pano2views

A frame is extracted from each video every N seconds, then converted like an image. Smaller = more frames + slower. Larger = fewer frames. Applies only to video inputs.

Base name for the ZIP file(s). Leave blank to derive from the input folder or file name. FOV and part-number suffixes are appended automatically.

90° = standard cubemap (faces tile exactly). Higher = overlap between faces (useful for photogrammetry); lower = gaps.

Larger faces = sharper but slower. "Match source" preserves the original detail.

Adds a tiny .xmp next to each face JPG with fixed calibration (focal from FOV, principal point at 0,0, no distortion, CalibrationPrior=exact). RealityScan reads these on import and won't re-solve the intrinsics. No pose is written — alignment is still free.