KDE Gear 21.12 has recently appeared as the latest release of the suite of applications that follow the parent project’s schedule (others like Krita and the Calligra suite do not). As usual in an environment as lively as KDE, applications tend to receive major improvements and changes at each iteration, so we take the opportunity to dig deeper into Kdenlive, the non-linear video editor.
Kdenlive is a well-known editor that little by little has made a place for itself in non-professional and even professional video editing that does not require the advanced possibilities offered by solutions such as DaVinci Resolve and Adobe Premiere. Within its segment, it is probably the most popular application among Linux users, although it also has a Windows user base. Its codebase is currently released under the GPLv3 license.
Kdenlive 21.12 comes with important new features that will surely make the application take an important qualitative leap, since it has incorporated a tracking algorithm based on deep learning. In other words, it has included support based on artificial intelligence, thus, at least roughly, in the wake of OpenShot.
Delving into the most interesting parts, Kdenlive 21.12 has included a automatic noise reduction filter and support for multiple containers in one project, which can import video footage or audio recording folders automatically ignoring any subdirectory structure on some devices such as Sony XDCam, Panasonic P2, Canon camcorders or Zoom audio recorders.
Among the new possibilities offered by the video editor is the offset crop mode, which moves the In and Out points of a clip forward or backward in one operation, keeping the same number of frames, the original duration, and without affecting adjacent clips. This feature is available by going to Tool> Scroll in the application menu.
Another of the most powerful novelties is the multi-camera tool, which allows you to view multiple video clips (or images) simultaneously by placing them on different tracks, but occupying the same position on the timeline. This way you can trim clips by pressing on the key assigned to the track while the playback continues through the timeline. As an alternative, you can press the different visualizations on the project monitor instead of on the keys.
And we come to the most interesting part of Kdenlive 21.12: the motion tracking. To start we have the DaSiamRPN tracking algorithm, which is supported by deep learning models to deliver accurate results. However, this feature requires downloading the artificial intelligence models, which must be obtained by following. On the other hand, the motion tracking filter comes with pixelated, dull or blur possibilities to hide a face (or anything else).
To everything mentioned so far can be added the official availability of a compilation for macOS x86 (Intel), the user manual migration to a new architecture called Sphinx which should be more user-friendly and easier to maintain, as well as improving the experience with existing tools and filters.
All the details of Kdenlive 21.12 are published in the, while the application can be obtained from the for Linux, Windows and macOS x86 (Intel). On Linux, the AppImage builds, Y are already in the version at hand, while the compilation in should be arriving shortly.