feat(host): X11 capture backend + shared pipeline extraction

Extract the display-agnostic encode/mux tail out of wayland.rs into a new
host/pipeline.rs: CaptureHandle + lifecycle, audio routing setup, the gst
arg builder, the spawn, and Serve::bind now live there. Backends supply
only their video-source element args plus a post-spawn hook (Wayland uses
it to close its leaked pipewire fd; X11 passes a no-op). capture.rs
collapses to a thin dispatcher; its CaptureHandle enum is gone.

Add host/x11.rs: ximagesrc (use-damage=false show-pointer=true), whole
root window by default or a single window via --window (xwininfo
click-picker → xid). x11rb reads geometry for an info log, justifying the
previously-vestigial dep. No portal, no fd dance — capture starts
silently when the first viewer connects (the ticket is the access
control). Viewer is display-agnostic and unchanged.

Wire --no-hwencode for real (was a no-op): the shared tail now selects
x264enc(tune=zerolatency,ultrafast)/I420 vs vah264enc/NV12 and switches
the videoconvert target format to match. Applies to both backends.

deps.rs: check_host_binaries now takes &HostOpts and checks shared
elements for both backends, encoder by --no-hwencode, source per backend
(pipewiresrc/ximagesrc), and xwininfo only when X11 + --window. Install
hints added for x264enc, ximagesrc, xwininfo.

Verified: warning-free build; smoke test still passes (tail unchanged);
ximagesrc + both encoder tails produce mpv-decodable H.264 against an
Xwayland root. Interactive cross-machine end-to-end pending.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

README.md

@@ -16,8 +16,10 @@ without spinning up a Discord call or fighting with NAT.
 Working:
 - Wayland capture via the screencast portal (KDE Plasma 6 confirmed; other
   Wayland compositors with the portal should work but are untested)
+- X11 capture via `ximagesrc` (whole screen, or a single window with
+  `--window`); selected automatically, or forced with `--display-server x11`
 - VAAPI H.264 encode in GStreamer (RDNA3 confirmed; other VAAPI-capable
-  GPUs should work)
+  GPUs should work), with a software x264 fallback via `--no-hwencode`
 - Audio capture of the default sink's monitor, with optional per-app
   routing (`--app <name>`) and microphone mixing (`--mic`)
 - `--repair` cleanup of orphaned PipeWire state left by a crashed host
@@ -30,8 +32,10 @@ Working:
   `~/.config/pixelpass/config.toml` and used to auto-size the default
   viewer cap
 
-Not yet working:
+Not yet built (deferred, not blocking):
-- X11 capture (stubbed, returns an error — Phase 2 follow-up)
+- Per-monitor selection on a multi-monitor X11 host — `ximagesrc` grabs the
+  whole root canvas; single-monitor cropping needs xrandr region coords
+- `use-damage=true` CPU optimization for the X11 capture path
 
 ## Quick start
 
@@ -68,7 +72,7 @@ pixelpass <ticket>
 
 ## Requirements
 
-- Linux (Wayland session for now; X11 stubbed)
+- Linux (Wayland or X11; the backend is autodetected)
 - A VAAPI-capable GPU and the right driver:
   - AMD: `libva-mesa-driver`
   - Intel: `intel-media-driver` (modern iGPUs) or `intel-vaapi-driver` (older)
@@ -117,10 +121,10 @@ cargo build --release
 ```
   Host                                                          Viewer
   ────                                                          ──────
-  Wayland portal (ashpd) ──> PipeWire fd
+  Wayland portal (ashpd) ──> PipeWire fd ─┐   (X11: ximagesrc, no portal)
-                              │
+                                          │
-                              ▼
+                                          ▼
-              gst-launch: pipewiresrc -> videorate -> vah264enc ->
+              gst-launch: <source> -> videorate -> vah264enc/x264enc ->
                           h264parse -> mpegtsmux
                           (audio:    pulsesrc <sink>.monitor ->
                                      avenc_aac -> aacparse ─┘)
@@ -142,7 +146,7 @@ cargo build --release
 The viewer's player connects to a localhost HTTP server, which is
 just one end of the iroh tunnel. The host's HTTP server sits on the
 other end and streams GStreamer's stdout (an MPEG-TS containing
-hardware-encoded H.264 + AAC) through with no demux or remux.
+H.264 + AAC) through with no demux or remux.
 
 iroh handles NAT traversal: direct UDP if hole-punching succeeds,
 relay path otherwise. Both have been verified end-to-end.
@@ -201,6 +205,27 @@ If a host crashes mid-session it can leave orphaned `pixelpass_capture_*`
 null-sinks and their paired loopbacks loaded in PipeWire. Run
 `pixelpass --repair` to unload them and exit.
 
+## Display server
+
+The capture backend is autodetected from the environment
+(`WAYLAND_DISPLAY` → Wayland, else `DISPLAY` → X11, else
+`XDG_SESSION_TYPE`). Override it with `--display-server wayland|x11` — for
+example to force the X11 path while running inside a Wayland session (an
+Xwayland or Xephyr `DISPLAY`).
+
+- **Wayland** goes through the screencast portal: a "Share Screen?" dialog
+  appears when the first viewer connects, and you pick the monitor (or
+  window, with `--window`) there.
+- **X11** uses `ximagesrc` and starts silently when the first viewer
+  connects — the ticket is the access control, there's no portal gate.
+  `--window` runs an `xwininfo` picker (click the window you want to
+  share); without it the whole root window is captured.
+
+Encoding is hardware VAAPI (`vah264enc`) by default. `--no-hwencode`
+switches to software x264 (`x264enc tune=zerolatency`) for hosts without a
+working VAAPI H.264 entrypoint — higher CPU, no GPU needed. This applies
+to both backends.
+
 ## Multi-viewer
 
 One gst capture pipeline fans out to N concurrent viewers via a

src/common/deps.rs

@@ -2,21 +2,45 @@ use anyhow::{Result, bail};
 use std::path::PathBuf;
 use std::process::Command;
 
+use crate::cli::HostOpts;
 use crate::common::display::DisplayServer;
 
-pub fn check_host_binaries(display: DisplayServer) -> Result<()> {
+pub fn check_host_binaries(display: DisplayServer, opts: &HostOpts) -> Result<()> {
-    if display == DisplayServer::Wayland {
+    // Unknown is handled (and rejected) by the caller; nothing to check here.
-        require("gst-launch-1.0")?;
+    if display == DisplayServer::Unknown {
-        require("gst-inspect-1.0")?;
+        return Ok(());
-        require("pactl")?;
-        require_gst_element("pipewiresrc")?;
-        require_gst_element("vah264enc")?;
-        require_gst_element("h264parse")?;
-        require_gst_element("mpegtsmux")?;
-        require_gst_element("pulsesrc")?;
-        require_gst_element("avenc_aac")?;
-        require_gst_element("aacparse")?;
     }
+
+    // Shared across both backends: the gst tools, audio routing, and the
+    // encode/mux tail elements.
+    require("gst-launch-1.0")?;
+    require("gst-inspect-1.0")?;
+    require("pactl")?;
+    require_gst_element("h264parse")?;
+    require_gst_element("mpegtsmux")?;
+    require_gst_element("pulsesrc")?;
+    require_gst_element("avenc_aac")?;
+    require_gst_element("aacparse")?;
+
+    // Encoder depends on --no-hwencode (software x264 vs hardware VAAPI).
+    if opts.no_hwencode {
+        require_gst_element("x264enc")?;
+    } else {
+        require_gst_element("vah264enc")?;
+    }
+
+    // Per-backend video source, plus the X11 window-picker when --window is set.
+    match display {
+        DisplayServer::Wayland => require_gst_element("pipewiresrc")?,
+        DisplayServer::X11 => {
+            require_gst_element("ximagesrc")?;
+            if opts.window {
+                require("xwininfo")?;
+            }
+        }
+        DisplayServer::Unknown => unreachable!("early-returned above"),
+    }
+
     Ok(())
 }
 
@@ -69,6 +93,13 @@ fn install_hint_for_bin(bin: &str) -> String {
             Some("opensuse" | "opensuse-tumbleweed" | "opensuse-leap") => "pulseaudio-utils",
             _ => "pulseaudio-utils (provides `pactl`)",
         },
+        "xwininfo" => match distro.as_deref() {
+            Some("arch" | "cachyos" | "manjaro" | "endeavouros") => "xorg-xwininfo",
+            Some("debian" | "ubuntu" | "pop" | "linuxmint") => "x11-utils",
+            Some("fedora" | "nobara") => "xorg-x11-utils",
+            Some("opensuse" | "opensuse-tumbleweed" | "opensuse-leap") => "xwininfo",
+            _ => "xwininfo (X11 window-info utility)",
+        },
         _ => bin,
     };
     install_command(&distro, pkg)
@@ -91,6 +122,20 @@ fn install_hint_for_gst_element(name: &str) -> String {
             Some("opensuse" | "opensuse-tumbleweed" | "opensuse-leap") => "gstreamer-plugins-bad",
             _ => "the GStreamer VA-API plugin (requires an H.264-capable GPU; almost all modern GPUs)",
         },
+        "x264enc" => match distro.as_deref() {
+            Some("arch" | "cachyos" | "manjaro" | "endeavouros") => "gst-plugins-ugly",
+            Some("debian" | "ubuntu" | "pop" | "linuxmint") => "gstreamer1.0-plugins-ugly",
+            Some("fedora" | "nobara") => "gstreamer1-plugins-ugly",
+            Some("opensuse" | "opensuse-tumbleweed" | "opensuse-leap") => "gstreamer-plugins-ugly",
+            _ => "the GStreamer x264 plugin (plugins-ugly)",
+        },
+        "ximagesrc" => match distro.as_deref() {
+            Some("arch" | "cachyos" | "manjaro" | "endeavouros") => "gst-plugins-good",
+            Some("debian" | "ubuntu" | "pop" | "linuxmint") => "gstreamer1.0-plugins-good",
+            Some("fedora" | "nobara") => "gstreamer1-plugins-good",
+            Some("opensuse" | "opensuse-tumbleweed" | "opensuse-leap") => "gstreamer-plugins-good",
+            _ => "the GStreamer X11 plugin (plugins-good)",
+        },
         "h264parse" | "mpegtsmux" | "aacparse" => match distro.as_deref() {
             Some("arch" | "cachyos" | "manjaro" | "endeavouros") => "gst-plugins-bad",
             Some("debian" | "ubuntu" | "pop" | "linuxmint") => "gstreamer1.0-plugins-bad",

src/host/capture.rs

@@ -1,40 +1,20 @@
-//! Capture dispatcher. Selects the per-display-server pipeline and returns its
+//! Capture dispatcher. Picks the per-display-server source backend and returns
-//! [`CaptureHandle`]. Each backend owns its own children and tears them down
+//! the shared [`pipeline::CaptureHandle`]. The handle itself, its teardown, and
-//! via its own Drop / shutdown.
+//! the whole encode/serve tail are display-agnostic and live in
+//! [`super::pipeline`]; the backends differ only in how they obtain a video
+//! source element.
 
-use anyhow::{Result, bail};
+use anyhow::Result;
 
 use crate::cli::HostOpts;
 use crate::common::display::DisplayServer;
-use crate::host::wayland;
+use crate::host::pipeline::CaptureHandle;
-
+use crate::host::{wayland, x11};
-pub enum CaptureHandle {
-    Wayland(wayland::CaptureHandle),
-}
-
-impl CaptureHandle {
-    pub fn local_port(&self) -> u16 {
-        match self {
-            CaptureHandle::Wayland(h) => h.local_port(),
-        }
-    }
-
-    pub async fn shutdown(self) {
-        match self {
-            CaptureHandle::Wayland(h) => h.shutdown().await,
-        }
-    }
-}
 
 pub async fn spawn(display: DisplayServer, opts: &HostOpts) -> Result<CaptureHandle> {
     match display {
-        DisplayServer::Wayland => {
+        DisplayServer::Wayland => wayland::start(opts).await,
-            let h = wayland::start(opts).await?;
+        DisplayServer::X11 => x11::start(opts).await,
-            Ok(CaptureHandle::Wayland(h))
-        }
-        DisplayServer::X11 => {
-            bail!("X11 capture pipeline not yet implemented (Phase 2 follow-up)");
-        }
         DisplayServer::Unknown => unreachable!("caller guarantees display != Unknown"),
     }
 }

src/host/mod.rs

@@ -1,7 +1,9 @@
 pub mod audio;
 mod capture;
+mod pipeline;
 mod serve;
 mod wayland;
+mod x11;
 
 use anyhow::{Result, bail};
 use iroh::endpoint::{Connection, presets};
@@ -17,7 +19,7 @@ use crate::common::{
     tunnel,
 };
 
-use self::capture::CaptureHandle;
+use self::pipeline::CaptureHandle;
 
 /// Messages from per-viewer tasks to the capture supervisor.
 enum SupervisorMsg {
@@ -32,7 +34,7 @@ enum SupervisorMsg {
 
 pub async fn run(opts: HostOpts) -> Result<()> {
     let display = DisplayServer::resolve(opts.display_server);
-    deps::check_host_binaries(display)?;
+    deps::check_host_binaries(display, &opts)?;
 
     if display == DisplayServer::Unknown {
         bail!(
@@ -249,7 +251,7 @@ fn print_host_banner(
     eprintln!("│  display server : {display:?}");
     eprintln!("│  capture        : {}", capture_summary(opts));
     eprintln!("│  bitrate / fps  : {} kbps @ {} fps", opts.bitrate, opts.framerate);
-    eprintln!("│  hw encode      : {}", if opts.no_hwencode { "off" } else { "auto (VAAPI if available)" });
+    eprintln!("│  hw encode      : {}", if opts.no_hwencode { "off (software x264)" } else { "on (VAAPI H.264)" });
     eprintln!("│  max viewers    : {} ({})", resolution.value, resolution.source.label());
     eprintln!("│");
     if clipboard_ok {

src/host/pipeline.rs

@@ -0,0 +1,255 @@
+//! Display-server-agnostic capture pipeline. The video *source* element is the
+//! only part that differs between Wayland (`pipewiresrc`, after a portal
+//! handshake) and X11 (`ximagesrc`); everything downstream — the videorate cap,
+//! the encoder, `h264parse`, `mpegtsmux`, the whole audio branch, the gst spawn,
+//! the [`Serve`] fanout binding, and the [`CaptureHandle`] lifecycle — is shared
+//! and lives here. Backends call [`spawn`] with just their source-element args.
+
+use anyhow::{Context, Result, bail};
+use nix::sys::signal::{Signal, kill};
+use nix::unistd::Pid;
+use std::process::Stdio;
+use std::time::Duration;
+use tokio::process::{Child, Command};
+use tokio::time::timeout;
+
+use super::audio::Routing;
+use super::serve::Serve;
+use crate::cli::HostOpts;
+
+pub struct CaptureHandle {
+    gst: Option<Child>,
+    audio: Option<Routing>,
+    serve: Option<Serve>,
+}
+
+impl CaptureHandle {
+    pub fn local_port(&self) -> u16 {
+        self.serve
+            .as_ref()
+            .expect("serve is always Some until shutdown")
+            .local_port()
+    }
+
+    /// Graceful teardown: SIGTERM gst, give it ~1s to exit, then SIGKILL,
+    /// unload audio routing (if any), then tear down the serve layer.
+    /// The serve reader will see EOF on gst stdout and exit on its own;
+    /// serve.shutdown() is the backstop.
+    pub async fn shutdown(mut self) {
+        if let Some(child) = self.gst.as_mut()
+            && let Some(pid) = child.id()
+        {
+            let _ = kill(Pid::from_raw(pid as i32), Signal::SIGTERM);
+        }
+        if let Some(child) = self.gst.as_mut() {
+            let _ = timeout(Duration::from_millis(1000), child.wait()).await;
+            let _ = child.start_kill();
+        }
+        if let Some(audio) = self.audio.take() {
+            audio.shutdown();
+        }
+        if let Some(serve) = self.serve.take() {
+            serve.shutdown().await;
+        }
+    }
+}
+
+impl Drop for CaptureHandle {
+    fn drop(&mut self) {
+        if let Some(child) = self.gst.as_mut() {
+            let _ = child.start_kill();
+        }
+        // Routing's and Serve's own Drop impls handle the rest.
+    }
+}
+
+/// Spawn the shared gst pipeline for a backend that supplies `source_args`
+/// (the video-source element + its properties, e.g. `["pipewiresrc", "fd=7",
+/// …]` or `["ximagesrc", "use-damage=false", …]`). `after_spawn` runs once,
+/// immediately after the gst child is launched — Wayland uses it to `close`
+/// the pipewire fd it leaked into the child; X11 passes a no-op.
+pub async fn spawn(
+    opts: &HostOpts,
+    source_args: Vec<String>,
+    after_spawn: impl FnOnce(),
+) -> Result<CaptureHandle> {
+    let (audio_routing, audio_device) = setup_audio(opts).await?;
+    let args = build_args(&source_args, &audio_device, opts);
+
+    let mut gst_cmd = Command::new("gst-launch-1.0");
+    gst_cmd
+        .args(&args)
+        .stdin(Stdio::null())
+        .stdout(Stdio::piped())
+        .stderr(Stdio::inherit());
+    if std::env::var_os("PIXELPASS_GST_DEBUG").is_some() {
+        gst_cmd.env("GST_DEBUG", "3");
+    }
+    let mut gst = gst_cmd.spawn().context("failed to spawn gst-launch-1.0")?;
+
+    // Backend-specific post-spawn cleanup (Wayland closes its leaked pw fd here,
+    // once gst has inherited its own copy).
+    after_spawn();
+
+    let gst_stdout = gst
+        .stdout
+        .take()
+        .context("gst-launch-1.0 stdout pipe unavailable")?;
+
+    // Hand stdout to the serve layer, which binds the localhost HTTP listener
+    // and runs the broadcast fanout. No demux/remux, no codec assumptions.
+    let serve = Serve::bind(gst_stdout).await?;
+
+    Ok(CaptureHandle {
+        gst: Some(gst),
+        audio: audio_routing,
+        serve: Some(serve),
+    })
+}
+
+/// Decide whether per-app audio routing is active and produce the `device=…`
+/// argument for `pulsesrc`. Routing activates when either `--app` is set
+/// (per-stream rerouting to a per-PID null-sink) or `PIXELPASS_AUDIO_VIA_NULL_SINK=1`
+/// is set (no app filter — captures everything via the null-sink, used for
+/// dogfooding the loopback path). Otherwise we capture the default sink's
+/// monitor (system audio out), not the default source (the mic).
+async fn setup_audio(opts: &HostOpts) -> Result<(Option<Routing>, String)> {
+    let routing_requested =
+        opts.app.is_some() || std::env::var_os("PIXELPASS_AUDIO_VIA_NULL_SINK").is_some();
+    let audio_routing = if routing_requested {
+        Some(
+            Routing::start(opts)
+                .await
+                .context("audio routing setup failed")?,
+        )
+    } else {
+        None
+    };
+    let audio_device = if let Some(r) = &audio_routing {
+        format!("device={}.monitor", r.sink_name())
+    } else {
+        let default = default_audio_monitor().await?;
+        format!("device={default}")
+    };
+    Ok((audio_routing, audio_device))
+}
+
+/// Build the full gst-launch argument vector: MPEG-TS mux + fdsink, then the
+/// video branch (caller's `source` → videorate cap → encoder → h264parse →
+/// mux.), then the audio branch (pulsesrc → AAC → mux.). The encoder and the
+/// `videoconvert` target format are selected by `opts.no_hwencode`:
+/// hardware VAAPI wants NV12, software x264 wants I420.
+fn build_args(source: &[String], audio_device: &str, opts: &HostOpts) -> Vec<String> {
+    let key_interval = (opts.framerate * 2).to_string();
+    let bitrate = opts.bitrate.to_string();
+    let framerate_caps = format!("video/x-raw,framerate={}/1", opts.framerate);
+
+    let (raw_format, encoder_args): (&str, Vec<String>) = if opts.no_hwencode {
+        (
+            "video/x-raw,format=I420",
+            vec![
+                "x264enc".into(),
+                "tune=zerolatency".into(),
+                "speed-preset=ultrafast".into(),
+                format!("bitrate={bitrate}"),
+                format!("key-int-max={key_interval}"),
+            ],
+        )
+    } else {
+        (
+            "video/x-raw,format=NV12",
+            vec![
+                "vah264enc".into(),
+                "rate-control=cbr".into(),
+                format!("bitrate={bitrate}"),
+                format!("key-int-max={key_interval}"),
+            ],
+        )
+    };
+
+    // muxer + sink
+    let mut args: Vec<String> = vec![
+        "mpegtsmux".into(),
+        "name=mux".into(),
+        "!".into(),
+        "queue".into(),
+        "!".into(),
+        "fdsink".into(),
+        "fd=1".into(),
+    ];
+
+    // video branch — videorate caps to the target fps so we don't ship at the
+    // monitor's refresh rate (e.g. 180Hz) and pile up frames in the demuxer
+    // queue faster than realtime.
+    args.extend(source.iter().cloned());
+    args.extend([
+        "!".into(),
+        "videorate".into(),
+        "!".into(),
+        framerate_caps,
+        "!".into(),
+        "queue".into(),
+        "!".into(),
+        "videoconvert".into(),
+        "!".into(),
+        raw_format.into(),
+        "!".into(),
+    ]);
+    args.extend(encoder_args);
+    args.extend([
+        "!".into(),
+        "h264parse".into(),
+        "config-interval=-1".into(),
+        "!".into(),
+        "video/x-h264,stream-format=byte-stream,alignment=au".into(),
+        "!".into(),
+        "mux.".into(),
+    ]);
+
+    // audio branch — capture the default sink's MONITOR (system audio out),
+    // not the default source (which is the mic).
+    args.extend([
+        "pulsesrc".into(),
+        audio_device.to_string(),
+        "do-timestamp=true".into(),
+        "!".into(),
+        "queue".into(),
+        "!".into(),
+        "audioconvert".into(),
+        "!".into(),
+        "audioresample".into(),
+        "!".into(),
+        "audio/x-raw,rate=48000,channels=2".into(),
+        "!".into(),
+        "avenc_aac".into(),
+        "bitrate=128000".into(),
+        "!".into(),
+        "aacparse".into(),
+        "!".into(),
+        "mux.".into(),
+    ]);
+
+    args
+}
+
+async fn default_audio_monitor() -> Result<String> {
+    let output = Command::new("pactl")
+        .arg("get-default-sink")
+        .output()
+        .await
+        .context("failed to run `pactl get-default-sink` (install pulseaudio-utils or pipewire-pulse)")?;
+    if !output.status.success() {
+        bail!(
+            "pactl get-default-sink failed: {}",
+            String::from_utf8_lossy(&output.stderr).trim()
+        );
+    }
+    let sink = String::from_utf8(output.stdout)
+        .context("default sink name was not UTF-8")?
+        .trim()
+        .to_string();
+    if sink.is_empty() {
+        bail!("pactl get-default-sink returned no name (is a sound server running?)");
+    }
+    Ok(format!("{sink}.monitor"))
+}

src/host/wayland.rs

@@ -1,8 +1,9 @@
-//! Wayland capture: ashpd ScreenCast portal → PipeWire fd → gst-launch.
+//! Wayland capture: ashpd ScreenCast portal → PipeWire fd → `pipewiresrc`.
-//! Builds the gst pipeline that produces MPEG-TS on stdout, then hands
+//! This module owns only the portal handshake and the source-element args;
-//! that stdout to [`super::serve::Serve`] which handles the HTTP fanout.
+//! the shared encode/mux tail, gst spawn, and serving live in
+//! [`super::pipeline`].
 
-use anyhow::{Context, Result, bail};
+use anyhow::{Context, Result};
 use ashpd::{
     WindowIdentifier,
     desktop::{
@@ -11,64 +12,12 @@ use ashpd::{
     },
 };
 use nix::fcntl::{FcntlArg, FdFlag, fcntl};
-use nix::sys::signal::{Signal, kill};
+use nix::unistd::close;
-use nix::unistd::{Pid, close};
 use std::os::fd::{AsFd, IntoRawFd, OwnedFd, RawFd};
-use std::process::Stdio;
-use std::time::Duration;
-use tokio::process::{Child, Command};
-use tokio::time::timeout;
 
-use super::audio::Routing;
+use super::pipeline::{self, CaptureHandle};
-use super::serve::Serve;
 use crate::cli::HostOpts;
 
-pub struct CaptureHandle {
-    gst: Option<Child>,
-    audio: Option<Routing>,
-    serve: Option<Serve>,
-}
-
-impl CaptureHandle {
-    pub fn local_port(&self) -> u16 {
-        self.serve
-            .as_ref()
-            .expect("serve is always Some until shutdown")
-            .local_port()
-    }
-
-    /// Graceful teardown: SIGTERM gst, give it ~1s to exit, then SIGKILL,
-    /// unload audio routing (if any), then tear down the serve layer.
-    /// The serve reader will see EOF on gst stdout and exit on its own;
-    /// serve.shutdown() is the backstop.
-    pub async fn shutdown(mut self) {
-        if let Some(child) = self.gst.as_mut()
-            && let Some(pid) = child.id()
-        {
-            let _ = kill(Pid::from_raw(pid as i32), Signal::SIGTERM);
-        }
-        if let Some(child) = self.gst.as_mut() {
-            let _ = timeout(Duration::from_millis(1000), child.wait()).await;
-            let _ = child.start_kill();
-        }
-        if let Some(audio) = self.audio.take() {
-            audio.shutdown();
-        }
-        if let Some(serve) = self.serve.take() {
-            serve.shutdown().await;
-        }
-    }
-}
-
-impl Drop for CaptureHandle {
-    fn drop(&mut self) {
-        if let Some(child) = self.gst.as_mut() {
-            let _ = child.start_kill();
-        }
-        // Routing's and Serve's own Drop impls handle the rest.
-    }
-}
-
 pub async fn start(opts: &HostOpts) -> Result<CaptureHandle> {
     // 1. Negotiate the screencast session with the portal.
     let proxy = Screencast::new()
@@ -108,124 +57,23 @@ pub async fn start(opts: &HostOpts) -> Result<CaptureHandle> {
     tracing::info!(node_id, width = w, height = h, "portal handshake complete");
     // The fd is CLOEXEC by default; the gst child needs to inherit it across
     // exec. We then leak it via into_raw_fd so its lifetime spans the spawn,
-    // and close the parent's copy once gst is running.
+    // and close the parent's copy once gst is running (the pipeline's
+    // after_spawn hook below).
     clear_cloexec(&pw_fd)?;
     let raw_fd: RawFd = pw_fd.into_raw_fd();
 
-    // 2. Spawn gst-launch with the full pipeline: video AND audio captured,
+    let source_args = vec![
-    //    encoded, and muxed into MPEG-TS inside gst. Output goes to stdout,
+        "pipewiresrc".to_string(),
-    //    which the serve layer pipes to its HTTP fanout — no demux/remux,
+        format!("fd={raw_fd}"),
-    //    no codec assumptions.
+        format!("path={node_id}"),
-    let key_interval = (opts.framerate * 2).to_string();
+        "do-timestamp=true".to_string(),
-    let bitrate = opts.bitrate.to_string();
+    ];
 
-    // Audio routing activates when either:
+    pipeline::spawn(opts, source_args, move || {
-    // - `opts.app` is set (per-stream rerouting to a per-PID null-sink),
+        // Parent no longer needs the pipewire fd — gst inherited its own copy.
-    // - or `PIXELPASS_AUDIO_VIA_NULL_SINK=1` is set (no app filter, just
+        let _ = close(raw_fd);
-    //   captures everything via the null-sink → useful for development
-    //   and dogfooding the loopback path before app filtering is picked).
-    let routing_requested =
-        opts.app.is_some() || std::env::var_os("PIXELPASS_AUDIO_VIA_NULL_SINK").is_some();
-    let audio_routing = if routing_requested {
-        Some(
-            Routing::start(opts)
-                .await
-                .context("audio routing setup failed")?,
-        )
-    } else {
-        None
-    };
-    let audio_device = if let Some(r) = &audio_routing {
-        format!("device={}.monitor", r.sink_name())
-    } else {
-        let default = default_audio_monitor().await?;
-        format!("device={default}")
-    };
-    let mut gst_cmd = Command::new("gst-launch-1.0");
-    gst_cmd
-        .args([
-            // muxer + sink
-            "mpegtsmux",
-            "name=mux",
-            "!",
-            "queue",
-            "!",
-            "fdsink",
-            "fd=1",
-            // video branch — videorate caps to 30fps so we don't ship at the
-            // monitor's refresh rate (e.g. 180Hz) and pile up frames in mpv's
-            // demuxer queue faster than realtime.
-            "pipewiresrc",
-            &format!("fd={raw_fd}"),
-            &format!("path={node_id}"),
-            "do-timestamp=true",
-            "!",
-            "videorate",
-            "!",
-            &format!("video/x-raw,framerate={}/1", opts.framerate),
-            "!",
-            "queue",
-            "!",
-            "videoconvert",
-            "!",
-            "video/x-raw,format=NV12",
-            "!",
-            "vah264enc",
-            "rate-control=cbr",
-            &format!("bitrate={bitrate}"),
-            &format!("key-int-max={key_interval}"),
-            "!",
-            "h264parse",
-            "config-interval=-1",
-            "!",
-            "video/x-h264,stream-format=byte-stream,alignment=au",
-            "!",
-            "mux.",
-            // audio branch — capture the default sink's MONITOR (system audio
-            // out), not the default source (which is the mic).
-            "pulsesrc",
-            &audio_device,
-            "do-timestamp=true",
-            "!",
-            "queue",
-            "!",
-            "audioconvert",
-            "!",
-            "audioresample",
-            "!",
-            "audio/x-raw,rate=48000,channels=2",
-            "!",
-            "avenc_aac",
-            "bitrate=128000",
-            "!",
-            "aacparse",
-            "!",
-            "mux.",
-        ])
-        .stdin(Stdio::null())
-        .stdout(Stdio::piped())
-        .stderr(Stdio::inherit());
-    if std::env::var_os("PIXELPASS_GST_DEBUG").is_some() {
-        gst_cmd.env("GST_DEBUG", "3");
-    }
-    let mut gst = gst_cmd.spawn().context("failed to spawn gst-launch-1.0")?;
-    // Parent no longer needs the pipewire fd — gst inherited its own copy.
-    let _ = close(raw_fd);
-
-    let gst_stdout = gst
-        .stdout
-        .take()
-        .context("gst-launch-1.0 stdout pipe unavailable")?;
-
-    // 3. Hand stdout to the serve layer, which binds the localhost HTTP
-    //    listener and runs the broadcast fanout.
-    let serve = Serve::bind(gst_stdout).await?;
-
-    Ok(CaptureHandle {
-        gst: Some(gst),
-        audio: audio_routing,
-        serve: Some(serve),
     })
+    .await
 }
 
 fn clear_cloexec(fd: &impl AsFd) -> Result<()> {
@@ -235,25 +83,3 @@ fn clear_cloexec(fd: &impl AsFd) -> Result<()> {
     fcntl(fd.as_fd(), FcntlArg::F_SETFD(flags)).context("F_SETFD on pipewire fd")?;
     Ok(())
 }
-
-async fn default_audio_monitor() -> Result<String> {
-    let output = Command::new("pactl")
-        .arg("get-default-sink")
-        .output()
-        .await
-        .context("failed to run `pactl get-default-sink` (install pulseaudio-utils or pipewire-pulse)")?;
-    if !output.status.success() {
-        bail!(
-            "pactl get-default-sink failed: {}",
-            String::from_utf8_lossy(&output.stderr).trim()
-        );
-    }
-    let sink = String::from_utf8(output.stdout)
-        .context("default sink name was not UTF-8")?
-        .trim()
-        .to_string();
-    if sink.is_empty() {
-        bail!("pactl get-default-sink returned no name (is a sound server running?)");
-    }
-    Ok(format!("{sink}.monitor"))
-}

src/host/x11.rs

@@ -0,0 +1,94 @@
+//! X11 capture: `ximagesrc` → the shared encode/mux tail in [`super::pipeline`].
+//! Unlike Wayland there's no portal and no fd hand-off — `ximagesrc` opens its
+//! own X connection from `$DISPLAY`. The whole root window is captured by
+//! default; `--window` resolves a single window's XID via an `xwininfo`
+//! click-picker. The ticket is the access control, so capture starts silently
+//! when the first viewer connects (no host-side consent prompt).
+
+use anyhow::{Context, Result, bail};
+use tokio::process::Command;
+use x11rb::connection::Connection;
+use x11rb::protocol::xproto::ConnectionExt;
+
+use super::pipeline::{self, CaptureHandle};
+use crate::cli::HostOpts;
+
+pub async fn start(opts: &HostOpts) -> Result<CaptureHandle> {
+    let xid = if opts.window {
+        Some(pick_window().await?)
+    } else {
+        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:#}"),
+    }
+
+    let mut source_args = vec![
+        "ximagesrc".to_string(),
+        // Full frames (no damage regions) to avoid partial-update artifacts;
+        // use-damage=true is a later CPU optimization. show-pointer matches
+        // Wayland's CursorMode::Embedded.
+        "use-damage=false".to_string(),
+        "show-pointer=true".to_string(),
+    ];
+    if let Some(xid) = xid {
+        source_args.push(format!("xid={xid}"));
+    }
+
+    // X11 has no leaked fd to clean up, so the post-spawn hook is a no-op.
+    pipeline::spawn(opts, source_args, || {}).await
+}
+
+/// Run `xwininfo` and let the user click the window they want to share, then
+/// parse the `Window id: 0x…` line out of its output. Returns the numeric XID.
+async fn pick_window() -> Result<u32> {
+    eprintln!("[pixelpass] click the window you want to share…");
+    let output = Command::new("xwininfo")
+        .output()
+        .await
+        .context("failed to run `xwininfo` (install xorg-xwininfo)")?;
+    if !output.status.success() {
+        bail!(
+            "xwininfo failed: {}",
+            String::from_utf8_lossy(&output.stderr).trim()
+        );
+    }
+    let text = String::from_utf8_lossy(&output.stdout);
+    for line in text.lines() {
+        // e.g.  "xwininfo: Window id: 0x3a00007 \"xterm\""
+        if let Some((_, rest)) = line.split_once("Window id: ") {
+            let token = rest.split_whitespace().next().unwrap_or("");
+            let hex = token.strip_prefix("0x").unwrap_or(token);
+            if let Ok(xid) = u32::from_str_radix(hex, 16) {
+                return Ok(xid);
+            }
+        }
+    }
+    bail!("could not parse a window id from xwininfo output");
+}
+
+/// Read pixel dimensions: the selected window's geometry when `--window` was
+/// used, otherwise the root window of the screen named by `$DISPLAY`.
+fn read_geometry(xid: Option<u32>) -> Result<(u16, u16)> {
+    let (conn, screen_num) =
+        x11rb::connect(None).context("could not connect to the X server (is DISPLAY set?)")?;
+    match xid {
+        Some(id) => {
+            let geo = conn
+                .get_geometry(id)
+                .context("GetGeometry request failed")?
+                .reply()
+                .context("GetGeometry reply failed")?;
+            Ok((geo.width, geo.height))
+        }
+        None => {
+            let screen = &conn.setup().roots[screen_num];
+            Ok((screen.width_in_pixels, screen.height_in_pixels))
+        }
+    }
+}