Skip to content

sylphlin/multicam-video-preprocessing

v1.7.0MIT

Universal multi-camera video processing pipeline and AI editing suite for 2 to 6 cameras. Features MFCC acoustic time alignment with subframe refinement (<0.125ms), EBU R128 loudness normalization (-14 LUFS), multi-in-one compact grid composition (10 fps, 1.2 Mbps, 1s GOP), Gemini 3.8 Flash multimodal rough-cut with silence-aware smart segmentation (30-40 min windows) and auto-cleanup, deterministic 8-check EDL semantic validation, FCP7 XML timeline export, and single-pass hardware video rendering.

GitHubOpen Website
multicam-video-preprocessing

Universal multi-camera video preprocessing and AI editing suite for 2 to 6 camera setups. Executes MFCC acoustic time alignment with subframe refinement (<0.125ms), EBU R128 two-pass linear loudness normalization (-14 LUFS), synchronized full-length camera master exporting (frame-accurate hardware re-encoding), Multi-in-One compact grid composition (canvas <= 1920x1080, min >= 640x480/CAM), zero-split Gemini 3.8 Flash Agentic Video EDL generation (100% Google Cloud Vertex AI via ADC + GCS storage with 2-day lifecycle auto-cleanup), FCP7 XML timeline export with NTSC fractional fps & drop-frame support (Primary), and direct video rendering (Secondary). Keywords: multicam, multi-camera, dual-cam, 4-cam, 6-cam, time alignment, audio sync, loudness normalization, video preprocessing, multicam pipeline, multi-in-one, token optimization, fcp7 xml, agentic video, vertex ai, gcs, adc.