What Is FACS Capture?
FACS capture is the 3D scanning of an actor’s face in a library of individual expressions, each based on the Facial Action Coding System. The result is a set of high-resolution head scans that VFX artists turn into blendshapes: the building blocks that let a digital face smile, frown and speak exactly like the real performer.
FACS scans are usually captured alongside a full cyber scan of the actor. This guide explains what FACS is, how a FACS session works, how scans become blendshapes, what VFX teams receive, and how it differs from facial performance capture. Every head in the video and images below is a real low-poly FACS scan from a Visualskies session.
Key takeaways
- FACS, the Facial Action Coding System, breaks every visible facial movement into numbered action units.
- A FACS scan is a series of head scans, one per expression, captured with a synchronised, cross-polarised camera rig.
- Artists align every expression to the actor’s neutral face and turn the differences into blendshapes (also called morph targets or shape keys).
- Animators or performance capture data mix the blendshapes to build expressions that were never scanned on their own.
- On Another Simple Favour, Visualskies captured 31 FACS expressions as part of 54 scans over two days in Rome.
What is FACS?
The Facial Action Coding System was developed by psychologists Paul Ekman and Wallace V. Friesen and first published in 1978. It describes the face in terms of action units (AUs): the smallest visible movements a face can make, each linked to the muscles that produce it. Raising the inner brows is one action unit. Pulling the lip corners up, the core of a smile, is another. Dropping the jaw is another.
Almost any expression can be written as a combination of action units. That is what makes FACS so useful for visual effects: instead of scanning thousands of random faces, a production can scan a structured set of movements and recombine them to build any expression it needs.
| Action unit | Name | What the face does |
|---|---|---|
| AU1 | Inner brow raiser | Lifts the inner ends of the eyebrows |
| AU2 | Outer brow raiser | Lifts the outer ends of the eyebrows |
| AU4 | Brow lowerer | Pulls the brows down and together into a frown |
| AU6 | Cheek raiser | Lifts the cheeks and creates crow’s feet |
| AU12 | Lip corner puller | Pulls the lip corners up and back, as in a smile |
| AU18 | Lip pucker | Pushes the lips forward into a pucker |
| AU24 | Lip pressor | Presses the lips together |
| AU26 | Jaw drop | Lets the jaw fall open |
| AU43 | Eyes closed | Closes the eyelids |
How a FACS scanning session works
A FACS session takes place in a dedicated head rig: a ring of synchronised, cross-polarised cameras and lights focused on the face. Every expression is captured in a fraction of a second, so the actor only needs to hold each pose for a moment.
- Neutral scan. The session starts with a relaxed, neutral face. Every other expression will be measured against it.
- Expression list. The actor works through an agreed list of expressions, from single action units such as a brow raise to combined shapes such as a wide smile or a snarl. The list is set by the VFX team according to how the digital face will be used.
- Review on the spot. Each capture is checked for sharpness and coverage before moving on, so a missed shape can be retaken while the actor is still in the chair.
- Processing. The photographs for each expression are colour-calibrated and reconstructed into a textured 3D head with photogrammetry, exactly as for a full cyber scan.
An example FACS expression list
Combining action units keeps a session short. These are the shapes from one Visualskies FACS session, the same scans used in the video above:
- Neutral, and neutral with eyes closed
- Smile
- Frown with brow furrow
- Brows up with jaw open
- Eyes closed with pucker
- Show teeth with brows down
- Funneler with squint
- Lips pressed with eyes wide
- Lips “oo” with brows up
- Compress
From scans to blendshapes
A raw expression scan is not yet useful to an animator. To become part of a facial rig, every expression has to share the same mesh structure as the neutral head. Artists wrap a clean, consistent topology onto each scan, then subtract the neutral shape to leave only the movement. Those differences are the blendshapes, called morph targets in Unreal Engine and shape keys in Blender.
In the finished rig, animators or facial performance data mix the blendshapes together. Sixty per cent of one shape plus forty-five per cent of another produces an expression that was never scanned on its own, but still moves with the actor’s real anatomy, because every ingredient came from their face.
Why the detail matters
Skin does not just move; it folds. Crow’s feet appear in a smile, forehead lines appear in a brow raise. Because every FACS scan records texture as well as shape, artists can bring in those wrinkles only when the matching expression is used, which is a large part of what stops a digital face looking rubbery.
What VFX teams receive from a FACS session
Deliverables are agreed with the VFX supervisor before the shoot. A typical FACS package can include:
- Raw, colour-calibrated photographs for every expression, so the facility can reprocess them in its own pipeline.
- A textured 3D head scan per expression, aligned in the same space as the neutral.
- Low-poly versions, like the ones used in this article, for quick review, layout and look development.
- Clean, wrapped meshes on a shared topology when the production wants scans that are ready to become blendshapes.
FACS scanning vs facial performance capture
The two are often confused, but they record different things:
- FACS scanning records the shape of the face in a library of held expressions, at very high resolution.
- Facial performance capture records how the face moves over time during a performance, usually with a head-mounted camera or a capture volume.
Most digital human pipelines use both: the FACS library gives the digital face its range of motion, and the performance data drives it shot by shot. For capturing a whole moving performance in 3D, see volumetric capture.
When does a production need FACS scans?
FACS scans are worth capturing whenever a digital face will be seen moving at close range: digital doubles that speak or emote, de-ageing and face replacement, creature work that is driven by an actor’s performance, and real-time characters for games or virtual production. For a background double seen at a distance, a full body and neutral head scan is usually enough.
Case study: FACS scanning for Another Simple Favour
For Paul Feig’s Another Simple Favour, Visualskies drove its mobile cyber rig from London to Cinecittà Studios in Rome to scan cast for VFX company Crafty Apes. Full body scanning took place on the first day, and head and FACS scanning on the second.
- 54 scans in two days on set
- 31 FACS expressions
- 14 full body scans and 9 head scans
We have run FACS sessions for games and animation too. At our New York studio we scanned Lenny Kravitz for 007 First Light, delivering more than 40 expressions for his character Bawma, and Uma Thurman for Quentin Tarantino's Yuki's Revenge, where 16 expressions were processed and aligned within a few working days.
FACS capture FAQs
What is FACS capture?
FACS capture is the 3D scanning of an actor’s face in a library of individual expressions, each based on the Facial Action Coding System. VFX artists turn the scans into blendshapes so a digital face can smile, frown and speak exactly like the real performer.
What does FACS stand for?
FACS stands for the Facial Action Coding System, a method for describing every visible facial movement as a set of numbered action units. It was developed by psychologists Paul Ekman and Wallace V. Friesen and first published in 1978.
What is a FACS scan?
A FACS scan is a series of 3D head scans of an actor holding individual facial expressions, each based on one or more FACS action units. VFX artists use the set to build blendshapes so a digital face can move exactly like the real one.
How many expressions are in a FACS session?
It depends on how close the digital face will be seen. A background double may need a handful of key expressions, while a hero digital human can need dozens. On Another Simple Favour, Visualskies captured 31 FACS expressions as part of 54 scans over two days at Cinecittà Studios in Rome.
What is the difference between a FACS scan and facial performance capture?
A FACS scan records the shape of the face in a library of held expressions. Facial performance capture records how the face moves over time during a take. The FACS library gives the digital face its range; the performance data drives it.
What is the difference between blendshapes and morph targets?
They are the same thing under different names. Blendshapes (Maya), morph targets (Unreal Engine and many game engines) and shape keys (Blender) all store the difference between a neutral mesh and an expression, so the expression can be dialled in by weight.
Why is cross-polarised lighting used for FACS scanning?
Cross-polarised filters on the lights and cameras remove the shine from skin, so each scan records true skin colour and fine detail such as pores and wrinkles. Wrinkles that appear in one expression and not another are part of what makes a digital face believable.
Can FACS scans be captured on location?
Yes. Visualskies’ head and full body cyber rigs are mobile. For Another Simple Favour the rig was driven from London to Cinecittà Studios in Rome, with full body scanning on the first day and head and FACS scanning on the second.
