Changelog

5.2 — Export Progress & UX Improvements

Real-time export progress — progress bar now reads out_time= from ffmpeg's -progress file, showing actual encoding progress (works at any loglevel, even error or quiet)
Configurable ffmpeg loglevel — dropdown selector in Export panel lets you choose between info, error, warning, verbose, debug to control log verbosity
Auto-load conv.isw — if conv.isw exists in the startup directory, it is automatically loaded on launch (useful for --pipe mode workflows)
Cleanup temporary files checkbox — toggle to auto-delete temp files (.bat, .isw, .log, _progress.txt) after export completes, enabled by default

-progress file — ffmpeg now writes machine-readable progress to temp/img2spec_export_progress.txt independently of loglevel
Progress fallback — parses both HH:MM:SS.xxxxxx and numeric (seconds) formats from -progress output

Export first-frame freeze — removed -fflags nobuffer from decoder ffmpeg, which caused frames to be output in decode order (instead of presentation order), leading to 150 duplicate/skipped frames (~2.5s) due to H.264 B-frame reordering
Export framerate consistency — -framerate now used for rawvideo input in encoder ffmpeg (replaces -r), matching rawvideo demuxer semantics
Export decoder output — removed redundant -r and -vsync options from decoder ffmpeg, letting it output at native source framerate without vsync interference

All first-frame diagnostic code (prev_buf, memcmp, dup_warn) removed from pipe_loop() after issue resolution

Video Mode — load, scrub, and export video files (MP4, MOV, AVI, etc. supported by ffmpeg)
- Timeline slider with frame-by-frame navigation
- Play/pause with forward/backward skip buttons
- Real-time preview of all modifiers applied to video frames
- Export to HEVC (NVENC/AMF) or H.264 (x264) with configurable quality and scale
- Audio remux from source video after export
- Default export filename: <input>_spmz.mp4
--pipe Mode — process frames via stdin/stdout pipeline
- img2spec workspace.isw --pipe --width W --height H
- Reads raw RGB24 frames from stdin, outputs processed RGBA frames to stdout
- Supports ScalePosModifier (full-resolution source preserved)
- Frame-accurate pipeline for integration with external tools
--batch-stdin Mode — batch process multiple images with different workspaces
- Reads JSON lines from stdin: {"src":"input.png","workspace":"file.isw","dst":"output.png"}
- Each line is a self-contained export job
- Avoids Windows 32K-char command-line limit via streaming input
Headless CLI Mode — --headless flag suppresses all GUI output
- Combined with --batch-stdin for fully automated batch processing
- Can run on machines without a display

Original release by Jari Komppa. Image-to-spectrum conversion GUI tool with: