Added optional Whisper installation to install.sh:
- New --skip-whisper flag to disable Whisper installation
- Interactive prompt asking if user wants Whisper for subtitling
- Automatic installation of openai-whisper via pip3
- PATH configuration hints for ~/.local/bin
- Python 3 and pip3 dependency checks
Whisper enables automated subtitle generation from audio using
OpenAI's speech recognition model.
The banner was displaying every time a new shell was opened,
which was intrusive. Now the aliases load silently.
Commands are still available (VideoTools, VideoToolsRebuild,
VideoToolsClean) but without the banner on every terminal load.
Replaced all emojis with text-based indicators:
- ✓ → [OK] (with GREEN color)
- ✗ → [ERROR] (with RED color)
- ⚠ → WARNING: (with YELLOW color)
Color coding is preserved for visual distinction while remaining
accessible in all terminals and avoiding encoding issues.
Added support for installing Real-ESRGAN NCNN during setup:
**New Features:**
- Prompts user to install AI upscaling tools (Real-ESRGAN NCNN)
- Downloads and installs binary from GitHub releases
- Tries /usr/local/bin with sudo, falls back to ~/.local/bin
- Supports x86_64 architecture, warns for ARM
- Added --skip-ai flag to skip AI tools prompt
**Installation Flow:**
1. Ask: "Install DVD authoring tools?" (dvdauthor + xorriso)
2. Ask: "Install AI upscaling tools?" (Real-ESRGAN NCNN)
3. Auto-install all missing dependencies user requested
**Benefits:**
- Users get AI upscaling without manual download
- FFmpeg native upscaling still works without it
- Optional enhancement, not required
Previously, install.sh would:
1. Ask "Install DVD authoring tools?"
2. Then ask again "Install missing dependencies now?"
Now it automatically installs any missing dependencies after the
user confirms they want DVD tools, eliminating the redundant prompt.
Changes:
- Removed second confirmation prompt for dependency installation
- Automatically installs missing deps when detected
- Shows clear message: "Missing dependencies: ... Installing..."
The dependency check was accepting any ISO tool (mkisofs, genisoimage,
or xorriso), but the install commands now specifically install xorriso.
Updated checks to require xorriso specifically since:
- It handles both ISO9660 and UDF formats
- Required for RIP module to extract UDF DVD ISOs
- Now installed by all package managers in the script
This ensures the install script will detect and install xorriso even
if older tools like genisoimage are already present.
**Metadata Panel Fix:**
- Simplified makeRow layout to properly display values
- Changed from complex Border layout to simple HBox
- Metadata values now visible in Convert module
**RIP Module Improvements:**
- Added mount-based ISO extraction for UDF ISOs
- Added 7z extraction fallback
- Prioritizes xorriso > 7z > bsdtar for ISO extraction
- Handles both ISO9660 and UDF format DVDs
**Installation Script:**
- Updated all package managers to install xorriso
- Ensures proper UDF ISO support when disc modules enabled
- apt/dnf/zypper now install xorriso instead of genisoimage
- Wrap entire subtitle module content in droppable container for drag and drop anywhere
- Remove all emojis from build scripts (build-linux.sh, build.sh, build-windows.sh, alias.sh)
- Subtitle module now accepts video/subtitle file drops on any part of the view
Created setup-realesrgan-linux.sh for easy one-command installation:
- Downloads Real-ESRGAN ncnn Vulkan binary from GitHub releases
- Installs to ~/.local/bin/realesrgan-ncnn-vulkan
- Installs all AI models to ~/.local/share/realesrgan/models/
- Sets proper permissions
- Provides PATH setup instructions
Installation:
./scripts/setup-realesrgan-linux.sh
Models included (45MB):
- realesr-animevideov3 (x2, x3, x4) - Anime/illustration upscaling
- realesrgan-x4plus - General photo/video upscaling
- realesrgan-x4plus-anime - Anime-specific upscaling
Tested and working on Fedora 43. Makes AI upscaling fully automated
for Linux users - no manual downloads or configuration needed.
Next step: Add in-app "Install AI Upscaling" button to VideoTools UI
for even easier setup without using terminal.
Documents the complete snippet system redesign with dual output modes:
- "Snippet to Default Format": Stream copy mode for bit-perfect source preservation
- "Snippet to Output Format": Conversion preview using actual settings
Updates ai-speak.md with comprehensive snippet testing requirements and Jake's AI communication. Updates DONE.md with detailed feature breakdown, technical improvements, and bug fixes. Includes testing checklist for both snippet modes and batch generation.
Enhances screenshot module with comprehensive technical metadata display including audio bitrate, adds 8px padding between thumbnails for professional contact sheets. Implements new Player module for video playback access. Adds complete Filters and Upscale modules with traditional FFmpeg scaling methods (Lanczos, Bicubic, Spline, Bilinear) and resolution presets (720p-8K). Introduces configurable snippet length (5-60s, default 20s) with batch generation capability for all loaded videos. Fixes snippet duration precision by re-encoding instead of stream copy to ensure frame-accurate cutting at configured length.
🔧 File Detection
- Replaced nullglob with explicit file scanning
- Added more video formats (flv, webm, m4v, 3gp, mpg, mpeg)
- Better error reporting showing supported formats
⚡ Hardware Detection
- Added lshw support for Linux hardware detection
- Conditional Windows commands - only run wmic on Windows
- Improved GPU detection for Linux systems
⏱️ Timeout Handling
- Cross-platform timeout support:
- Linux: timeout
- macOS: gtimeout
- Windows: Background process with manual kill
📁 Path Handling
- Robust script directory detection for different shells
- Absolute module sourcing using SCRIPT_DIR
🖥️ Drag & Drop
- Better argument handling for Wayland desktop environments
- Comprehensive file extension support
Now works on:
- ✅ Windows x64 (Git Bash, WSL)
- ✅ Linux (Wayland, X11)
- ✅ macOS (Terminal)
