Repository navigation
Releases: h4nz4/segmenta
Release list
v0.1.4
🚀 v0.1.4
Added
Added --concat-copy-mp4 mode to skip HEVC reencoding and concatenate timestamped MP4 segments directly into a single output file.
Preserved existing chronological grouping behavior, so MP4 segments are still merged in parsed timestamp order.
Changed
Updated the CLI flow to bypass HEVC encoder detection and selection when concat-copy mode is used.
Kept output naming, folder organization, preview generation, and source-file cleanup behavior consistent with the standard workflow.
Validation
Added strict input validation for concat-copy mode: each file in a group must be .mp4.
Improved failure reporting so ffmpeg stream-copy errors are surfaced clearly when input MP4 streams are incompatible.
v0.1.3
v0.1.2
Release Notes: Segmenta v0.1.2
🖼️ Integrated Video Previews (Thumbnailer)
Automatic Generation: Segmenta now automatically generates a high-quality .jpg preview sheet for every session after successful transcoding.
Optimized Grid: Defaults to a tall 3 x 9 grid with fixed 400px tile widths, designed for efficient scanning of long recordings.
Technical Headers: Previews include a comprehensive technical header containing file size, duration, exact resolution, bitrate, and audio channel metadata.
Timestamp Overlays: Each tile includes a precise white timestamp overlay with a black shadow for high visibility against any content.
Industry-Standard Performance:
Multithreaded Decoding: Enabled AUTO thread detection to utilize all CPU cores for HEVC decoding.
Fast Keyframe Sampling: Employs industry-standard "fast" sampling (nearest keyframe) for near-instant preview generation.
FFmpeg Scaling: Utilizes FFmpeg filter graphs (scale) to resize frames during the decoding process, significantly reducing memory and CPU overhead compared to Python-level resizing.
⚙️ CLI Logic & Workflow Refactor
Non-Interactive Deletion: Removed the interactive "Delete original processed .ts files?" prompt to streamline automated workflows.
New CLI Flags:
--delete-original: Explicitly opt-in to deleting source .ts segments after a successful run.
--no-thumbnail: Disable preview sheet generation if needed.
Reliability: The --keep-ts flag now takes absolute precedence, ensuring source files are preserved even if --delete-original is toggled.
📊 Enhanced Feedback
Dual Progress Tracking: Added a dedicated real-time ASCII progress bar for the thumbnail generation phase, separate from the main transcoding progress.
Expanded Summaries: The final run summary now explicitly lists the paths for both the output .mp4 and the companion _preview.jpg files.
🐛 Bug Fixes
Fixed a critical crash caused by module 'av' has no attribute 'AVError' appearing in certain environments.
Fixed a logic bug where preview sheets would sometimes contain 27 identical frames due to improper decoder buffer management.
📦 New Dependencies
av: High-performance video decoding.
pymediainfo: Comprehensive media metadata extraction.
Pillow: Image composition and text rendering.