What's new since openscreen

Capturia is a fork of openscreen, re-baselined onto upstream v1.10.0. Everything in the table below arrived after that baseline, in Capturia 2.1 and 2.2. Feature names link to the page that documents them.

Feature	openscreen 1.10	Capturia 2.1–2.2
Area recording	Whole screen or one window	Drag a rectangle on a screen and record just that. Not offered on Linux.
Saved looks	Not in 1.10	Save background, effects, cursor and caption styling as a named look; star one as the default for new projects.
Zooms at flagged moments	No way to mark a moment while recording	Flag a moment from the HUD while recording; auto-enhance turns each flag into one zoom.
Right-click region menu + paste attributes	Ctrl/Cmd + C / V only — the menu was open as upstream #24	Copy, Paste at playhead, Delete on a region pill; Split at playhead on a clip; Ctrl/Cmd + Shift + V pastes attributes onto a region you already have.
Poster thumbnails	Recent projects listed without poster frames	Each recent project shows a frame grabbed from its own footage.
Speech-model choice	One built-in Whisper model, not configurable	Fast, Balanced, or Accurate — switch without losing existing transcripts.
Recordings folder	Default location only	Point new takes at any folder, and reset back. Earlier recordings stay where they were saved.
Pre-release channel	Stable releases only	Get pre-release builds in settings: opt in to release candidates, off by default.
Click highlight	Not in 1.10	An opt-in ring under the cursor on every recorded click, drawn by the same cursor pass in preview and export, so it follows zoom and tilt. Off by default; saved looks carry it.
Speed up idle	Not in 1.10	Auto-enhance lays 3× speed regions over stretches where the cursor sits still and nobody is speaking.
Export dialog memory	Defaults on every open	Format, quality, frame rate, codec, aspect ratio and GIF options come back from the last export that succeeded.
Auto-level	Open on the roadmap as voice enhancement	One click measures the take's loudness on this device and sets the output gain so speech lands around −16 LUFS, within ±12 dB.
Linux hardware encode	Software H.264 encode on Linux	H.264 export goes through h264_vaapi from the compositor's own dmabuf, and falls back to software when the driver refuses the imported frame. HEVC and non-VAAPI stacks stay software-encoded.

The roadmap tracks what is still open, including the items above that are only partly done.