Function decide_constant_quality 

fn decide_constant_quality(
    st: &mut StripeState,
    settings: &CaptureSettings,
    damage: Damage,
    is_animated: bool,
    recovery_idr: bool,
    encoder: EncoderQuality,
) -> HwFrameDecision

decide_hw_fullframe for a constant-quality session: the cleanup moves the session’s own quality to the paint-over one, as main’s paint-over did, rather than holding a frame at a quantizer. Each encoder then codes the cleanup the way it codes any frame at that quality: x264’s and x265’s rate factor give a key frame their intra offset (about three steps finer than the paint-over index), SVT-AV1 its key-frame qindex, and the encoders whose quality change re-opens the session (x265, SVT-AV1, kvazaar) open on a key frame at it. A frame held at the paint-over quantizer itself came out coarser than main’s (SVT-AV1 49.6 against 61.0 dB, x264 53.7 against 56.5 dB).

The session stays at the paint-over quality while the region stays clean: from its cleanup until motion (more than a small change), so the frames Turbo sends of a still screen, or a caret blinking on it, do not take a re-opening encoder back to the session’s quality, a key frame at it, after QP_HYSTERESIS_LIMIT frames. The refresh and the bursts come when cleanup_due says, as at a constant rate, and no key-frame cleanup follows: on NVENC at CRF 25 and paint-over 18 it lifted the end 0.8 dB for five times the bytes, one key frame of 415 kB where the refresh is 100. A requested or scheduled key frame is coded at the session’s own quality, as on main, and its burst at the paint-over one. An encoder whose quality change re-opens it (EncoderQuality::reopens: x265, kvazaar, SVT-AV1) codes the refresh that moves it as a key frame, as main’s paint-over did.