v0.9.3 Meshroom core + dev maintenance

🚀 videostrip v0.9.3

This release marks a stabilization milestone for videostrip, separating the Meshroom plugin into its own repository and solidifying the CLI + core library for long-term use in underwater video transect processing.

🔑 Highlights

• Meshroom plugin split:
  The Meshroom node has moved to [videostrip-meshroom](https://github.com/cappelletto/videostrip-meshroom).
  videostrip now focuses purely on the CLI and core library.
• Stable CLI API:
  Arguments and output schema (frames.csv, summary.yaml) are locked at Schema v1.
  Future changes will maintain backward compatibility.
• Bundled OpenCV 4.12:
  Deterministic builds via vendored OpenCV; avoids system ABI conflicts.
  Licenses included in share/licenses/opencv/.
• Cross-platform CI/CD:
  Continuous builds & tests on Linux + Windows.
  Prebuilt binaries available in release assets.
• Enhancement pipeline:
  Configurable pre-export image enhancement (contrast, gamma, CLAHE, grayworld).
• Improved packaging:
  Optional DEB/TAR packages install into /opt/videostrip (avoids overwriting system libraries).

📦 Assets

• Linux: videostrip-v0.9.3-Linux-x86_64.tar.gz, .deb

• Windows: videostrip-v0.9.3-win64.zip

• Includes:

  • videostrip_cli binary
  • libvideostrip_core.a (static core library)
  • Configs + licenses

v0.9.0

Core feature complete public release under Apache 2.0 license

v0.8.2

Full Changelog: v0.7.1...v0.8.2

📦 Release Notes — v0.8.2

🚀 New Features

• Image Enhancement Module

  • Introduced a modular enhance/ pipeline with YAML-driven configuration.
  • Supports configurable enhancement sequence (contrast, gray-world white balance, gamma, CLAHE).
  • CLI override syntax for rapid testing.
  • Enhancement stage executes before feature extraction/export.

• Feature Density Normalization

  • Added grid-based feature density normalization (#23).
  • Provides more uniform feature distribution for SfM workflows.
  • Includes configuration structures and working implementation.

• Documentation & Packaging

  • Added manpage target in CMake build system.
  • Updated README with enhancement pipeline usage and feature list.
  • Listed packaging and manpages as supported features.

🛠 Fixes & Improvements

• Patched yaml-cpp detection across Linux/Nix environments.
• Fixed missing CONFIG stage in top-level CMake.
• Corrected enhancement sequence call ordering and enable flag propagation.
• Enhanced EnhanceStage.process() to handle 8/16-bit grayscale and multi-channel images.
• Disabled CLAHE test for grayscale deterministic inputs (non-reproducible).
• Fixed namespace issues in enhancement units.
• Added 3 unit tests for enhancement pipeline (config, stage, ops).
• Re-enabled YAML container for enhancement config.

🔧 Chores & Maintenance

• Promoted enhancement code into its own <enhance> module.
• Refined CMake structure and removed stale TODOs.
• Added TODO notes for missing dependencies (system-wide yaml).
• Updated .gitignore for generated docs.

📑 Versioning

• v0.7.1 → v0.8.0: Initial enhancement pipeline, CMake manpage target.
• v0.8.0 → v0.8.1: Test suite integration for enhancement.
• v0.8.1 → v0.8.2: Grid-based feature density normalization, bug fixes, expanded docs.

Stable Metadata Integration: MVP+ release with locked Schema v1, reproducible metadata outputs (CSV + YAML)

📦 Release v0.6.0 — Stable Metadata Integration

Highlights

This release delivers the first stable contract (Schema v1) for frame extraction metadata, ensuring repeatability, reproducibility, and downstream compatibility with SfM/MVS pipelines.

🚀 New Features

• Schema v1 locked for output contract:
  • frames.csv with per-frame metadata (frame_idx, timestamp_ms, output_image, feature_count, quality_score, georef).
  • summary.yaml with schema version, run info, configuration snapshot, and list of extracted images.
  • Stable file/folder layout: images/, features/, frames.csv, summary.yaml, run.log.
• MetadataWriter module (videostrip_core/io/metadata_writer):
  • Centralized CSV + YAML writing.
  • Append-safe CSV (row written immediately per frame).
  • YAML written once at run completion.
• VideoFrameExtractor::run refactor:
  • Integrated MetadataWriter for schema v1 outputs.
  • Removed legacy writeMetadataCSV() / writeSummaryYAML() functions.
  • Guarantees reproducible metadata even if errors occur during run.
• Logging:
  • Errors, warnings, and run completion messages consistently recorded in run.log.

🛠 CI/CD & Testing

• Added unit tests for schema writers (CSV + YAML regression).
• GitHub Actions now builds, runs smoke tests, and validates schema outputs.
• Both Linux and Windows targets covered.

📚 Documentation

• Updated user guide with schema v1 contract.
• Provided sample YAML config and usage examples.

---

Upgrade Notes

• Consumers of metadata must use frames.csv and summary.yaml (schema v1). These formats are stable going forward.
• Old ad-hoc CSV/YAML functions have been retired.

---

Next Steps

• Add optional frame enhancement filters (CLAHE, WB).
• Introduce grid-based feature density normalization as an experimental mode.
• Expand Windows CI/CD to cover feature parity tests with Linux.

---

✅ This release marks the first stable MVP+ slice: reliable frame extraction, reproducible metadata, and direct downstream compatibility with Meshroom/Metashape pipelines.