Can MRZ scanning work in poor lighting or from an angle?

To a degree, yes, though there are real limits. Modern MRZ scanning software applies image processing techniques, adjusting brightness and contrast, correcting for perspective distortion when a document is photographed at a slight angle, and sometimes combining multiple frames from a live camera feed to get a cleaner composite image, that make it noticeably more tolerant of imperfect conditions than older, stricter hardware-based passport readers used to be.

Moderate angles are usually handled well. If a document is tilted 10 or 15 degrees relative to the camera, good software can typically correct for that perspective distortion mathematically before running OCR, effectively "flattening" the image as if it had been shot straight-on.

Very extreme angles, where the MRZ's characters become significantly compressed or distorted in the captured image, are harder to correct reliably and more likely to produce reading errors or an outright failure to extract data.

Lighting has a similar story. Dim but even lighting is usually workable, since software can boost brightness and contrast digitally to compensate, though at some point noise in the image becomes too significant to correct.

Glare is a tougher problem than dim light in many cases, because a bright reflection can completely obscure the characters underneath it rather than just making them slightly harder to read. No amount of digital enhancement recovers text that was never actually captured in the image in the first place.

The most reliable approach, and the one most identity verification products take, is prevention rather than correction: giving the user real-time feedback during capture, prompting them to reduce glare, find better lighting, or hold the document more level, before an image is actually processed. This catches problems where they're easiest to fix, by asking the user to adjust and try again, rather than trying to recover a usable result from a genuinely poor capture after the fact.

ScanDoc's scanning flow includes this kind of live capture guidance specifically to reduce failed reads caused by lighting and angle issues, aiming to get a clean image on the first or second attempt rather than relying purely on post-capture image correction to compensate for a poor original shot.