Spaces:
Sleeping
Sleeping
Commit History
-
🎭 Add complete OmniAvatar-14B integration for avatar video generation
bravedims✨ Features: - Full OmniAvatar-14B engine integration with adaptive body animation - Audio-driven lip-sync and character animation - Multi-modal input support (text prompts, audio, reference images) - Performance optimization for various GPU configurations - Cross-platform setup scripts for model downloading 📦 New Files: - omniavatar_engine.py: Core avatar generation engine - setup_omniavatar.py/.ps1: Automated model download scripts - configs/inference.yaml: Optimized inference configuration - scripts/inference.py: Enhanced inference script - examples/infer_samples.txt: Sample input formats - OMNIAVATAR_README.md: Comprehensive documentation - OMNIAVATAR_INTEGRATION_SUMMARY.md: Implementation overview 🔧 Updates: - requirements.txt: Added OmniAvatar-compatible dependencies - Enhanced error handling and fallback systems - Hardware-specific performance optimizations 🚀 Usage: 1. Run setup: python setup_omniavatar.py 2. Start app: python app.py 3. Generate avatars via Gradio UI or API…e7ffb7d -
Make video duration dynamic based on text-to-speech length
bravedims🎯 Feature: Video duration now matches text/audio length instead of fixed 5 seconds ✅ Improvements: - Calculate duration based on word count (2.5 words/second average speaking rate) - Try to get actual audio file duration using librosa if available - Dynamic video generation with appropriate length - Add text animation showing speech content progressively - Show duration info and frame counter in video - Set reasonable bounds: minimum 3s, maximum 30s 🎬 Enhanced Video Features: - Color-changing animated circle - Progressive text display (typing effect) - Text wrapping for long content (max 3 lines) - Frame counter and time display - Better visual feedback 📝 Logic: - If audio file exists: Use actual audio duration - If no audio: Estimate from text (word_count / 2.5 words per second) - Add 20% padding for natural speech patterns - Fallback to 5 seconds if no text provided 🔧 Technical: - Update input file format: prompt@@image@@audio@@text_content - Pass text_to_speech content to in…efb1c49 -
Fix inference script NameError and add placeholder video generation
bravedims🐛 Issue: NameError: name 'logger' is not defined in inference.py line 69 ❌ Error: Logger was being used but not properly initialized 🔧 Fixes: - Add proper logging.basicConfig() and logger initialization - Fix get_device() function to use initialized logger - Add comprehensive main() function with error handling - Implement placeholder video generation using OpenCV - Add proper sample processing and output handling ✅ Features: - Creates animated placeholder videos (moving circle + text) - Proper device auto-detection (CPU/GPU) - Error handling for each sample - Detailed logging throughout the process - MP4 output with configurable FPS and duration 📝 Current Behavior: - Generates 5-second placeholder videos instead of actual avatars - Shows 'Avatar Placeholder Frame X' with animated circle - Ensures the full pipeline works end-to-end - Future: Will be replaced with actual OmniAvatar model inference7b4fc5d -
Fix device configuration and hardware requirements
bravedims- Update README.md to request a10g-small GPU hardware instead of t4-medium - Fix inference.yaml to use auto device detection instead of hardcoded cuda - Disable xformers and flash_attention for CPU compatibility - Add device auto-detection to inference script - This should fix the CPU/GPU mismatch causing generation failurese29fad2 -
Deploy OmniAvatar-14B with ElevenLabs TTS integration to Hugging Face Spaces
bravedimsbd1f2b1