Photo guidelines
CardOffice checks every photo automatically and then shows it to a person. This page explains what both are looking for — useful whether you are taking the photo or reviewing it.
The short version
Section titled “The short version”A good ID photo looks like a passport photo: head and shoulders, facing the camera, evenly lit, plain background, in focus.
The requirements
Section titled “The requirements”| Requirement | Why | |
|---|---|---|
| Size | At least 600 × 800 pixels | Below this, the face is visibly soft at 300 dpi |
| Shape | Portrait, not landscape | Cards use a 35 × 45 mm photo area |
| Format | JPEG, PNG, WebP or HEIC | What phones and cameras produce |
| File size | Up to 15 MB | Larger files are almost always screenshots of screenshots |
| Face | Head fills 50–85% of the frame height | Any smaller and the face is unrecognisable on a card |
| Background | Plain and light | A busy background makes the face hard to pick out |
| Lighting | Even, no harsh shadows | Shadows across the face defeat visual identification |
| Focus | Sharp | Motion blur cannot be recovered |
| Expression | Neutral, eyes open, mouth closed | Standard for identity documents |
| Head coverings | Only for religious reasons, face visible | Consistent with passport rules |
What the automated checks measure
Section titled “What the automated checks measure”Each photo gets a score out of 100. It is a triage aid, not a verdict — a person still approves or rejects every photo unless your organisation has explicitly turned that off.
| Check | What it looks at | Weight |
|---|---|---|
| Resolution | Pixel dimensions against the minimum | 18 |
| Focus | Edge sharpness (variance of the Laplacian) | 18 |
| Orientation | Aspect ratio — portrait vs landscape | 16 |
| Lighting | Average brightness | 16 |
| Background | Variation across the outer border | 12 |
| Face | Skin-tone region size and position | 10 |
| Contrast | Tonal range | 4 |
| Colour | Colour cast, and whether it’s greyscale | 3 |
| Centring | Horizontal offset of the face | 3 |
Each check comes back as pass, warning or fail. Anything with a fail stays in the review queue no matter what its score is — a hard failure is never auto-approved.
What this is and isn’t. These are image-statistics checks modelled on the ICAO 9303 portrait requirements. There is no trained face-recognition model here, and this is not certified biometric conformance testing. It reliably catches the failures that waste cards — dark rooms, sideways phones, blurry snaps, cluttered backgrounds — and leaves judgement to a person.
Reviewing photos
Section titled “Reviewing photos”Open Photo review in the sidebar. Photos are oldest-first by default, which is the order people are waiting in.
Each card shows the photo, the score, and any checks that did not pass. That is usually enough to decide without opening anything.
Keyboard shortcuts
Section titled “Keyboard shortcuts”Reviewing 200 photos with a mouse is miserable. Use these instead:
| Key | Does |
|---|---|
A |
Approve the focused photo |
R |
Reject it (opens the reason box) |
Space |
Add it to the selection |
→ or J |
Next photo |
← or K |
Previous photo |
? |
Show this list |
Select several with Space, then use Approve n to clear a batch at once.
Writing a rejection reason
Section titled “Writing a rejection reason”The reason is emailed to the person, so write it as an instruction they can follow:
- ✅ “The photo is too dark. Please retake it facing a window.”
- ❌ “Poor quality.”
There are one-tap presets for the common cases.
Duplicate detection
Section titled “Duplicate detection”Every photo gets a perceptual hash. If the same face is submitted under two different identities, the review screen flags it — the hash survives resizing, recompression and small crops, which a checksum does not.
Photos and privacy
Section titled “Photos and privacy”- Original uploads are kept unmodified, so a bad crop can always be redone without asking someone to photograph themselves again.
- Rejected photos are deleted after 90 days by default.
- Superseded photos are deleted after 5 years by default.
- The photo on someone’s current card is never automatically deleted.
- Both retention periods are configurable per organisation in Settings → Photos.
Tips for taking photos at a counter
Section titled “Tips for taking photos at a counter”If you photograph people at a service desk:
- Put a plain light-grey or white backdrop behind the chair
- Use a lamp either side, or one large soft light in front — never overhead alone, which puts shadows in the eye sockets
- Mark the floor where people should stand, so framing is consistent
- Hold the camera at eye height, not above; looking down distorts faces
- Take two and keep the better one — it costs nothing
© 2026 Secure ID Ltd. All rights reserved.Documentation licence and terms