Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion api/fishjam-server
72 changes: 72 additions & 0 deletions docs/tutorials/agents.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -206,6 +206,78 @@ You can interrupt the currently played audio chunk. See the example below.

</Tabs>

### Making the Agent see

Agents can also request video frames (JPEG images) from peers' video tracks.
Unlike audio, which streams continuously, video frames must be explicitly requested and arrive asynchronously.

:::important
Video frame capture is rate-limited to one frame per second per track.
:::

<Tabs groupId="language">
<TabItem value="ts" label="TypeScript">

```ts
// @noErrors
import { RoomId, FishjamClient, TrackId } from '@fishjam-cloud/js-server-sdk';

const fishjamId = '';
const managementToken = '';
const fishjamClient = new FishjamClient({ fishjamId, managementToken });
const room = await fishjamClient.createRoom();
const { agent } = await fishjamClient.createAgent(room.id, {});
const trackId: TrackId = '' as TrackId;

// ---cut---
import type { IncomingTrackImage } from '@fishjam-cloud/js-server-sdk';

// Listen for incoming video frames
agent.on('trackImage', (message: IncomingTrackImage) => {
const { contentType, data } = message;
// process the image data
});

// Request a frame periodically
setInterval(() => {
// [!code highlight:1]
agent.captureImage(trackId);
}, 1000);

```

</TabItem>

<TabItem value="python" label="Python">

```python
import asyncio

from fishjam import FishjamClient
from fishjam.agent import IncomingTrackImage

fishjam_client = FishjamClient(fishjam_id, management_token)

agent = fishjam_client.create_agent(room_id)

async with agent.connect() as session:
# Request a frame
# [!code highlight:1]
await session.capture_image(track_id)

# Captured frames arrive as IncomingTrackImage messages
async for message in session.receive():
match message:
case IncomingTrackImage() as msg if msg.track_id == track_id:
data = msg.data
# process the image data
pass
```

</TabItem>

</Tabs>

### Disconnecting

After you're done using an agent, you can disconnect it from the room.
Expand Down