Features
Part of How to tell a costume department's work from a real actor lookalike
Actor lookalikes quiz: the questions that matter
Actor lookalikes quizzes score film stills, which are character designs. The self-test on your own face that measures what a still can move.
What to take away
- Only one quiz item measures face matching: two frames, one question, same person or not.
- Naming a performer from a single still measures how many films you have seen.
- The real test can be run at home with twelve photographs of your own face.
- Score it by counting how many frames of you a stranger files under someone else.
- Candid shots are fairer than film stills, but they still carry lens, light, hair and expression.
The three question types a quiz can ask
The three question types
What a correct answer draws on
- Name the performer in this still
- Memory of the production
- Same person, or two people?
- Comparison of the two frames
- Which of these four plays this kind of part?
- Casting convention
Face matching
- Name the performer in this still
- No
- Same person, or two people?
- Yes
- Which of these four plays this kind of part?
- No
Only the middle row is a face test. The other two borrow its authority, and forced choice is why the borrowing works. A click between fixed options reads as knowledge even when you arrived by memory.
The naming item has a fallback you can catch in yourself. When the production is unfamiliar, you pick by part rather than by person. You choose the build inside the costume. Casting conventions reward that shortcut.
Why a single frame cannot carry the answer
A lens changes the shape of a head before the quiz reaches you. A wide lens close in stretches a face. A long lens from across the street flattens it. Both choices belong to the production, not to the performer.
[figure 1]
Seven layers between performer and still
- Makeup: corrective to prosthetics
- Hair: wig or character cut
- Wardrobe: silhouette and neckline
- Lighting: half-face shadow possible
- Lens: wide stretches, long flattens
- Color grade: shifts skin tone
- Digital work: may alter face
Example items worth putting on a card
The same-or-different format is not a gimmick. The Glasgow Face Matching Test, a published set of face pairs used in perception research, works this way. A border officer runs the same check at a desk.
Here are four real cases. Each is a same-or-different item. The answer to every one is the same person.
- Gary Oldman as Winston Churchill in Darkest Hour (2017), against a press photograph of Oldman from the same awards season.
- Nicole Kidman in The Hours (2002), where she wore a prosthetic nose, against a red carpet frame from 2001.
- Charlize Theron as Aileen Wuornos in Monster (2003), against Theron at a premiere that year.
- Tom Cruise as Les Grossman in Tropic Thunder (2008), bald cap and padded suit, against an interview still from 2008.
None of these is a trick. The performer is the same in both frames. What changed is exactly what a still-based quiz will not hold constant. Telling a costume department's work from the performer is the next thing to learn.
The self-test on your own face
You control every layer in this version. Shoot two frames under each of six conditions, twelve in total.
- Print all twelve at the same size on the same paper, so no frame reads as the important one.
- Add eight to twelve photographs of other people, shot the same way.
- Shuffle the pile and hand it to someone who has never met you.
- Give one instructionsort these into people. Say nothing about how many people are in the pile.
The six conditions:
- Arm's length, phone at face height
- Across the room, camera on a shelf
- Hair tied back
- Hair down and loose
- Daylight from a window, then overhead room light
- Neutral face, then a full laugh
The strangers are what make the sort honest, because two photographs of one face can read as two people and the sorter has no way to know the count.
[figure 2]
Six conditions for your self-test
- Phone at arm's length, then across room
- Beside window, then ceiling light
- Neutral face, then full smile
- Hair as worn, then pulled back
- With and without glasses, hat, scarf
- Color, then same frame monochrome
Scoring the self-test
Two numbers matter, and both are counts rather than marks out of ten. The first is how many distinct piles contain at least one frame of you. A clean sort gives one. A split sort gives two or three.
The second is how many strangers' frames were placed in your pile. Every stranger who lands with you is a merge, and a merge is the same error as a split, read backwards.
If the first number stays at one, your twelve frames were too alike: one distance, one light, one expression. Loosen the set and run it again.
Watch which condition did the splitting. Arm's length against across the room is the usual cause. Hair back against hair down comes next.
Playing a live quiz anyway
Score the routes instead of the total. Mark each item as answered by recognising the film, by recognising the part, or by reading the face. The last count is usually the smallest, and it is the only one that speaks to face matching.
The stranger's confidence in the self-test is worth carrying over. A wrong sort arrives with the same certainty as a right one, which is why arguments over celebrity lookalikes pairs rarely close.
A set of twenty same-or-different items has a chance floor of ten correct, so a score near ten is a coin flip. Machine systems normalise pose, light and expression before they compare two faces, as the facial recognition systems literature sets out. A film still adds all three.
Common questions
What does an actor lookalikes quiz actually measure?
Usually recall of films and casting types. Face matching is measured only by same-or-different items, where both frames show a real person and the conditions differ.
Why does the self-test need other people's photographs?
Without them, the person sorting knows every frame is you, and the task stops measuring anything.
Is a candid shot a fairer quiz item than a film still?
Fairer, not fair. It removes the character design and keeps the lens, the light, the hair and the expression.
Can familiar actors be recognised from a single still?
Yes, easily. Familiarity survives bad photographs. The interesting cases are faces you know less well.
