Ear visibility and neutral expression in ID photos: what the specs actually require

A neutral expression is a hard requirement in the biometric standard behind passport photos. ISO/IEC 39794-5:2019 defines it as not smiling, mouth closed, no teeth showing, and a closed-mouth smile does not pass either ¹. Ear visibility is a different kind of rule. The US and UK photo pages never mention ears at all ²³. India's official OCI specification asks for "both edges of the face" rather than ears ⁴. The reason ears keep coming up is that the standard uses them as the two points that define the width of your face, and software cannot locate them reliably when hair is in the way ¹.

Two pieces of passport photo advice get repeated everywhere: keep a neutral face, and tuck your hair behind your ears. The first is a genuine requirement with a documented technical reason. The second is mostly folklore built on top of a real measurement problem, and almost no government page actually says it.

The gap matters because the two rules fail differently. Get the expression wrong and you have broken a written rule that a reviewer or an automated checker can point at. Get your hair over your ears and you have made the geometry harder to measure, which may or may not cost you anything depending on the country and the tool.

Here is what the standard says, what the research behind it found, and what the US, UK, and Indian authorities each put in writing. If you want the measurements checked before you submit anything, an on-device app like SpecSnap validates each crop against the selected document's published tolerances while you shoot, and does it without uploading your face anywhere.

Jump to: Neutral expression · What expression does to matching · The mouth evidence · Ears · Country rules · Getting it right · FAQ · Sources

Neutral expression is written into the standard#

The biometric spec underneath modern passport photos is ISO/IEC 39794-5:2019. Section D.1.4.3.2 requires a neutral facial expression, and it spells out what that means: the subject does not smile, the mouth is closed, and no teeth show ¹.

One detail surprises people. Even a smile with a closed mouth is non-compliant under that section ¹. The rule is about the whole expression, not only about whether your teeth are visible.

The successor quality standard goes further. Under ISO/IEC WD4 29794-5:2022, expression neutrality is a mandatory image quality aspect when capturing reference samples for identity documents ¹. That is the category a passport photo falls into. It is not a stylistic preference that a friendly reviewer might wave through; it is listed as mandatory for exactly this use case.

National wording drifts from the standard, though, and that drift is worth knowing about. The UK asks for "a plain expression and your mouth closed" ². The US asks for a neutral expression with both eyes open and visible and the mouth closed ³. India's OCI specification says something looser: "The expression on the face should look natural" ⁴. Three authorities, one underlying standard, three different words.

What your expression actually does to a face-matching system#

The usual explanation is that smiling moves your facial landmarks and throws off the measurements. That is not quite what the research found.

A 2022 state-of-the-art review prepared for Germany's Federal Office for Information Security surveyed the evidence on how expression affects face recognition. Current recognition systems cope well with slight variations in expression. It is the larger variations between two images of the same person that can cause a false rejection ¹.

Which expressions cause trouble is not intuitive either. In the studies reviewed, sadness and disgust could significantly move the comparison scores, while joy and surprise had less negative effect ¹. Grimacing is a category of its own: extreme expressions have been shown to defeat state-of-the-art recognition systems entirely, which is to say a determined person can pull a face and not be recognised ¹.

So why require neutrality at all, if a small smile is survivable? Two reasons, both about consistency rather than any single photo. With a neutral face, your features are usually clearly visible, whereas other expressions can partially hide them behind wrinkles or squinted eyes. And a neutral face has lower variance than an expressive one, because there are fewer ways to do it ¹. Using neutral images only is therefore expected to improve recognition performance, and that has been confirmed empirically ¹.

That is the real argument. Your passport photo has to match a live scan of your face years later, at a gate you have never seen. Neutral is the one expression you can reproduce on demand.

The open-mouth evidence is thinner than the advice#

Here the same review is refreshingly honest, and it is worth knowing before you panic about a slightly parted mouth.

The finding that an open mouth significantly affects recognition accuracy, while minor movements at the corners of the mouth or a slight smile do not, comes from an analysis that used only simple recognition algorithms based on PCA and LDA. Those are not state of the art ¹. Modern recognition networks are trained on large in-the-wild datasets full of open and closed mouths, so it is plausible they handle the difference far better ¹.

The review still lands on closing your mouth, for reasons that have nothing to do with the matching score. Presentation-attack detection, morphing-attack detection, and manual identity verification by a human officer all benefit from a closed mouth ¹. Your photo is not only fed to a matcher. It is also checked for tampering and looked at by people.

Practical upshot: close your mouth because the standard says so and because it helps the other checks, not because a parted lip will fail a modern matcher.

Ears are not a biometric feature, they are a measuring point#

This is the part almost every "tuck your hair back" article gets wrong. Nothing in the standard is reading your ears as an identifying feature. Your ears are where the ruler goes.

ISO/IEC 39794-5:2019 uses the ears to define the width of the face, and the crown of the head to define its height ¹. Those measurements feed the head-size and positioning rules: whether your head is centred, whether it is too big for the frame, whether it is cut off at an edge ¹. If you have ever wondered how a checker decides your face fills the right proportion of the picture, this is the machinery. Our head size guide covers what those proportions come out to per country.

Then comes the problem. Facial landmark extractors do not output landmarks for the ears, and the review names the reason: ears are often occluded by hair ¹. So software has to estimate where your ears are by interpolating from the eyes and the contour of the face. Because the ratio between inter-eye distance and head size varies from person to person, that estimate has to be made carefully ¹.

The standard itself concedes the difficulty. If the position of the crown, chin, or ears cannot be determined exactly because of poor positioning, ISO/IEC 39794-5:2019 says "a good guess shall be made" ¹.

A guess. That is the actual cost of hair over your ears: it moves your face width from a measurement to an estimate, and it does the same for any tool trying to judge whether your crop is compliant. It also degrades unrelated checks. One published method for assessing background uniformity defines its sample regions relative to the subject's ears, and the review flags its inability to locate those ears as a limitation ¹.

None of this makes covered ears a rejection in most countries. It makes your photo harder to verify, and harder to verify is where borderline cases come from.

What each authority actually writes down#

Three things stand out.

No one asks for ears. Not the US, not the UK, not India's own OCI specification. What India asks for is that head coverings must still leave "the facial features from bottom of chin to top of forehead and both edges of the face" clearly shown ⁴. "Both edges of the face" is the requirement that ear advice is a clumsy proxy for. Your ears do not have to be on display. The sides of your face do have to be findable.

Hair rules are about the eyes. The UK's line is "not have hair in front of your eyes", plus nothing covering your face ². That is a different concern from the ear geometry, and it is the one that gets photos bounced.

Every one of them bans editing. The UK requires the photo be unaltered by computer software ². India says do not retouch, enhance, or soften it ⁴. This is why a validating app and a beautifying app are opposites: one tells you the crop is wrong, the other quietly changes your face and creates a compliance problem. See why was my passport photo rejected for the full rejection list, and the biometric requirements checklist for the numbers per country.

Getting both right in one shot#

Neither rule needs a studio. They need one deliberate take.

1. Set your face before you shoot, not after. Mouth closed, no smile, eyes open and looking at the lens. Do it in a mirror once so you know what your neutral face feels like, because most people's default is a faint smile.
2. Move hair off the sides of your face, not just off your ears. The goal is that both edges of your face are unambiguous. If your hair falls forward over your jawline, push it back. Ears themselves are optional in every spec above.
3. Kill the shadows. Every authority in the table bans them, on your face and behind you. Even, front-facing light is the whole fix; see lighting requirements for a clean ID photo indoors and removing shadows from your ID photo background.
4. Validate the geometry, do not edit the face. Head height, centring, and face width are measurements you cannot judge by eye, and the editing bans mean you cannot fix a bad capture afterwards. Retake instead.

That last point is where an on-device validator earns its place. SpecSnap covers 46 government photo specifications across 20 countries and measures each crop against the published tolerances for the document you picked, returning a pass or a warning in real time rather than after an upload. The full pipeline runs locally with zero server uploads, so your face never leaves the phone, which matters more than usual for a biometric image. It is free to try every spec, with a one-time in-app purchase to export without a watermark. Our on-device validation comparison puts it against the alternatives.

FAQ#

Do my ears have to be visible in a passport photo?

In the US, UK, and Indian OCI specifications, no. None of those three photo requirements mention ears ²³⁴. India asks instead that "both edges of the face" be clearly shown ⁴. The reason ear advice persists is that ISO/IEC 39794-5:2019 uses the ears as the two points that define the width of your face, and landmark software cannot locate them reliably when hair covers them, so the position has to be estimated ¹.

Is a small closed-mouth smile acceptable?

Not under the standard. ISO/IEC 39794-5:2019 Section D.1.4.3.2 requires no smile, a closed mouth, and no visible teeth, and a smile combined with a closed mouth is explicitly non-compliant ¹. National pages vary in strictness of wording, from the US "neutral expression" to India's "should look natural" ³⁴, but a plainly neutral face satisfies all of them.

Why does smiling matter if face recognition is so good now?

Current systems handle slight expression differences well; it is bigger differences between two images of the same person that can cause a false rejection ¹. Neutrality is required for consistency: a neutral face keeps features clearly visible instead of hidden by wrinkles or squinted eyes, and it has lower variance than expressive faces, so restricting reference images to neutral ones improves recognition performance ¹.

Does an open mouth really hurt the match?

The evidence is weaker than the advice suggests. The study showing that an open mouth significantly affects accuracy used older PCA and LDA algorithms, not state-of-the-art recognition networks, which are likely far less sensitive to it ¹. Closing your mouth is still the right move, because the standard requires it and because attack-detection systems and human reviewers benefit from it ¹.

Can I just edit my expression or hair afterwards?

No. The UK requires your photo to be unaltered by computer software ², and India's specification says do not retouch, enhance, or soften it ⁴. Retake the photo instead. Validating the capture against the spec is allowed and useful; changing your face is what gets photos refused.

Sources#