Why does the picture-in-picture window cover a face in dual camera video?
The inset is not smart about what is under it. Picture-in-picture places a small window in one of four corners of the frame, and it simply sits on top of whatever the main camera happens to be showing in that corner — a face included. If a person in the main shot is standing, sitting or walking through that corner, the inset covers them for as long as it stays there.
This is not a bug to report or a setting hiding somewhere. It is a direct consequence of how picture-in-picture works: one camera fills the canvas, the other is drawn on top of a fixed region of it. The fix is to know where that region is and keep the parts of the shot that matter out of it — or switch to a layout that does not have this problem at all.
Why the inset always wins
In DualCam, picture-in-picture is drawn last, after the main camera frame has been composited. So whatever the small window overlaps — a face, a hand, a sign someone is holding — is replaced pixel-for-pixel by the inset for as long as it sits there. There is no transparency and no attempt to detect what is underneath; the app has no way to know that a face happens to be under that corner right now.
The inset also carries a small margin so it never touches the very edge of the frame, which slightly enlarges the area it blocks compared to what you might expect from its stated size alone. A “small” inset in a corner still removes a real, fixed rectangle of the main image for the whole time it is there.
It is baked into the frame — you cannot move it after recording
DualCam composites both cameras into the recorded frame in real time, so the picture-in-picture window is not a separate layer you can move in an editor afterwards. Once a take is recorded, whatever the inset covered is gone from the file the same way a sticker glued over a photo removes what is underneath it — there is no original frame left to uncover.
That is why this problem is worth solving before and during a take rather than after. Most other things about a shot can be salvaged in an editor; a face hidden behind the inset cannot.
Check the frame before you press record
Before you start, look at the four corners of what the main camera actually sees, not just the centre of the frame. If someone is going to be standing, sitting or walking near a specific corner during the shot, do not put the inset there. This matters most in loosely blocked scenes — a room tour, an interview where the other person can shift in their seat, a walk down a street — where you cannot fully predict where people end up.
The inset can be dragged to any of the four corners and resized, so treat corner choice as a framing decision, the same way you would decide where to stand a tripod. The corner that is empty in your shot, not the one that looks tidiest in the abstract, is the right one.
Fix it live, because the preview is the recording
Because the viewfinder shows exactly what is being written to the file, you do not have to guess whether the inset is covering something — you can see it happening in real time and act. Layout, the picture-in-picture corner, its size and its shape can all be changed mid-recording in DualCam, so if a person walks into the corner the inset occupies partway through a take, drag the window to a different corner rather than losing the shot.
A smaller inset covers less, but it is not a substitute for choosing the right corner — a small window sitting directly on a face still hides the face. Treat resizing as a secondary adjustment, made after the corner is already right.
When to stop using picture-in-picture altogether
Some shots are picture-in-picture in name only. A two-person interview where both faces need to stay visible for the whole conversation, or a scene where people move around too much for any fixed corner to stay clear, is better served by a top-and-bottom or left-and-right split, because both camera feeds get equal, dedicated space and neither one can be drawn over by the other.
The trade-off is that a split crops both source frames to fit the canvas, which picture-in-picture does not do to its main frame. Neither layout is free of compromise; the question is only which compromise fits what you are filming — an occluded face is usually the worse of the two.
Common follow-up questions
Can I make the inset transparent so it does not fully block anything?
No. The inset is a solid rectangle drawn on top of the main frame, and there is no opacity control, because a transparent inset would mean blending two live camera feeds pixel by pixel, which is a different feature from picture-in-picture.
Does the inset always cover the same amount of the frame?
The area is fixed by whichever of the three sizes you pick, plus a small margin from the edge. It does not grow or shrink based on what is happening underneath it.
If I notice too late, can I re-record just that section?
No — DualCam records one continuous take into one file, so there is no partial re-record. If the inset covered something important, the fix has to happen by stopping and starting a new take, or by catching it live and moving the window before it matters.
Want to just do this?
DualCam records the iPhone front and back cameras at the same time and writes one finished MP4 while you shoot. Free, no account, no ads, nothing leaves the phone.