Install ClipMill.

Set up your workspace and make your first cut. Everything you need, right here.

Install on macOS

Apple silicon (M1 or newer) · macOS 14 or later

  1. Download and open the disk image

    Choose the macOS build on the download page and open the .dmg file.

  2. Move ClipMill to Applications

    Drag ClipMill into your Applications folder, then open it from there.

  3. Allow the first launch

    The current Mac app is not notarized. If macOS says it cannot verify ClipMill, choose Done. Open System Settings → Privacy & Security → Open Anyway, then choose Open.

Check your download first.

Use the installer from this site. Its SHA-256 checksum is listed on the download page. You only need to allow the Mac app once. An Intel Mac build is not currently available.

Install on Windows

Beta

Windows 10 or 11 · 64-bit (x64)

  1. Run the setup file

    Choose the Windows build on the download page and open the file ending in _x64-setup.exe.

  2. Review the first-run prompt

    The installer is not signed yet. After checking that you downloaded it from this site, select More info → Run anyway if Windows displays “Windows protected your PC.”

  3. Open ClipMill from Start

    The app installs for your user account and appears in the Start menu. The installer adds Microsoft Edge WebView2 if it is missing.

The installer also adds Microsoft’s Visual C++ runtime when needed, with a Windows permission prompt. If you decline that step, install the runtime before running ClipMill’s component setup.

Windows runs the local Qwen editorial model through llama.cpp, using a suitable Vulkan graphics card or the CPU. A built-in picker remains available without the editorial model.

Install on Linux

x86_64 · Ubuntu 22.04 or a distribution with a compatible glibc

Choose an AppImage or Debian package on the download page. Run the command from the folder containing your download.

AppImage

Make the file executable, then open it. The AppImage includes its WebView libraries.

chmod +x ClipMill_0.4.0_amd64.AppImage
./ClipMill_0.4.0_amd64.AppImage

Debian or Ubuntu

Install the package with apt. It installs the required WebView, OpenGL, and GLib libraries too.

sudo apt install ./ClipMill_0.4.0_amd64.deb

Linux runs the local Qwen editorial model through llama.cpp, using a suitable Vulkan graphics card or the CPU. A built-in picker remains available without the editorial model.

A little setup. Then it’s yours.

On first launch, Set up ClipMill shows the processing components and recommended models for your computer, including their approximate download sizes.

  1. Start setup

    Keep your internet connection active while ClipMill downloads and verifies its components and models.

  2. Let the local models get ready

    ClipMill checks that the editorial model runs on your computer before using it. Apple silicon uses MLX; Windows and Linux use llama.cpp. CPU-only machines can take several minutes. Check progress, retry, or choose a smaller model in Models.

  3. Bring your first recording

    Create a project and choose a video. Once setup is complete, local analysis and export work offline.

Components come from pinned Python and package downloads; models come from Hugging Face. Downloaded files are verified against their expected hashes. Your project content is not sent during setup.

Read about local processing

Bringing a video from YouTube

Setup installs the YouTube import component. It also needs Node.js 22 or newer installed on your computer. ClipMill can find installations from the official installer, Homebrew, nvm, Volta, or fnm.

YouTube import uses the internet to download the selected video before local processing begins. Only import videos you have permission to use.

Keeping ClipMill up to date

Download the latest installer from this site and install it over your existing app. Your projects and models live separately from the application.

ClipMill can check GitHub once a day and show a notice when a newer version is available. It does not download or install an app update automatically. Turn off Say when a new version is out in Settings → About to disable those network checks.

The release notes explain what changed in each version. The download page follows the latest published release automatically.

A place for your projects

Projects, components, models, and logs stay in your local application data folder. Settings → Storage shows what uses space and what you can remove.

macOS
~/Library/Application Support/dev.clipmill.ClipMill
Windows
%LOCALAPPDATA%\clipmill\ClipMill\data
Linux
~/.local/share/clipmill

The engine/ folder holds components, models/ holds model files, and logs/ contains process logs. Quitting ClipMill stops its background processes; pending work resumes when you next open it.

If you need to uninstall

On macOS, move ClipMill from Applications to the Bin. On Windows, use Settings → Apps → Installed apps → ClipMill → Uninstall. On Linux, remove the Debian package with your package manager or delete the AppImage.

To remove all local data as well, delete the corresponding folder above. This also deletes your saved projects, so back up anything you want to keep first.

Getting unstuck

Setup stopped before it finished

Choose Try again in setup. The full installer output is in logs/engine-install.log inside your application data folder.

A component keeps restarting

ClipMill retries a stopped process with a longer pause between attempts. Check logs/workers/ in your application data folder for the process log.

The Linux window is blank on an NVIDIA GPU

Start ClipMill with the following environment setting. For an AppImage, use:

WEBKIT_DISABLE_DMABUF_RENDERER=1 ./ClipMill_0.4.0_amd64.AppImage
Where are the included software licences?

They are in the app’s resources/licenses folder, starting with README.txt. On macOS, Control-click ClipMill, choose Show Package Contents, then open Contents/Resources/resources/licenses.