Why does your face look distorted in the picture-in-picture window?
Nothing about compositing two cameras into one frame is doing this to you. A face held close to almost any phone lens comes out a little wider, with the nose and forehead pushed forward and the ears pulled back — a well-known effect of perspective in wide-angle lenses, present in a normal selfie taken at arm’s length and worse the closer the lens gets. Picture-in-picture does not introduce this distortion; it just gives you a smaller, more critical stage to notice it on.
The instinct that makes it worse is a reasonable one: the inset window looks small on screen, so it feels like the face inside it should be brought closer to “fill” it properly. That instinct is exactly backwards for how the inset actually works, and understanding why fixes the problem without touching a single setting.
A lens problem, not a compositing problem
Perspective distortion comes from distance, not from any particular camera or app. The closer a subject sits to the lens, the more the parts nearest the lens — usually the nose and forehead in a face — are magnified relative to the parts further away, like the ears and the sides of the head. Move back, and the difference between “near” and “far” parts of the face shrinks, so the distortion shrinks with it. This is true of the front camera on any iPhone, in any app, whether or not a second camera is involved.
DualCam composites whatever the camera sends it; it does not add a second layer of warping on top. A face already distorted by being close to the lens goes into the frame distorted, and a face at a normal distance goes in normal — the compositing step is not where the shape of a face gets decided.
Why the small window makes it look worse
The picture-in-picture inset is a smaller version of the source frame, scaled down and cropped into whichever shape you picked — the same crop-not-stretch approach used everywhere in DualCam, so it is not adding distortion of its own. What it does change is how much of the frame a face takes up and how much else is around it to compare against.
A distorted face at arm’s length in a full-screen shot has plenty of other things in frame — shoulders, background, the edges of a room — that read as normal, so the eye discounts a small amount of stretch near the centre. Crop that same face down into a small corner window with almost nothing else visible, and the widened nose or forehead is now most of what the window contains. The distortion has not gotten stronger; it has simply lost the context that used to make it easy to ignore.
Backing off does not lose you anything
Because the inset is already a scaled-down copy of the source frame, moving the phone slightly further away does not make the face smaller inside the window in any way that matters — DualCam is going to scale that source down to fit the inset regardless of whether it started large or already reasonably sized. What backing off actually buys is distance from the lens, which is the one thing that determines how distorted the face is before any cropping or scaling happens.
A normal arm’s length, the same distance that works for an ordinary selfie, is enough. There is no framing benefit to moving closer than that for a picture-in-picture inset, because the window was always going to be small regardless of how close the source was shot.
The same physics applies when the back camera is the inset
Picture-in-picture is not always a face in the small window. A tutorial or a craft video often puts the back camera — close on hands, a keyboard, an ingredient — into the inset while the front camera or a wider shot fills the main frame. The same rule holds: whatever is nearest the back camera’s lens gets magnified relative to what is further from it, so a close macro-style shot of hands can pick up its own version of this stretch, most visible at the edges of the frame nearest the lens.
Backing the back camera off a few inches and letting the inset’s scaling do the work of making the subject look close reads better than pressing the lens right up against it.
Check it live, because the preview is the recording
The viewfinder shows the exact frame that gets written to the file, distortion included, so there is no guessing involved — hold the phone at your intended distance before you press record and look at the inset itself, not just the main frame. If the face still looks wide, the fix is more distance, not a different corner or a different shape.
Inset size, shape and corner can all be changed mid-recording, so if something looks off once you are already rolling, the fastest fix is still the simplest one: move the phone back a little, not resize or reshape the window around the problem.
Common follow-up questions
Will picking a bigger inset size fix the distortion?
No. Size changes how much of the frame the window fills on the canvas, not how close the lens was to the face when it captured the source image. A bigger window just shows the same distorted proportions at a larger scale.
Does switching to a different inset shape help?
No. Square, circle and the tall rounded shape all crop the same scaled-down source frame differently; none of them undo distortion that was already present in that source before cropping.
Can this be fixed after recording?
No. DualCam composites both cameras into the recorded frame in real time, so whatever shape the face had going in is the shape it has in the file. The fix has to happen at the distance you hold the phone, before or during the take.
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.