Clip Playback
Open a clip from the Timeline to enter the playback view. This is where you watch recorded clips, review AI analysis frame-by-frame, and interact with the agent about specific moments.
Video Player
Standard video controls:
- Play / Pause — spacebar or click the player
- Seek bar — scrub to any point in the clip
- Volume control
- Playback speed — 0.25×, 0.5×, 1×, 1.5×, 2×
- Full screen toggle — escape to exit
The player supports H.264 and H.265 video codecs. If a clip uses H.265 and your system doesn't support it natively, Aegis transcodes on-the-fly automatically.
Playback Speed Use Cases
| Speed | When to Use |
|---|---|
| 0.25× | Examining fast events frame-by-frame — someone running, a car passing quickly |
| 0.5× | Detailed review of a specific moment — reading a license plate, identifying a visitor |
| 1× | Normal playback — reviewing a clip at real speed |
| 1.5× | Scanning through a long clip quickly |
| 2× | Fast-forwarding through uneventful sections |
AI Analysis Panel
If video analysis is enabled, an analysis panel appears alongside the player. This shows the VLM's description of each frame as it's analyzed.
How Frame Analysis Works
- The VLM receives a frame from the clip
- The configured analysis prompt guides what the model looks for
- The model generates a natural-language description of that frame
- The description appears in the analysis panel, timestamped to match the video
Analysis Panel Features
- Real-time descriptions — update as the video plays, synchronized with frame timestamps
- Timestamped entries — each description shows the exact time in the clip it corresponds to
- Persistent storage — analysis results are stored with the clip, so you only need to analyze once
- Independent scrolling — scroll through analysis results independently of video playback to review earlier descriptions
Starting Analysis
If auto-start is disabled in Video Analysis Settings, click the Start Analysis button to begin. Analysis runs frame-by-frame on the selected VLM and produces natural language descriptions of what's happening.
If the clip was previously analyzed, stored descriptions are loaded immediately — no need to re-run analysis.
Clip Information
The clip info panel shows metadata about the current clip:
- Camera name — which camera recorded this clip
- Recording timestamp — when the clip was captured (date and time)
- Duration — length of the clip
- File size — how much disk space the clip uses
- AI-generated description — summary text if the clip has been analyzed. This is the same description that appears in the Timeline and is used by event handlers for alert evaluation.
- Favorite status — toggle the heart icon to mark/unmark this clip as a favorite
Asking About a Clip
While viewing a clip, you can switch to the Aegis Home chat and ask questions about it. The agent has context about the currently viewed clip and can answer questions like:
- "What's happening in this clip?"
- "Is this the same person from earlier?"
- "Describe everything you see at the 0:15 mark."
- "How many people are visible?"
- "What type of vehicle is in the driveway?"
- "Does this look suspicious?"
The agent combines the VLM analysis, clip metadata, its memory, and any relevant context from other cameras to give you an informed answer. This is significantly more powerful than re-watching footage yourself — the AI can cross-reference information across cameras and time periods.
Navigation
Use the Previous (←) and Next (→) buttons to navigate between clips from the same camera without returning to the Timeline. This lets you quickly scan through a sequence of events from one camera — for example, reviewing all motion clips from the front door over the past few hours.
Exporting and Sharing
Clips are stored as standard video files on your local disk. You can locate any clip through the Storage settings to access the raw video file for external use — sharing with neighbors, providing to authorities, or backing up.