preset emotion
python run_phase2c_emotion_to_image.py --prompt "mystical forest" --emotion "mysterious"
# Custom emotion generation
python run_phase2c_emotion_to_image.py --prompt "urban landscape" --emotion "🤩" --type custom
# Descriptive emotion generation
python run_phase2c_emotion_to_image.py --prompt "mountain vista" --emotion "I feel a sense of wonder" --type text
# Batch emotion processing
python run_phase2c_emotion_to_image.py --batch-emotions "joyful,sad,mysterious" --prompt "abstract art"
# Interactive mode
python run_phase2c_emotion_to_image.py --interactive
Phase 2.D: Real-Time Data Feeds to Image Generation 🌎
# Streamlit UI (Recommended)
streamlit run src/ui/compi_phase2d_streamlit_ui.py
# Command line generation with weather data
python run_phase2d_realtime_to_image.py --prompt "cityscape" --weather --city "Tokyo"
# News-driven generation
python run_phase2d_realtime_to_image.py --prompt "abstract art" --news --category "technology"
# Multi-source generation
python run_phase2d_realtime_to_image.py --prompt "world state" --weather --news --financial
# Temporal series generation
python run_phase2d_realtime_to_image.py --prompt "evolving world" --weather --temporal "0,30,60"
# Interactive mode
python run_phase2d_realtime_to_image.py --interactive
Phase 2.E: Style Reference/Example Image to AI Art 🖼️
# Streamlit UI (Recommended)
streamlit run src/ui/compi_phase2e_streamlit_ui.py
# Command line generation with reference image
python run_phase2e_refimg_to_image.py --prompt "magical forest" --reference "path/to/image.jpg" --strength 0.6
# Web URL reference
python run_phase2e_refimg_to_image.py --prompt "cyberpunk city" --reference "https://example.com/artwork.jpg"
# Batch generation with multiple variations
python run_phase2e_refimg_to_image.py --prompt "fantasy landscape" --reference "image.png" --num-images 3
# Style analysis only
python run_phase2e_refimg_to_image.py --analyze-only --reference "artwork.jpg"
# Interactive mode
python run_phase2e_refimg_to_image.py --interactive
🧪 NEW: Ultimate Multimodal Dashboard (True Fusion) 🚀
Revolutionary upgrade with REAL processing of each input type!
# Launch the upgraded dashboard with true multimodal fusion
python run_ultimate_multimodal_dashboard.py
# Or run directly
streamlit run src/ui/compi_ultimate_multimodal_dashboard.py --server.port 8503
Key Improvements:
- ✅ Real Audio Analysis: Whisper transcription + librosa features
- ✅ Actual Data Processing: CSV analysis + formula evaluation
- ✅ True Emotion Analysis: TextBlob sentiment classification
- ✅ Live Real-time Data: Weather/news API integration
- ✅ Advanced References: img2img + ControlNet processing
- ✅ Intelligent Fusion: Actual content processing (not static keywords)
Access at: http://localhost:8503
See: ULTIMATE_MULTIMODAL_DASHBOARD_README.md for detailed documentation.
🖼️ NEW: Phase 3.C Advanced Reference Integration 🚀
Professional multi-reference control with hybrid generation modes!
Key Features:
- ✅ Role-Based Reference Assignment: Select images for style vs structure
- ✅ Live ControlNet Previews: Real-time Canny/Depth preprocessing
- ✅ Hybrid Generation Modes: CN + IMG2IMG simultaneous processing
- ✅ Professional Controls: Independent strength tuning for style/structure
- ✅ Seamless Integration: Works with all CompI multimodal phases
See: PHASE3C_ADVANCED_REFERENCE_INTEGRATION.md for complete documentation.
🗂️ NEW: Phase 3.D Professional Workflow Manager 🚀
Complete creative workflow platform with unified logging, presets, and export bundles!
Key Features:
- ✅ Unified Run Logging: Auto-ingests from all CompI phases
- ✅ Professional Gallery: Advanced filtering and search
- ✅ Preset System: Save/load complete generation configs
- ✅ Export Bundles: ZIP packages with metadata and reproducibility
- ✅ Annotation System: Ratings, tags, and notes for workflow management
Launch: python run_phase3d_workflow_manager.py | Access: http://localhost:8504
See: docs/PHASE3D_WORKFLOW_MANAGER_GUIDE.md for complete documentation.
⚙️ NEW: Phase 3.E Performance, Model Management & Reliability 🚀
Production-grade performance optimization, model switching, and intelligent reliability!
Key Features:
- ✅ Model Manager: Dynamic SD 1.5 ↔ SDXL switching with auto-availability checking
- ✅ LoRA Integration: Universal LoRA loading with scale control across all models
- ✅ Performance Controls: xFormers, attention slicing, VAE optimizations, precision control
- ✅ VRAM Monitoring: Real-time GPU memory usage tracking and alerts
- ✅ Reliability Engine: OOM-safe auto-retry with intelligent fallbacks
- ✅ Batch Processing: Seed-controlled batch generation with memory management
- ✅ Upscaler Integration: Optional 2x latent upscaling for enhanced quality
Launch: python run_phase3e_performance_manager.py | Access: http://localhost:8505
See: docs/PHASE3E_PERFORMANCE_GUIDE.md for complete documentation.
🧪 ULTIMATE: Phase 3 Final Dashboard - Complete Integration! 🎉
The ultimate CompI interface that integrates ALL Phase 3 components into one unified creative environment!
Complete Feature Integration:
- ✅ 🧩 Multimodal Fusion (3.A/3.B): Real audio, data, emotion, real-time processing
- ✅ 🖼️ Advanced References (3.C): Role assignment, ControlNet, live previews
- ✅ ⚙️ Performance Management (3.E): Model switching, LoRA, VRAM monitoring
- ✅ 🎛️ Intelligent Generation: Hybrid modes with automatic fallback strategies
- ✅ 🖼️ Professional Gallery (3.D): Filtering, rating, annotation system
- ✅ 💾 Preset Management (3.D): Save/load complete configurations
- ✅ 📦 Export System (3.D): Complete bundles with metadata and reproducibility
Professional Workflow:
- Configure multimodal inputs (text, audio, data, emotion, real-time)
- Upload and assign references (style vs structure roles)
- Choose model and optimize performance (SD 1.5/SDXL, LoRA, optimizations)
- Generate with intelligent fusion (automatic mode selection)
- Review and annotate results (gallery with rating/tagging)
- Save presets and export bundles (complete reproducibility)
Launch: python run_phase3_final_dashboard.py | Access: http://localhost:8506
See: docs/PHASE3_FINAL_DASHBOARD_GUIDE.md for complete documentation.
🎯 CompI Project Status: COMPLETE ✅
CompI has achieved its ultimate vision: the world's most comprehensive and production-ready multimodal AI art generation platform!
✅ All Phases Complete:
- ✅ Phase 1: Foundation (text-to-image, styling, evaluation, LoRA training)
- ✅ Phase 2: Multimodal integration (audio, data, emotion, real-time, references)
- ✅ Phase 3: Advanced features (fusion dashboard, advanced references, workflow management, performance optimization)
🚀 What CompI Offers:
- Complete Creative Platform: From generation to professional workflow management
- Production-Grade Reliability: Robust error handling and performance optimization
- Professional Tools: Industry-standard features for serious creative and commercial work
- Universal Compatibility: Works across different hardware configurations
- Extensible Foundation: Ready for future enhancements and integrations
CompI is now the ultimate multimodal AI art generation platform - ready for professional creative work! 🎨✨
🎯 Core Features
- Text Analysis: Emotion detection and sentiment analysis
- Image Generation: Stable Diffusion integration with advanced conditioning
- Audio Processing: Music and sound analysis with Whisper integration
- Data Processing: CSV analysis and mathematical formula evaluation
- Emotion Processing: Preset emotions, custom emotions, emoji, and contextual analysis
- Real-Time Integration: Live weather, news, and financial data feeds
- Style Reference: Upload/URL image guidance with AI-powered style analysis
- Multi-modal Fusion: Combining text, audio, data, emotions, real-time feeds, and visual references
- Pattern Recognition: Automatic detection of trends, correlations, and seasonality
- Poetic Interpretation: Converting data patterns and emotions into artistic language
- Color Psychology: Emotion-based color palette generation and conditioning
- Temporal Awareness: Time-sensitive data processing and evolution tracking
🔧 Tech Stack
- Deep Learning: PyTorch, Transformers, Diffusers
- Audio: librosa, soundfile
- UI: Streamlit/Gradio
- Data: pandas, numpy
- Visualization: matplotlib, seaborn
📝 Usage
Coming soon - basic usage examples and API documentation.
🤝 Contributing
This is a development project. Feel free to experiment and extend functionality.
📄 License
MIT License - see LICENSE file for details.