Skip to content

Install and Launch

Download a release

Open Ravo Releases and choose the package matching your operating system and processor.

Computer Download and launch
Apple-silicon macOS ARM64 DMG; copy Ravo Studio to Applications and open it
Intel macOS x86_64 DMG; copy Ravo Studio to Applications and open it
Windows x86_64 ZIP; extract the complete folder, then run ravo_studio.exe
Linux x86_64 or ARM64 Matching AppImage: mark it executable and open it; DEB: install with your package manager

Keep the complete packaged runtime together. Copying only the executable can leave Qt plugins, translations or codec libraries behind. Release notes describe the features and installation constraints of that download; the handbook follows the current development branch and can be ahead of it.

For Linux, packages target the documented modern distribution baseline. DEB installation should use the distribution package manager so it resolves desktop, PipeWire and VA-API dependencies. AppImage users need the corresponding host libraries and a working display session. On Ubuntu-family systems the multimedia libraries include libpipewire-0.3-0, libva2, libva-drm2 and libva-x11-2. See Packaging for the complete runtime baseline.

Ravo remains in active development. Choose the exact release notes and platform evidence for your download rather than assuming all features on main have already shipped.

First launch

Open Studio and create a library at a writable local location, or open an existing Ravo .sqlite catalog. The first-launch filename suggestion is Ravo Library.sqlite in Pictures.

Studio restores its last window geometry and workspace panel sizes. A stored window outside the current displays is fitted onto a connected screen. Choose File → Settings → General to select the interface language. Settings also groups workspace sizes, current-catalog backups and Assistant connection preferences.

Continue with First launch and import.

Build from source

These instructions are for contributors and people testing an unreleased development baseline.

You need CMake 3.26+, a C++20 compiler, Python, Ninja and Qt 6.11.2 with Core, Gui, Sql, Network, Qml, Quick, Quick Controls/Dialogs/Layouts, LinguistTools, Svg, ShaderTools and Multimedia. Install the complete image-format and SQLite runtime plugins. The selected Qt kit supplies matching FFmpeg 7.1.5 libraries.

Dependency preparation needs access to the pinned source-root repositories. Follow the dependency guide for kit paths and dependency access.

For a new checkout:

git submodule update --init FreeCM
python3 configs/source_root_workflow.py --init
python3 configs/source_root_workflow.py --update

For a prepared checkout, inspect its current roots before changing them:

python3 configs/source_roots.py show --format json
python3 configs/source_roots.py resolve --format json
python3 configs/source_roots.py verify

--init is the network-enabled preparation step. --update materialises existing dependencies offline and generates host presets; ordinary builds do not do either implicitly.

macOS

cmake --preset mac_clang_release -DBUILD_TESTING=ON
cmake --build --preset mac_clang_release
./build/mac_clang_release/Ravo/desktop/ravo_studio.app/Contents/MacOS/ravo_studio

Windows

From a matching MSVC developer environment:

cmake --preset win_msvc_release -DBUILD_TESTING=ON
cmake --build --preset win_msvc_release
build/win_msvc_release/Ravo/desktop/ravo_studio.exe

Linux

cmake --preset linux_clang_release -DBUILD_TESTING=ON
cmake --build --preset linux_clang_release
./build/linux_clang_release/Ravo/desktop/ravo_studio

Use the matching *_debug preset for Debug. Generated presets are host-specific; cmake --list-presets lists those prepared for the checkout. Build and package qualification is platform-specific.

Open a known library

Studio accepts these launch arguments on every supported host:

ravo_studio --catalog "/path/to/Ravo Library.sqlite" --language en_US

Use the actual executable path from the package or build tree. The catalog option also accepts --catalog=/path/to/library.sqlite. Language choices are English, German, Spanish, French, Brazilian Portuguese, Simplified/Traditional Chinese, Japanese and Korean. A missing selected language package reports an error.

Verify a source build

ctest --test-dir build/mac_clang_release --output-on-failure --parallel 4

Substitute your preset's build directory. For a short user-run workflow check, follow the smoke test.