Tap the preview to focus/expose there (on cameras that support it). A zoom slider appears when the camera supports zoom.
Sourcing
Opt-in: adds rounded coordinates (e.g. 40.71N-74.01W) to photo/take/transcript names for sourcing and verification. Location is read at capture time and never leaves the device except in the file name.
On-device models
Models download once on first use and then work offline. Voice labeling uses a separate ~100 MB model.
Cloud upload (Nextcloud / WebDAV)
One-tap Upload in the take modal pushes the recording plus its transcript. Use a dedicated app password, not your main password — it is stored only in this browser. For iCloud Drive / Google Drive, use the Share button (native share sheet) instead.