Merge pull request #4370 from pipecat-ai/aleix/daily-screen-video-destination

feat(daily): support screenVideo destination and configurable camera send settings

changelog/4370.added.2.md

@@ -0,0 +1,12 @@
+- Added `camera_out_send_settings` to `DailyParams`. This dict is passed verbatim to the Daily client's camera publishing settings, allowing applications to fully control encoding, codec, bitrate, and framerate.
+
+  ```python
+  params = DailyParams(
+      camera_out_send_settings={
+          "maxQuality": "high",
+          "encodings": {
+              "high": {"maxBitrate": 2_000_000, "maxFramerate": 30}
+          },
+      },
+  )
+  ```

changelog/4370.added.md

@@ -0,0 +1,16 @@
+- Added support for Daily's built-in `screenVideo` destination in `DailyTransport`. When `"screenVideo"` is included in `video_out_destinations` transport parameter, a dedicated screen video track is created at join time and frames with `transport_destination="screenVideo"` are routed to it.
+
+  ```python
+  params = DailyParams(
+        video_out_enabled=True,
+        video_out_is_live=True,
+        video_out_width=1280,
+        video_out_height=720,
+        video_out_destinations=["screenVideo"]
+  )
+
+  ...
+
+  frame = OutputImageRawFrame(...)
+  frame.transport_destination = "screenVideo"
+  ```

changelog/4370.deprecated.md

@@ -0,0 +1 @@
+- Deprecated `TransportParams.video_out_bitrate` for the Daily transport. Use `DailyParams.camera_out_send_settings` instead to configure camera publishing encodings (bitrate, framerate, codec, etc.).

src/pipecat/transports/base_output.py

@@ -90,6 +90,18 @@ class BaseOutputTransport(FrameProcessor):
         # it.
         self._media_senders: dict[Any, BaseOutputTransport.MediaSender] = {}
 
+        if params.video_out_bitrate is not None:
+            import warnings
+
+            with warnings.catch_warnings():
+                warnings.simplefilter("always")
+                warnings.warn(
+                    "Transport parameter `video_out_bitrate` is deprecated and will be removed in a future "
+                    "version. Use provider specific settings instead.",
+                    DeprecationWarning,
+                    stacklevel=2,
+                )
+
     @property
     def sample_rate(self) -> int:
         """Get the current audio sample rate.

src/pipecat/transports/base_transport.py

@@ -47,7 +47,7 @@ class TransportParams(BaseModel):
         video_out_is_live: Enable real-time video output streaming.
         video_out_width: Video output width in pixels.
         video_out_height: Video output height in pixels.
-        video_out_bitrate: Video output bitrate in bits per second.
+        video_out_bitrate: [DEPRECATED] Video output bitrate in bits per second.
         video_out_framerate: Video output frame rate in FPS.
         video_out_color_format: Video output color format string.
         video_out_codec: Preferred video codec for output (e.g., 'VP8', 'H264', 'H265').
@@ -76,7 +76,7 @@ class TransportParams(BaseModel):
     video_out_is_live: bool = False
     video_out_width: int = 1024
     video_out_height: int = 768
-    video_out_bitrate: int = 800000
+    video_out_bitrate: int | None = None
     video_out_framerate: int = 30
     video_out_color_format: str = "RGB"
     video_out_codec: str | None = None

src/pipecat/transports/daily/transport.py

@@ -325,6 +325,7 @@ class DailyParams(TransportParams):
         api_key: Daily API authentication key.
         audio_in_user_tracks: Receive users' audio in separate tracks
         camera_out_enabled: Whether to enable the main camera output track.
+        camera_out_send_settings: Camera output track publishing settings.
         custom_audio_track_params: Per-destination configuration for custom audio tracks.
         custom_video_track_params: Per-destination configuration for custom video tracks.
         dialin_settings: Optional settings for dial-in functionality.
@@ -337,6 +338,7 @@ class DailyParams(TransportParams):
     api_key: str = ""
     audio_in_user_tracks: bool = True
     camera_out_enabled: bool = True
+    camera_out_send_settings: dict[str, Any] | None = None
     custom_audio_track_params: Mapping[str, DailyCustomAudioTrackParams] | None = None
     custom_video_track_params: Mapping[str, DailyCustomVideoTrackParams] | None = None
     dialin_settings: DailyDialinSettings | None = None
@@ -555,6 +557,7 @@ class DailyTransportClient(EventHandler):
         self._camera_track: DailyVideoTrack | None = None
         self._microphone_track: DailyAudioTrack | None = None
         self._custom_audio_tracks: dict[str, DailyAudioTrack] = {}
+        # Custom video tracks will also include `screenVideo`.
         self._custom_video_tracks: dict[str, DailyVideoTrack] = {}
 
     def _speaker_name(self):
@@ -666,19 +669,19 @@ class DailyTransportClient(EventHandler):
         self._client.update_publishing(publishing)
 
     async def register_video_destination(self, destination: str):
-        """Register a custom video destination for multi-track output.
+        """Register a video destination for multi-track output.
+
+        Built-in destination ("camera") is configured at join time so it's
+        skipped here.
 
         Args:
             destination: The destination identifier to register.
+
         """
-        params = (self._params.custom_video_track_params or {}).get(destination)
-        self._custom_video_tracks[destination] = await self.add_custom_video_track(
-            destination, params=params
-        )
-        publishing: dict[str, Any] = {"customVideo": {destination: True}}
-        if params and params.send_settings:
-            publishing["customVideo"][destination] = {"sendSettings": params.send_settings}
-        self._client.update_publishing(publishing)
+        if destination == "screenVideo":
+            await self._register_screen_video_destination()
+        else:
+            await self._register_custom_video_destination(destination)
 
     async def write_audio_frame(self, frame: OutputAudioRawFrame) -> bool:
         """Write an audio frame to the appropriate audio track.
@@ -719,7 +722,7 @@ class DailyTransportClient(EventHandler):
         """
         destination = frame.transport_destination
         video_source: CustomVideoSource | None = None
-        if not destination and self._camera_track:
+        if (not destination or destination == "camera") and self._camera_track:
             video_source = self._camera_track.source
         elif destination and destination in self._custom_video_tracks:
             track = self._custom_video_tracks[destination]
@@ -795,7 +798,11 @@ class DailyTransportClient(EventHandler):
                 self._callback_task_handler(self._video_queue),
                 f"{self}::video_callback_task",
             )
-        if self._params.video_out_enabled and not self._camera_track:
+        if (
+            self._params.video_out_enabled
+            and self._params.camera_out_enabled
+            and not self._camera_track
+        ):
             video_source = CustomVideoSource(
                 self._params.video_out_width,
                 self._params.video_out_height,
@@ -804,7 +811,11 @@ class DailyTransportClient(EventHandler):
             video_track = CustomVideoTrack(video_source)
             self._camera_track = DailyVideoTrack(source=video_source, track=video_track)
 
-        if self._params.audio_out_enabled and not self._microphone_track:
+        if (
+            self._params.audio_out_enabled
+            and self._params.microphone_out_enabled
+            and not self._microphone_track
+        ):
             logger.debug(
                 f"Creating custom audio source, auto silence {self._params.audio_out_auto_silence}"
             )
@@ -899,28 +910,19 @@ class DailyTransportClient(EventHandler):
                 },
                 "publishing": {
                     "camera": {
-                        "sendSettings": {
-                            "maxQuality": "low",
-                            **(
-                                {"preferredCodec": self._params.video_out_codec}
-                                if self._params.video_out_codec
-                                else {}
-                            ),
-                            "encodings": {
-                                "low": {
-                                    "maxBitrate": self._params.video_out_bitrate,
-                                    "maxFramerate": self._params.video_out_framerate,
-                                }
-                            },
-                        }
+                        "isPublishing": camera_enabled,
+                        "sendSettings": self._params.camera_out_send_settings
+                        if self._params.camera_out_send_settings
+                        else {},
                     },
                     "microphone": {
+                        "isPublishing": microphone_enabled,
                         "sendSettings": {
                             "channelConfig": "stereo"
                             if self._params.audio_out_channels == 2
                             else "mono",
                             "bitrate": self._params.audio_out_bitrate,
-                        }
+                        },
                     },
                 },
             },
@@ -1317,23 +1319,17 @@ class DailyTransportClient(EventHandler):
         """
         future = self._get_event_loop().create_future()
 
-        width = params.width if params else self._params.video_out_width
-        height = params.height if params else self._params.video_out_height
-        color_format = params.color_format if params else self._params.video_out_color_format
-
-        video_source = CustomVideoSource(width, height, color_format)
-
-        video_track = CustomVideoTrack(video_source)
+        video_track = self._create_video_track(params)
 
         self._client.add_custom_video_track(
             track_name=track_name,
-            video_track=video_track,
+            video_track=video_track.track,
             completion=completion_callback(future),
         )
 
         await future
 
-        return DailyVideoTrack(source=video_source, track=video_track)
+        return video_track
 
     async def remove_custom_video_track(self, track_name: str) -> CallClientError | None:
         """Remove a custom video track.
@@ -1344,6 +1340,8 @@ class DailyTransportClient(EventHandler):
         Returns:
             error: An error description or None.
         """
+        if track_name == "screenVideo":
+            return
         future = self._get_event_loop().create_future()
         self._client.remove_custom_video_track(
             track_name=track_name,
@@ -1424,6 +1422,56 @@ class DailyTransportClient(EventHandler):
         )
         return await future
 
+    async def _create_video_track(
+        self,
+        params: DailyCustomVideoTrackParams | None = None,
+    ) -> DailyVideoTrack:
+        """Create a video track for the given parameters."""
+        future = self._get_event_loop().create_future()
+
+        width = params.width if params else self._params.video_out_width
+        height = params.height if params else self._params.video_out_height
+        color_format = params.color_format if params else self._params.video_out_color_format
+
+        video_source = CustomVideoSource(width, height, color_format)
+
+        video_track = CustomVideoTrack(video_source)
+
+        return DailyVideoTrack(source=video_source, track=video_track)
+
+    async def _register_screen_video_destination(self):
+        """Register screen video destination track."""
+        params = (self._params.custom_video_track_params or {}).get("screenVideo")
+
+        video_track = await self._create_video_track(params)
+        self._custom_video_tracks["screenVideo"] = video_track
+
+        # screenVideo inpupts settings.
+        inputs: dict[str, Any] = {
+            "screenVideo": {
+                "isEnabled": True,
+                "settings": {"customTrack": {"id": video_track.track.id}},
+            }
+        }
+        self._client.update_inputs(inputs)
+
+        # screenVideo publishing settings.
+        publishing: dict[str, Any] = {"screenVideo": True}
+        if params and params.send_settings:
+            publishing["screenVideo"] = {"sendSettings": params.send_settings}
+        self._client.update_publishing(publishing)
+
+    async def _register_custom_video_destination(self, destination: str):
+        """Register a custom video destination for multi-track output."""
+        params = (self._params.custom_video_track_params or {}).get(destination)
+        self._custom_video_tracks[destination] = await self.add_custom_video_track(
+            destination, params=params
+        )
+        publishing: dict[str, Any] = {"customVideo": {destination: True}}
+        if params and params.send_settings:
+            publishing["customVideo"][destination] = {"sendSettings": params.send_settings}
+        self._client.update_publishing(publishing)
+
     #
     #
     # Daily (EventHandler)