If you're building your own scanning flow, two endpoints let you do it a page at a time instead of uploading a whole document:
POST /exams/{id}/sessions/scan-imagessends photos of a student's pagesPOST /exams/{id}/sessions/scan-student-namereads the name off the first page, so you can confirm who you're looking at before you commit the rest
POST /exams/{id}/sessions/scan-document is still there for whole PDFs.