Teeth whitening before and after: what actually changes, and why photos lie

A tooth does not have a colour in the way a paint chip has a colour. What reaches the camera is light from some source, partly reflected off the enamel surface, partly scattered inside the tooth, partly returning from the dentine underneath, and then filtered through whatever the camera decides the scene white point is. Change any link in that chain and the number the sensor records changes, whether or not the tooth did.

This is why the before-and-after format is so persuasive and so weak at the same time. It presents two images as if the only difference between them is the treatment, when in practice the two images were usually taken on different days, in different rooms, with a phone that re-decided its white balance and exposure both times. The comparison feels like evidence. Structurally, it is closer to an anecdote with two data points and no controls.

None of that means whitening does nothing. It means that photographs are a poor instrument for measuring how much it did, unless you deliberately turn them into a better one. The rest of this article is about which part of the change is chemistry and which part is optics.

The single most useful distinction in this whole topic is where the discolouration physically sits. Dentistry splits it into extrinsic and intrinsic, and the two behave so differently that treating them as one thing is the reason people end up disappointed.

Extrinsic: on the outside of the tooth

Enamel is covered by a thin protein film, the acquired pellicle, which forms continuously from saliva. Pigmented compounds in things that pass over the teeth — tea, coffee, red wine, some spices, tobacco smoke — bind into that film. Because this layer is superficial, it is the layer that responds most readily to surface-level intervention, and it is also the layer that comes back, because the pellicle reforms and you keep drinking coffee. Extrinsic staining is the reason a professional clean can produce a visible change with no bleaching chemistry involved at all.

Intrinsic: inside the tooth

Intrinsic discolouration lives in the tooth structure itself, mostly in the dentine. Causes are varied and they are not equivalent: developmental conditions such as fluorosis, certain medications taken during the years the teeth were forming, trauma to a single tooth that changes the pulp, and the slow physiological thickening of dentine that happens to everyone with age. These do not sit on a surface you can polish. Whether any given intrinsic case responds to whitening, and how much, is genuinely a clinical judgement — it depends on the cause, and a single dark tooth after an old injury is a different conversation from generalised age-related darkening.

Most real mouths have some of both. That mixture is exactly why generic before-and-after galleries generalise badly: you cannot tell from someone else's photograph which component of their discolouration moved, so you cannot infer what yours would do.

People tend to picture enamel as a white shell. It is closer to a semi-translucent glaze. Light entering a tooth passes through the enamel, reaches the dentine underneath, scatters, and comes back out. The dentine is the more yellow of the two tissues, so the perceived shade of a tooth is substantially a report on the dentine seen through a partly transparent layer.

Two consequences follow. First, thinner or more translucent enamel lets more of the underlying colour through, which is part of why teeth tend to look less bright over a lifetime even in people whose habits never changed. Second, at the incisal edge — the biting edge of the front teeth — there is often little dentine behind the enamel at all, so light passes through and the edge can read slightly grey or translucent rather than white. That is normal anatomy, not a defect, and no amount of whitening turns a translucent edge opaque.

This also explains why a whitened smile can still look uneven. If the underlying dentine varies between teeth, or the enamel thickness does, then lightening the whole arch does not necessarily equalise it. Understanding that in advance is the difference between a satisfying result and a disappointing one.

Colour is one axis. A smile has several, and it is worth being explicit about which ones whitening leaves completely untouched, because a lot of dissatisfaction after a course of treatment is really dissatisfaction with a different variable.

• Alignment and crowding. Rotation, spacing, overlap and bite are structural. A brighter crowded tooth is a brighter crowded tooth.
• Tooth shape and proportion. Length, width, the curve of the incisal edges and how much tooth the lip reveals are unaffected.
• Gum line and gum health. The height and symmetry of the gingival margin frames every tooth, and it is a clinical matter.
• Decay, wear and existing restorations. Whitening is cosmetic chemistry, not treatment.
• How the smile reads to another person. Whether a smile looks genuine is driven by the muscles around the eyes, not by shade.

That last one is the axis we can actually speak to. On /analyze we read whether a smile engages the orbicularis oculi — the eye muscle whose contraction produces the crease and cheek raise that observers use, largely unconsciously, to separate a felt smile from a performed one. That signal is independent of tooth shade. A person with a modest natural shade and a genuine smile reads warmer than a bright, tight, eyes-uninvolved one. It is also the cheapest thing on this entire list to change.

Here is the part that matters most if you are evaluating anyone's results, including your own. Each of these can shift apparent tooth brightness on its own, and in a typical unplanned before-and-after pair, all six are different.

1. White balance

Every camera has to guess what counts as white in the scene, then rebalance the whole image around that guess. A phone photographing a mouth in a warm domestic room will typically pull the image cooler to compensate for the tungsten cast; the same phone in a daylight-balanced clinic makes a different decision. Since a tooth is a near-neutral, slightly yellow object, it is precisely the kind of subject that moves most when the white point moves. Two photographs, two automatic decisions, one apparent transformation.

2. Exposure

Brighten an image and the lightest areas run toward white first. Teeth are usually among the brightest things in a close-up of a face, so a modest exposure increase disproportionately whitens them, and if the highlights clip, the yellow information is simply gone. An overexposed after photo will always beat a correctly exposed before photo, regardless of treatment.

3. The light source itself

A hard on-camera flash close to the lens does two things: it floods the tooth surface with specular reflection, which reads as brightness, and it fills in the shadow inside the mouth, which raises the contrast between teeth and the dark oral cavity behind them. Soft window light does neither. Ring lights, clinic overheads, phone torches and mixed sources all render the same tooth differently, and mixed colour temperatures in one frame make the camera's white balance guess even less stable.

4. Angle, lip position and lens

How far the lips retract changes how much tooth is visible and how much dark background surrounds it. A wider retraction shows more of the whiter middle of the tooth and less of the cervical region near the gum, which is naturally darker. Camera distance matters too: a phone held close applies noticeable perspective distortion, and a subtly different head angle changes which surfaces catch the light. This is the same family of effects that makes people look different across ordinary photographs, covered in more depth in why you look different in photos than in the mirror.

5. What is next to the teeth

Human colour perception is relational. A tooth against a dark lip colour looks brighter than the same tooth against a pale one. Skin tone in frame, lipstick, tan, background wall colour and even the colour of a shirt all shift the judgement. Photographers have exploited this forever, and it is not a trick you need to intend for it to work on you.

6. Computational processing and filters

Modern phones do not record a scene, they reconstruct one. Tone mapping, local contrast, skin smoothing and portrait modes all operate before you ever see the file, and some editing tools ship an explicit tooth-brightening control. Any comparison that goes through a beauty filter on either side is uninterpretable. If you cannot confirm the processing pipeline was identical, you do not have a comparison.

You cannot eliminate every variable with a phone, but you can hold enough of them still that the remaining difference means something. The principle is the same one any measurement discipline uses: fix the instrument, fix the conditions, and put a known reference in the frame so you can check afterwards that you succeeded.

1. One room, one time of day. Pick a spot and note it. Daylight changes hour by hour and season by season, so if you are relying on a window, use the same window at roughly the same time.
2. One light source, everything else off. Mixed sources are the main cause of unstable colour. Kill the overheads if you are using a window, or vice versa.
3. Lock white balance and exposure. On most phones a long press on the subject locks focus and exposure; a camera app with manual white balance is better. Auto anything defeats the whole exercise.
4. Put a neutral reference in frame. A grey card is ideal, but a piece of plain white printer paper held beside the mouth works. It gives you something with a fixed real-world colour to compare across the two images. If the paper is a different brightness in the after shot, so is everything else, and the teeth prove nothing.
5. Same distance, same height, same lens. Mark where you stand. Do not switch between the front and rear camera between shots — they are different sensors and different processing.
6. Same lip position. Choose one: a relaxed natural smile, or a deliberate wide retraction. Do not compare one against the other.
7. No filters, no beauty mode, no editing. Turn off portrait mode and any automatic retouching. Export the originals.
8. Compare on one screen, side by side. At the same brightness setting, in the same session. Screens auto-adjust too.

Done properly, this takes about five minutes to set up and thirty seconds to repeat. Done improperly, it produces a pair of images that tell you nothing and feel like they tell you everything, which is worse than not taking them.

We are not making a dental point here — we are making a measurement point, from our own data. RealSmile derives 17 facial-geometry metrics from 68 landmarks and turns them into a Face Score percentile. That pipeline is deliberately geometric, and geometry ought to be the most stable thing about a face. It still moves between photographs of the same person taken minutes apart.

When we looked at people rescanning themselves in a single session, the score for one face routinely shifted by several points across shots. The face did not change. The lighting, the camera height, the head angle and the expression did, and that was enough to move landmark placement and the metrics computed from it. We wrote that up plainly because it constrains what we are allowed to claim: a score from one photograph is a reading of a photograph, not a fixed property of a person.

If a geometric measurement is that sensitive to capture conditions, a colour judgement — which depends on white balance, exposure and surrounding tones, none of which affect geometry — is far more so. That is the whole argument of this article in one line. It is also why the 17-metric scan and the face rating tool both push you toward even, directional light and a level camera before they push you toward anything else, and why the same discipline that makes a whitening comparison honest also makes a facial measurement more repeatable.

We cannot tell you what to do, and we are not qualified to. What we can do is describe the shape of the conversation, so it is less opaque when you have it. A dental assessment of discolouration generally covers the cause of the colour rather than the colour itself: whether it is surface deposit or internal, whether one tooth differs from its neighbours and why, the condition of the enamel, the presence of any restorations, and whether there is anything — decay, gum inflammation, sensitivity, wear — that should be addressed before a cosmetic question is even on the table.

Shade itself is recorded against physical reference tabs held next to the tooth under controlled clinic lighting, rather than judged from a photo, precisely because everyone in the field knows how badly photographs behave. That is a useful thing to know: the profession that does this for a living does not trust a picture either.

Questions that tend to produce a clear answer:

• Is the discolouration I have mostly surface-level or internal, and how can you tell?
• Do I have any restorations that will not change colour, and would they need replacing to match?
• Is there anything you would want treated or monitored before a cosmetic step?
• What range of outcome is realistic for a case like mine, and what would make it less predictable?
• What are the known side effects, how common are they, and how long do they last?
• How is the result going to be recorded, so that any change is measured rather than remembered?

That last one is worth pushing on. If nobody records a baseline under controlled conditions, then in three weeks the only comparison available is between a new photograph and your memory of your old teeth, and memory for colour is poor. Sensitivity during and after whitening is a well-recognised effect and a normal thing to ask about directly rather than discover.

Shade is one input into how a smile reads, and it is neither the largest nor the most controllable. The things that move a first impression are, roughly in order: whether the smile is genuine, whether the photograph was taken under decent light at a sensible camera height, the framing of the mouth by lips and gums, and then colour. Whitening addresses the last item on that list. It is not nothing, but it is a strange place to start.

There is also a ceiling worth naming. Uniform, maximally bright teeth are not what people read as attractive so much as what they read as dental work. Natural teeth vary slightly in shade across the arch and are translucent at the edges; removing that variation entirely tends to register as manufactured even when a viewer cannot say why. Where the line sits is subjective, but the existence of a line is not.

The generalisable habit here is the same one behind every honest appearance measurement, and it is the same argument we make about skin in the men's skincare routine guide: control your capture conditions, keep a baseline, and judge changes against a fixed reference rather than against how you remember looking. Do that and you will know what worked. Skip it and you will have a folder of flattering photographs and no idea.