How to fill the screen when a podcast guest has no camera

Last updated 2026-09-01 · 8 min read

What is the short version?

One guest joins with the camera off and the episode becomes unpublishable as video — half of it is a black rectangle with a name in it. A photograph in a branded frame, appearing only while they speak, fixes it entirely.

1. Go to Vizard Agent and upload the recording.
2. Send a photograph of the person who had no camera.
3. Give it your intro and outro audio if you have them.

What do you need before you start?

The recording as it came out and one decent photograph. Vizard Agent finds the camera-off stretches itself, so you do not need to note the timestamps — which is fortunate, because a camera that drops in and out does so more often than anyone remembers.

• The recording. Unedited, cameras as they were.
• The photograph. Reasonably high resolution.
• The intro and outro. Audio files, if separate.
• Your branding. Colours and typeface, or a previous episode.
• The format. Landscape for a channel, vertical for clips.

What do you type into Vizard Agent?

Say who has no video and hand over the photograph. Vizard Agent works out when to show it, so what it needs from you is who the photograph belongs to and what the frame should look like around it.

Prompt

Variants worth knowing:

• Say whose photo it is. So it appears at the right times.
• Send the intro and outro as audio. They get levelled to match.
• Point at a previous episode. The branding comes from it.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real podcast where one participant had no camera. Step eight is the one that saves you an evening: it finds every camera-off and frozen stretch itself rather than asking you where they are.

1. Checks the workspace and downloads your files to see what they are.
2. Samples frames from the recording to see what is actually on screen.
3. Reviews those frames, then pulls a full-size one to look closely.
4. Starts the transcript running in the background.
5. Looks at the photograph you supplied.
6. Scans the whole recording for stretches where a camera is off or frozen.
7. Checks the transcript against what it found.
8. Checks the opening and the ending of the recording and reviews those frames.
9. Downloads your intro and outro audio.
10. Measures the levels and the silences across all three audio sources.
11. Measures the loudness and the lead-in silence so the joins do not jump.
12. Reads the intro and outro transcripts.
13. Samples the brand colours, the fonts and the dead air.
14. Locates the font files and builds the branded layout background.
15. Reviews the layout design before assembling anything.

Step ten matters more than it sounds. An intro recorded separately is almost never at the same level as the conversation, and a listener notices a jump at the join long before they notice anything else about the edit.

What does the result look like?

The episode this page is written from came back as a full edit: a branded layout background, the guest's photograph filling the frame for the stretches where their camera was off, an intro and an outro levelled to match the conversation, and the dead air trimmed.

The photograph is not on screen throughout. It appears for the specific stretches where there is nothing else to show, which is what stops it feeling like a placeholder and starts it feeling like a design decision.

When does this not work well?

A photograph is a completely static object sitting on a screen the audience expects to be moving, and there is a real limit to how long anybody will keep looking at one. These are the situations where a single still image stops being enough on its own.

• Very long audio-only stretches. One photo cannot hold ten minutes.
• Low-resolution photographs. They show at full frame.
• Cameras dropping constantly. The switching becomes distracting.
• No branding to work from. The frame looks generic.
• Nothing else to cut to. Consider a waveform or captions as well.

How do you fix a result that came back wrong?

Say which stretch of the episode is wrong and what is wrong with it. Vizard Agent keeps the map of every camera-off period it found, the built layout and all the measured audio levels, so a changed photograph or a different detection threshold is a re-render rather than another full scan.

• "It appears too often." Threshold raised so brief drops are ignored.
• "Use a different photo." Swapped into the same layout.
• "The intro is louder than the show." Re-levelled to match.

How does Vizard Agent compare to doing it yourself?

By hand this means watching the whole episode with a notepad, writing down every moment a camera went off, then building a layout and placing the photograph at each one. Vizard Agent scans for them and measures the audio joins at the same time.

Common questions

Do I need the timestamps? No. Vizard Agent scans the recording and finds them itself.

What if the camera froze rather than turned off? It catches both. Vizard Agent looks for frozen picture as well as absent picture.

Can it use several photos? Yes. Vizard Agent can rotate through them so one image is not held too long.

What about my other guests? Only the ones without video get the treatment. Vizard Agent leaves the rest as filmed.

Can it match my branding? Yes. Vizard Agent samples the colours and typefaces, or takes them from a previous episode.

Will the intro sound right? Yes. Vizard Agent measures all three sources and levels them together.

Can it add captions too? Yes, and it helps here. Captions give the eye something to follow during the photo stretches.

What about a waveform instead? That works as well. Vizard Agent can animate one under or beside the photograph.

Does it trim the dead air? Yes. Vizard Agent measures the silences and tightens them.

Can I get vertical clips from it? Yes. Vizard Agent re-frames the same layout for a vertical crop.

Is a photograph really enough? For a few minutes at a time, yes — audiences accept it because they understand what happened. What they will not sit through is a black rectangle with a name on it.

Can it handle a guest who joined late? Yes. Vizard Agent treats the missing period the same way.

Does it check the result? Yes. Vizard Agent reviews the branded layout and the rendered frames before committing to the full episode.

Will it detect a briefly frozen picture? Yes, and you can set how brief. Vizard Agent ignores momentary drops if you raise the threshold.