Workflow guide
An inset that clarifies instead of cluttering
Picture-in-picture shows two things at once — a reaction over gameplay, a face over slides, a demo over an interview. It fails when the inset covers the action or fights for sound. This guide covers sizing, placement, and audio so both layers earn their pixels.
From the Mellius product team · Updated September 28, 2026
Conspiracy theorists, you may hate this
AI Director
›
Just type what you need
- ›Remove all silent parts
- ›Add bold kinetic captions
- ›Only have Bill Ackman speaking
- ›Cut to highlights only
- ›Add cinematic transitions
- ›Match cuts to the beat
The short answer
Place the main story full-frame and the inset at fifteen to thirty percent of frame width in a corner with a subtle border and drop shadow, cut or move it when it covers key action, balance audio so one layer leads, and animate entrances over six to twelve frames.
Both layers stay readable
Text and faces in each layer remain legible at the delivery size, including phones.
Nothing important is covered
The inset never hides captions, graphics, faces, or the moment the viewer came to see.
One audio leader
A single layer carries the mix at a time while the other ducks or goes silent.
Size and place the inset deliberately
Size the inset between fifteen and thirty percent of frame width: smaller vanishes on phones, larger splits attention and covers the main story. Anchor it to a corner on the rule-of-thirds intersection opposite the main subject's gaze, and add a two to four pixel border with a soft drop shadow so it separates from busy backgrounds. Mirror the inset when the two subjects should face each other across the frame, which reads as conversation rather than collision.
Plan movement before animating. Slide or scale the inset in over six to twelve frames with an ease-out curve, and move it to another corner — or cut it away — the moment it covers slides, captions, or gameplay UI. Common pitfall: parking a webcam box over the exact corner where platform UI, subtitles, or kill feeds live. Check the composite inside a 9:16-safe overlay if the video ships to Shorts or TikTok.
- Keep insets at fifteen to thirty percent of frame width.
- Anchor opposite the main gaze with a subtle border.
- Move or cut the inset before it covers key action.
Balance audio and switch focus cleanly
Decide which layer leads and mix the other fifteen to twenty decibels down or mute it outright — two full-volume sources guarantee mud. When focus swaps, dip the outgoing layer first, then raise the incoming one over a quarter second so the handoff feels motivated. Keep room tone or game audio low under commentary rather than cutting to total silence, which reads as a technical fault.
Use the inset as punctuation, not wallpaper. Bring it in for reactions, demos, and proof, then clear it for full-frame emphasis at climaxes. In Mellius you can ask to add a picture-in-picture layer over a named section and reposition it conversationally, with both clips staying editable and every move reversible on the timeline.
- One audio leader at a time; duck or mute the other.
- Cross handoffs over a quarter second, never hard-swap.
- Clear the inset for full-frame climaxes.
Compare the workflows
| Layout | Best fit | Trade-off |
|---|---|---|
| Corner inset | Reactions, webcams, and demos over main content | Always risks covering something important |
| Side-by-side split | Comparisons and interviews needing equal weight | Shrinks both layers on vertical screens |
| Full-frame switching | Maximum clarity for each moment | Loses the simultaneity that defines PiP |
A practical sequence
Put it into practice
- 01
Stack the layers
Put the main story full-frame, add the inset above it, and size it to fifteen to thirty percent width.
- 02
Place and style
Anchor a clear corner, add a thin border and shadow, and check coverage against captions and UI.
- 03
Mix and animate
Pick one audio leader, duck the other layer, and animate entrances over six to twelve frames.
Frequently asked questions
What size should picture-in-picture be?
Fifteen to thirty percent of frame width suits most delivery. Below fifteen percent faces and text collapse on phones. Above thirty percent the inset competes with the main story instead of supporting it, so switch to side-by-side or full-frame cuts instead.
Where should I place the inset?
Anchor it to the corner opposite the main subject's gaze, on a rule-of-thirds intersection. Then check it against captions, graphics, gameplay HUDs, and platform interface overlays. Move corners between sections when the action shifts underneath.
How do I handle audio with two videos playing?
Choose one leader and duck the other fifteen to twenty decibels or mute it. Crossfade leadership over about a quarter second when focus swaps. Never run two dialogue sources at full volume — the result is always mud.
Continue reading
Primary sources
Product capabilities change. These official references support the comparisons above.
Describe the edit. Keep the timeline.
Mellius turns conversational direction into cuts, captions, graphics, motion, color, and sound you can still inspect and change.
Start editing free