How to Contour an Oval Face: A Decision Rule for a Shape That Needs Less

The five readings, and what a brush can move

The five readings are face length ratio, forehead ratio, jaw ratio, jaw angle and chin taper. Each is a quotient or an angle, each has a centre for the oval outline, and each has a tolerance published on this site's rule-set page. The table below puts the oval centre and one standard deviation around it beside the one stroke that can move that reading in each direction. The values are the tape-scale figures this site shows readers, not the detector's raw landmarks.

Two things in the table matter more than the others. The jaw angle has no stroke in either direction: it is a property of where the bone turns, and no quantity of pigment relocates a bone. The face length ratio has a stroke at both ends, but it is the reading most likely to be already inside its band on an oval, which is why the decision rule below so often returns zero. The other three readings are movable and are usually not the ones doing the work.

The bands are one standard deviation wide, not the full range of the outline. They are a reading aid, not a diagnostic cutoff. A reading just outside its band is not a different face shape; it is a reading that sits a little further from the oval centre than the sample's typical spread. The rule uses the distance to choose where to look, not to reclassify the face.

Centres and tolerances are this site's published tape-scale values, converted from the classifier's detector-scale constants by the same affine mapping the result page uses. They are not external norms, and the bands are one standard deviation rather than a clinical range. The stroke column describes the direction a pigment can push the reading, not a guarantee that the eye will read it that way under every light. The face length ratio and forehead ratio are both width ratios with different denominators; the jaw ratio is jaw width over cheekbone width; the chin taper is chin width over jaw width.

The decision rule: find the reading that is doing the work

For each reading, subtract the oval centre from your own value and divide by the band's standard deviation. That gives a z score: how many typical spreads your reading sits from the oval centre. The largest absolute z names the reading that is least oval-like. If every z is between −1 and +1, no reading is doing enough work to justify a corrective stroke, and the correct amount of contour is none.

A worked example, using the forehead ratio. A face with a 138 mm forehead width and a 136 mm cheekbone width gives 138 ÷ 136 = 1.015. The oval centre is 0.899 and the standard deviation is 0.081, so z = (1.015 − 0.899) ÷ 0.081 = 1.43. If the other four readings sit inside their bands, the forehead ratio is the one that earns a stroke, and the stroke is shade at the temples. Nothing is assigned to the jaw, the cheekbone, the hairline or the chin, because none of those readings is the one that is off.

The rule has a tie-break. If two readings are outside their bands, work on the one with the larger absolute z first, then re-read both before adding the second. The reason is not that one reading is more important. It is that a stroke placed for one reading changes the face another reading is measured on, and the second decision should be made after seeing the first result rather than in advance.

The rule also has a stopping condition. If the largest z is outside the band but only just — say 1.05 or 1.1 — the honest move is to treat it as optional. A single stroke at that distance is within the range where the measurement, the light and the blend can each account for more of the change than the stroke does. The rule is a way to avoid spending contrast on readings that are not deciding anything, and that logic applies at the edge as much as in the middle.

The strokes, and where they stop

Each stroke is deliberately small. The face length ratio is moved by a narrow band just inside the hairline at the top of the forehead and a narrow touch at the point of the chin; both shorten the visible vertical line without narrowing the mid-face. The forehead ratio is moved at the temples, from the hairline toward the top of the ear, and the stroke should stop at the level of the brow tail rather than run down the side of the face. The jaw ratio is moved along the outer jaw, from the corner toward the ear, but the surface technique — one plane rather than a drawn line — is on the square-face page and is not repeated here.

The chin taper is the least familiar of the four. A wide chin is narrowed by shading its two sides and leaving the centre clear; a pointed chin is brought forward by lighting across its width rather than by putting a bright point on the tip. Both are small moves, and both are easy to overdo because the chin is a small area: a stroke that looks restrained on the jaw is a stripe on the chin.

The jaw angle gets no stroke, and the page will not pretend otherwise. What shade along the jaw does is change where the eye guesses the corner is, which is a real effect and a different one from moving the angle. The difference matters because the guess is the part that fails under a light you did not pose for. If the jaw angle is the reading furthest from the oval centre, the useful response is not more contour; it is to accept the bone or to change the silhouette around it.

None of these strokes needs the wet-face boundary method repeated. That method — one light, wet skin, turn until the shine runs unbroken from the corner of the mouth past the cheekbone — is on the round-face page, and it is the right way to find the front-to-side boundary on any face. This page assumes you have a boundary and tells you which side of it to work on. It does not teach you to find it again.

What the published oval advice agrees and disagrees about

Four published positions on oval contour, spanning April 2020 to September 2026, are worth setting side by side because the disagreement is specific and it is the evidence for a decision rule. Three are external guides; the fourth is this site's own oval page, which calls the section the shortest on the page and says the work is optional.

Every date was read off the page or the page's own publication line. The first three rows are external guides and are identified by kind rather than by link; the fourth is this site's own oval page, which is the shortest makeup section on that page by its own description. The rows do not agree on where the shade goes: two put it under the cheekbone, one warns against a hard vertical stroke there, two put it at the temples or the sides of the forehead, and one leaves the forehead alone. They agree that an oval needs less correction than the other outlines, and none of them gives a rule for how much less. That gap is the reason for the decision rule above, not a defect in any one row.

Why the disagreement is the evidence for the rule

Read the fourth column down and the positions contradict each other in a way that cannot be explained by skill. Kendra Powell puts the shade under the cheekbone and along the jawline. Charlotte Tilbury puts it at the temples and deeper into the cheekbones, and warns that taking the hollow stroke too high elongates the face. Test My Face puts a light contour under the cheekbones and says the jaw and forehead should not be contoured aggressively. This site's own oval page puts the shade at the hairline and the chin tip and warns against a hard vertical stroke under the cheekbone. Four positions, four different maps.

The agreement is narrower but more useful: every one of the four says an oval needs less correction than the outlines around it. None of them says how much less, and that is the missing piece. A placement chart cannot answer the question because the answer depends on which reading is off. A face whose forehead ratio is 1.43 standard deviations above the oval centre needs a temple stroke; a face whose jaw ratio is 1.22 above needs a jaw stroke; a face with all five inside their bands needs neither. The same chart cannot be right for all three.

The decision rule also explains why the published maps disagree without anyone being careless. Each guide is describing the oval faces it has seen, and those faces did not all have the same reading out of band. A guide that saw mostly wide foreheads put the shade at the temples. A guide that saw mostly long ovals put it at the hairline and the chin. A guide that saw mostly balanced faces said the work was optional. Each is a local observation presented as a general placement, and the rule turns the local observations back into cases.

What the rule does not do is make the published maps useless. Once the reading names the area, the maps are still the place to learn how much product, which direction to blend, and what the area looks like on a real face. The rule chooses the area; the maps teach the hand.

Three checks before a second stroke

The first check is asymmetry. Measure the relevant width on both sides before you decide the reading is the face's and not the tape's. If the two sides differ by more than 3 mm — a threshold this page sets for itself, not a published one — treat them as two readings and work each side to its own number rather than splitting the difference. A face that is 4 mm wider at one temple is not a face with a forehead ratio; it is a face with two.

The second check is light. A stroke that reads as structure under a single overhead source can read as a grey smear under a window on one side. Check the finished face in at least two lights and from an angle other than straight on. The decision rule tells you which reading you are trying to move; the light tells you whether the stroke actually moved it or only moved the shadow.

The third check is the front plane. Stop the stroke before it reaches the area that faces you. Two zones meeting edge to edge read as a boundary drawn onto the face; a strip of unmodified skin between them reads as a surface turning away from the light. The round-face page's boundary method is the way to find that strip; this page's rule is the way to decide which side of it deserves the product.

If a second stroke is still tempting after all three checks, re-read the five numbers rather than adding product. The most common reason a first stroke did not do enough is that it was placed for the wrong reading, and a second stroke on top of the first is how a face ends up with two corrections for one problem.

Where it stops

Contour changes where an eye goes and what that eye takes to be the outline of a face. It does not change the outline. The five readings this site measures are taken off the bone and the skin, and a brush reaches neither. That is why the rule is written as a way to choose a stroke rather than a way to change a reading: the reading is the input, not the output.

The second limit is the one the published maps keep running into. Almost none of the lighting you will stand in is the single source the technique was demonstrated under. A placement that reads as structure in a bathroom mirror can read as a stain in daylight, and no amount of blending fixes a stroke that was placed for a light that is not there. The rule narrows the number of strokes; it does not make them light-independent.

Neither limit argues against doing any of this. Both say what it is. An oval face has less to correct than the outlines around it, and the most useful skill is knowing when the correct amount is one stroke and when it is none.

Questions this page answers