A common angular scale
Perspective uses width = 2f tan(H/2), height = f[tan(up) + tan(down)]. This is the correct camera projection for the selected angular inputs, but wide fields expand nonlinearly. A 106° field therefore appears about 2.23 times as wide as a 61.4° field, although it covers about 1.73 times the horizontal angle. Angular map instead uses equal pixels per degree, so its rectangle proportions directly match H:V in degrees. The panorama switches projection with the frames.
Both projections keep a minimum 140° horizontal and 110° vertical reference envelope, expanding one axis to fit the screen. The angular map is a diagram of direction, not a natural camera view. Dragging rotates the camera without changing headset angles. Neither mode is a life-size or through-the-lens simulation.
What these outlines mean
Frames approximate display bounds with rectangles. They do not reproduce lens contours, binocular overlap, eye relief, distortion, clarity, or your personal fit. Published nominal specs and measured effective FOV are different kinds of evidence. Vision Pro’s 95° × 60° is one 2024 test sample, not a universal hardware shape. The matching GearLab comparison also provides Quest 3 measurements (100° × 80°); the data-basis selector lets you avoid comparing its nominal value to that Vision Pro measurement.
For optical see-through glasses, the frame bounds digital imagery; the real room remains visible outside it. In isolate mode, those surroundings are dimmed for emphasis, not shown as physically blocked. For VR/passthrough headsets the dark surround illustrates the display boundary.