Skip to content

VideoHighlighter 0.12.1

Choose a tag to compare

@github-actions github-actions released this 22 Sep 05:18
· 65 commits to main since this release

VideoHighlighter 0.12.1

A patch release, mostly for macOS. On a tester's Mac, the 0.12.0 build kept using more memory until the machine had to be restarted. This release fixes that, and the macOS build is now tested before it ships. On AMD cards, action recognition can now reach the GPU even when torch-directml is installed.

7 commits on main since 0.12.0.

macOS

  • The app no longer fills the memory. A packaged build starts each helper process by running the app again, and every helper loaded the whole app (Qt, OpenVINO, the language models) before it realised it was a helper. Object detection alone starts six of them. Helpers now stop before any of that loads.
  • The debug log survives the run. Each helper also took itself for a fresh launch and moved the running app's debug.log aside, which is why the logs from the runs that mattered kept going missing.
  • The timeline filmstrip falls back to software decoding when the hardware decoder opens a file but cannot decode it, which happens with ProRes on some Macs. Before, the filmstrip stayed empty and nothing was logged.
  • The macOS build is tested before it becomes a DMG. The packaged app runs without a window on two generated clips, 720p H.264 and 1080p ProRes. The test checks that it imports, that helpers do not load the app again, that object detection starts its workers, that the filmstrip gets a frame and that the debug log survives, all within a memory limit. A failure stops the release.

GPU and compute

  • R3D on AMD cards with torch-directml installed. torch-directml cannot run the 3D convolutions R3D is made of. So when it was installed, R3D fell back to the processor without trying ONNX Runtime's DirectML path, which does run it (27.9 ms a window on an RX 570). It now tries that path before giving up on the card. The app does not install torch-directml itself, so this only affected people who had installed it by hand.

Documentation

  • The READMEs and CONTRIBUTING.md link the project's Discord server.

The project

VideoHighlighter is owned and maintained jointly by Przemysław Kreft (@Aseiel) and Meric Donmezer (@mericdonmezer-ops), its two copyright holders.

Full Changelog: 0.12.0...0.12.1

Download

Install VideoHighlighter for Windows

One file. Run it, click through, done — it fetches and unpacks
everything else itself. Installs per user, no admin rights needed.

Installer Portable
Windows (desktop app) Setup .exe .7z part 1 + part 2
Windows (web UI) — .7z part 1 + part 2
macOS — .dmg

The installer is not signed yet, so Windows SmartScreen shows
"Windows protected your PC" on first run — More info → Run anyway.

The portable downloads are split archives: every .7z.0NN part
listed under Assets must sit in the same folder, then extract the
.001 with 7-Zip. Downloading one part alone fails. The installer
above exists so you never have to do that.

The web UI is the Tauri shell around the React frontend with the
Python engine bundled; it has no installer because no installer can
hold a 1.7 GB sidecar.

00-VideoHighlighter-Windows-Setup.zip is the older script-based
installer, kept as a fallback.