What the US rules are trying to prevent
US passport and visa photo checks are designed to make identity review simple. The photo needs to show the real person clearly, without styling or editing that changes how the person looks.
That is why the practical checklist is strict:
- one person in the image
- face directed toward the camera
- eyes open and visible
- neutral expression
- white or off-white background
- no heavy shadows
- no digital alteration
The rule set is not only about pixel dimensions. A technically square image can still fail if the face is too small, the background is busy, or the image has been altered.
Digital size
For online use, the Department of State digital image guidance uses a square format. The minimum file dimensions are 600 x 600 pixels.
YapaPhoto's US preparation path therefore exports a 600 x 600 JPEG from the uploaded image. It does not synthesize a new face, replace the background with AI, or create a fictional passport portrait.
Face position and crop
The face should be centered and large enough to be reviewed. In practice, the crop must leave space around the head while keeping facial features clear.
Common crop failures include:
- top of head too close to the edge
- chin cut off
- face far too small in the square
- head tilted enough to make the crop look unnatural
Automated preparation can estimate these points from the detected face rectangle. It still cannot prove final acceptance, because the submission system and human review decide the final result.
Background and lighting
Use a plain white or off-white background. The background should be even, without patterns, strong shadows, or objects.
Lighting should be soft and direct enough to show both sides of the face. A photo taken near a window can work if it avoids glare and hard shadows.
What YapaPhoto can safely do for US photos
For US passport and visa use, YapaPhoto should prepare from a real uploaded image:
- verify that one face is detected
- estimate whether the face is large enough
- crop to a square digital output
- export a 600 x 600 JPEG
- label the result as a measured preparation, not guaranteed acceptance
That is intentionally different from an AI portrait-generation flow.