Open-source AI podcast clipper. Generate vertical clips with face tracking and burned-in captions. CLI, MCP server, web app.
Python ai-toolsauto-captionscaptionscliffmpegmcpmcp-serveropusclip-alternativepodcast-clipspodcastsshortform-videovertical-videowhisper
fix: use face_map.is_split_screen from full video, not per-segment detection
Nika Siradze911ad16041c0
diff --git a/backend/services/video_processor.py b/backend/services/video_processor.pyindex 46a68ed..e4c586f 100644--- a/backend/services/video_processor.py+++ b/backend/services/video_processor.py@@ -191,8 +191,12 @@ def crop_to_vertical(if strategy in ("face", "speaker"):# Split-screen: cut each speaker segment separately with static crop per half.- # Simple, reliable, no complex FFmpeg expressions.- is_split = _detect_split_screen(input_path, width, height)+ # Check face_map first (computed from full video), fall back to segment detection.+ is_split = False+ if face_map and face_map.get("is_split_screen"):+ is_split = True+ else:+ is_split = _detect_split_screen(input_path, width, height)if is_split and transcript_words:result = _crop_split_screen(