blancodagoat/DejaVu
Watch
1
0
Fork
You've already forked DejaVu
 0
Code Issues Pull requests Projects Releases 1 Packages Wiki Activity
1 release 21 tags
  • v1.5.0 148a9987bf

    v1.5.0 Stable

    blancodagoat released this 2026-11-07 21:01:02 +00:00 | 1 commits to main since this release

    DejaVu 1.5.0

    A maintenance release: saves work again on PCs where AV1 recording broke, and the
    clip can now carry more than one output device.

    Fixed: "Save failed" on some PCs with hardware AV1 (#9, #10)

    Thanks to @AaronTheFlash (#9) and @recui (#10) for the reports and the logs.

    On some PCs with an AV1-capable GPU, Windows' MP4 writer accepted the first few
    frames and then gave up (error 0xC00D4A45). DejaVu only stopped using AV1 when it
    failed before the first frame, so it picked AV1 again on every restart and failed
    about once a second. Each attempt left a broken file in the buffer, those pushed the
    good ones out, and pressing the hotkey ended in "Save failed".

    Now:

    • Before recording AV1, DejaVu writes a short test clip and reads it back. If that
      fails, or takes longer than a second, it records H.264 instead.
    • If AV1 fails later in the session, DejaVu switches to H.264 for the rest of the
      session, without a failure balloon. AV1 is tried again at the next launch.
    • A segment that won't play back is dropped instead of taking a slot in the buffer.
    • Issue reports say when AV1 was turned off.

    If you were affected, you should not need to change anything: install 1.5.0 and
    saves work again, in H.264.

    New: record more than one output device (#11)

    Thanks to @Greetedlight18 (#11) for the request.

    Virtual mixers like SteelSeries Sonar and Elgato Wave Link put the game and voice
    chat on separate output devices, so recording the default device alone missed one
    of them. The tray menu has a new Audio devices submenu: the default output is
    always recorded, and any other output you tick is recorded too, mixed into the same
    audio track.

    • Each device runs on its own clock, so its sound is resampled on the fly to stay in
      sync with the rest over the full 25-minute buffer.
    • A ticked device is recorded whole, Discord included. With Keep Discord out of clips
      on, Discord stays out of everything unless it plays only on ticked devices (for
      example Discord set to Sonar Chat, with Chat ticked). That is checked when each
      one-minute segment starts, so joining a call mid-segment can miss up to a minute.
    • A device that is unplugged, or disappears mid-recording (say, a Sonar restart),
      only loses its own sound. The buffer keeps going and picks the device up again at
      the next one-minute segment. Ticked devices that are unplugged show as
      "Disconnected device" so you can untick them.
    • If you make a ticked device your default output, it isn't recorded twice: DejaVu
      notices the switch straight away and leaves that device's own stream out until the
      next segment. With Captured app audio only on, that also leaves out its other
      sound (voice chat, for example) for up to a minute.
    • A virtual mixer's own effects (Sonar's ducking or voice processing, for example)
      are recorded the way they sound: DejaVu records each device's output as is.
    • Needs System audio on.
    • Your config carries over as is: the new extraAudioDevices key is optional, and a
      1.4 config without it isn't rewritten.

    Also

    • When a segment fails to finalize, its audio capture is still released.

    Launch video

    The README now shows DejaVu in 20 seconds instead of a screenshot; the full video with sound is attached below as dejavu-1.5.mp4. Music: Sascha Ende, ende.app (CC BY 4.0).

    Downloads