fix(quality): scale after videoconvert at exact even WxH
Live medium-quality stream errored with "negotiation problem" on the host and rendered a squashed, garbled picture in the viewer. Two causes, both from inserting videoscale before videoconvert with PAR+range caps: - videoscale was scaling pipewiresrc's raw output directly. The portal source's format/memory (e.g. DMABuf) isn't something software videoscale negotiates — the original pipeline always fed pipewiresrc through videoconvert first. Move videoscale *after* videoconvert so it operates on system-memory NV12/I420. - `pixel-aspect-ratio=1/1` + a width range over-constrained negotiation and risked a non-square-PAR / distorted result. Instead compute an exact even WxH from the known source dimensions (Wayland: portal size; X11: root/window geometry), preserving aspect, and pin it fully in the caps. This is also downscale-only now — a source already at/below the target height is left native instead of upscaled. Unknown dims (rare X11 geometry failure) fall back to the height-only + square-pixel + even width-range negotiation. source_dims threaded through pipeline::spawn from both backends. Smoke test updated to mirror the new ordering (1920x1080 -> 852x480, videoscale after videoconvert) and still asserts an even sub-source width. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
+16
-8
@@ -21,13 +21,21 @@ pub async fn start(opts: &HostOpts, quality: &EffectiveQuality) -> Result<Captur
|
||||
None
|
||||
};
|
||||
|
||||
// Geometry is informational (mirrors Wayland's portal-handshake log line);
|
||||
// a failure here shouldn't abort capture — ximagesrc will surface a real
|
||||
// error if the X connection is genuinely unusable.
|
||||
match read_geometry(xid) {
|
||||
Ok((w, h)) => tracing::info!(width = w, height = h, xid = ?xid, "X11 capture geometry"),
|
||||
Err(e) => tracing::warn!("could not read X11 geometry (capture will still try): {e:#}"),
|
||||
}
|
||||
// Geometry mirrors Wayland's portal-handshake log line and feeds the
|
||||
// downscale presets (so they can compute an exact target size). A failure
|
||||
// here shouldn't abort capture — ximagesrc will surface a real error if the
|
||||
// X connection is genuinely unusable, and the scaler falls back to a
|
||||
// height-only negotiation when dims are unknown.
|
||||
let source_dims = match read_geometry(xid) {
|
||||
Ok((w, h)) => {
|
||||
tracing::info!(width = w, height = h, xid = ?xid, "X11 capture geometry");
|
||||
Some((w as u32, h as u32))
|
||||
}
|
||||
Err(e) => {
|
||||
tracing::warn!("could not read X11 geometry (capture will still try): {e:#}");
|
||||
None
|
||||
}
|
||||
};
|
||||
|
||||
let mut source_args = vec![
|
||||
"ximagesrc".to_string(),
|
||||
@@ -42,7 +50,7 @@ pub async fn start(opts: &HostOpts, quality: &EffectiveQuality) -> Result<Captur
|
||||
}
|
||||
|
||||
// X11 has no leaked fd to clean up, so the post-spawn hook is a no-op.
|
||||
pipeline::spawn(opts, quality, source_args, || {}).await
|
||||
pipeline::spawn(opts, quality, source_dims, source_args, || {}).await
|
||||
}
|
||||
|
||||
/// Run `xwininfo` and let the user click the window they want to share, then
|
||||
|
||||
Reference in New Issue
Block a user