iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Face scanning is not “better” than hand modeling in every game. Scanning is the fastest route to a recognizable actor likeness and fine skin detail, while hand-created modeling gives artists control over proportions, stylization, topology, and gameplay readability. A scan is an input asset, not a finished character: it normally needs retopology, cleanup, texture work, rigging, facial animation, and optimization. For that reason, shipped games commonly use a hybrid pipeline—capture or photograph a real face, then have artists redesign, animate, and optimize it for the game.
What face scanning and hand-created modeling actually mean
Face scanning and photogrammetry
Photogrammetry reconstructs a three-dimensional face from photographs or synchronized camera captures. The result can preserve a real person’s proportions, pores, wrinkles, asymmetries, and other details that would take an artist considerable time to sculpt from reference. A scan is best understood as source data: a high-resolution representation of a head that still has to be made usable for animation and real-time rendering.
“Face scanning” can describe several capture setups. A photogrammetry rig may use many cameras firing at once, while a performance-capture system records facial movement with cameras, markers, or both. These are related but separate jobs. One captures the shape and surface of a neutral head; the other records how a performer moves, speaks, and emotes.
Hand-created character modeling
In a hand-created workflow, an artist sculpts or models the face directly from concept art, photographs, anatomical references, or a director’s instructions. The artist decides the silhouette, facial planes, stylization, age, proportions, topology, and level of exaggeration. A model can be deliberately attractive, strange, heroic, cartoon-like, or non-human without being constrained by a real person’s anatomy.
#1 Best Overall
“Hand-created” does not mean every vertex is placed without digital assistance. Artists may use digital sculpting, procedural tools, reference planes, and scanned textures. The defining difference is authorship and control: the final shape is designed for the character and the game rather than being reconstructed primarily from a captured subject.
The hybrid approach
A hybrid pipeline starts with scan data or photographs, then treats that material as reference or a high-detail base. Artists adjust the likeness to match the approved character design, rebuild topology for deformation, correct areas that cameras did not see clearly, and create the facial poses the game needs. This approach combines realistic surface information with deliberate design decisions.
How a scanned face becomes a game-ready character
The following stages explain why a scan cannot simply be dropped into a game engine.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
1. Capture the neutral head and reference material
The subject is photographed or scanned under controlled lighting and a consistent expression. A studio may also capture color charts, hair and eye references, multiple expressions, and side or rear views. The purpose is to obtain enough information to reconstruct shape and appearance, not to produce a final in-game mesh.
NVIDIA’s Universal Capture example in GPU Gems 3 illustrates the scale of a performance setup: eight infrared cameras, three synchronized high-definition color cameras, and approximately 70 facial markers. That is one documented system, not a universal hardware requirement. Its significance is that geometry capture and motion capture are coordinated but distinct data streams.
2. Reconstruct the surface
Software combines the images into a dense head model and projects color information onto it. Dense reconstruction can contain tens of thousands of points or more, along with holes, noise, overlapping surfaces, and areas hidden by hair, eyelids, nostrils, or the mouth. The mesh records what the cameras observed; it does not understand which parts must bend, slide, or remain rigid during an expression.
3. Clean and retopologize
Artists remove spikes and scanning artifacts, close gaps, smooth inconsistent areas, and create a new topology suitable for deformation. Retopology replaces an irregular capture mesh with intentional edge loops around the eyes, lips, cheeks, and jaw. Those loops let a rig or blendshape system produce predictable expressions and prevent texture stretching.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Ubisoft Montréal described this problem while developing Faceshifter. The team began with scans of about a thousand heads; a single high-resolution head contained tens of thousands of data points. The studio said scans were often messy and time-consuming to turn into animated characters, so it cleaned them, wrapped them into polygons, detected facial landmarks, and assembled a reusable database.
4. Repair what cameras could not capture
Occluded regions and moving features need artistic correction. The inside of the mouth, the contact between lips, the upper eyelid, the ears, and parts of the nostrils are frequently unreliable in a neutral scan. Epic’s digital-human documentation for Unreal Engine 4.27 describes scan data as a source of macro detail while artists correct the mouth, upper eyelid, and ears. These corrections are not a failure of photogrammetry; they are the design work required to make captured information behave correctly.
5. Build materials and textures
Captured color must be converted into game textures. Artists separate or author maps for base color, roughness, normal detail, and other material properties, then remove lighting baked into the photographs. They also balance skin response so that pores remain visible without making the face look dirty or overly sharp under every light. Hair, eyebrows, eyelashes, teeth, and eyes usually require their own assets and shading decisions.
6. Create the facial rig and expressions
The cleaned mesh needs controls that animation can drive. Depending on the project, those controls may be bones, blendshapes, a FACS-inspired pose library, a machine-learning solver, or a combination. A neutral scan supplies shape; it does not supply the complete set of mouth, eye, cheek, and jaw deformations needed for speech and emotion.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware match7. Optimize for the target hardware
Dense capture data is expensive to render and animate. Teams reduce polygon counts, establish level-of-detail versions, compress textures, simplify animation data, and test the result at the game’s target frame rate. The highest-resolution scan therefore may exist only as a source or cinematic asset, while gameplay uses a carefully optimized representation.
How a hand-created face is built
Start from the character brief
The artist begins with the approved concept: age, ancestry, occupation, emotional range, silhouette, and how recognizable the character must be from a distance. Unlike a scan, the design can intentionally depart from real anatomy. A hero can have enlarged eyes for readability, a villain can have sharpened planes, and a creature can combine human and animal structures.
Sculpt primary, secondary, and tertiary forms
Primary forms establish the skull, jaw, cheekbones, brow, and muzzle. Secondary forms define eyelids, lips, nasolabial folds, and muscles. Tertiary forms add pores, wrinkles, scars, and other surface information. The artist can place each layer where it supports the character’s identity and camera distance rather than preserving every incidental feature of a real subject.
Design topology around movement
Because the artist controls the base mesh, edge loops can be planned for blinking, smiling, speech, squinting, and jaw rotation from the beginning. This reduces the amount of corrective repair later. It also makes it easier to create a stylized model whose topology supports exaggerated expressions instead of forcing a realistic scan to perform actions it was never shaped for.
Recommended Free Tools
Author expressions and gameplay timing
Hand-authored poses let animators decide exactly when an expression appears, how long it holds, and how it blends with dialogue, camera cuts, player input, and scripted events. The trade-off is labor: building a convincing library for many characters requires substantial sculpting and animation time, especially when each face has a distinct anatomy.
Facial animation is a separate production problem
A scanned likeness can look perfect in a still image and still fail when it speaks. Facial animation must solve deformation, timing, eye focus, lip contact, teeth visibility, breathing, and transitions between emotions. Capture can provide performance nuance, but the data still has to be solved onto the game mesh and corrected.
Performance capture and marker solving
Autodesk defines performance capture as the evolution of motion capture in which an actor’s body movements, gestures, facial expressions, and voice are captured simultaneously through sensors, visual markers, or advanced camera techniques. In NVIDIA’s Universal Capture example, marker motion drives deformation of a static scanned and surfaced head. Additional animator keying handles details such as the inner-lip contour. The example demonstrates that a scanned head and captured performance are separate pipeline stages.
FACS and pose libraries
Many systems organize facial motion into combinations of anatomical action units. Epic’s Unreal Engine 4.27 digital-human documentation records 50 FACS poses. Those poses can be combined, corrected, and driven by dialogue or performance data. A project may use fewer or more controls; the useful number depends on the art style, language coverage, camera distance, and required emotional range.
Hand correction remains necessary
Solvers can misread a cheek movement, lose the lip seal, or produce an eyelid that intersects the eye. Animators correct those errors and author poses that capture did not produce cleanly. Motion graphs can connect captured clips with hand-animated facial clips, allowing expressions to blend and to be triggered by artificial-intelligence systems or player input.
What major game examples reveal
Photogrammetry can preserve likeness while artists retain authorship
Capcom director Koji Nakanishi explained the decision for Resident Evil 7’s Chris Redfield this way: “In order to achieve that kind of photorealistic quality while respecting Chris’ character, we decided to use a technique known as photogrammetry.” He also explained that designers adjusted the scanned face to match the intended Chris Redfield design. The scan supplied realistic information; artists remained responsible for the final character.
Large scan libraries need procedural organization
Ubisoft Montréal’s Faceshifter work shows the scaling challenge. Scanning about a thousand heads creates a valuable variety of faces, but each head still needs cleanup, landmark alignment, and animation-ready structure. The studio’s André described the goal this way: “With a system like Faceshifter, we can generate heads in a few seconds, which therefore allows us to generate, directly in the video game, lively and diverse crowds of various sizes, made up of a bunch of different faces.” Reusable components and automated processing make a library practical; they do not eliminate artistic review.
Hand-sculpted poses can support long-form emotional direction
PlayStation Blog reported that The Last of Us Part II used nearly 20 emotional states for 25 key characters and about 15,000 individual hand-sculpted poses. Keith Paciello described sculpting expressions for joy, sadness, fear, anger, surprise, disgust, and contempt, then layering eye motion, breathing, look-at behavior, and script triggers over the facial model. He said, “We could emotionally pace out our characters from the absolute beginning of the game to the absolute end of the game and blend in and out of cinematics seamlessly.” This illustrates a gameplay-animation objective that a raw scan cannot provide by itself.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsFace scanning versus hand modeling: the practical trade-offs
| Decision factor | Face scanning | Hand-created modeling | Hybrid workflow |
|---|---|---|---|
| Likeness fidelity | Strong when a real actor or model must be recognizable; captures individual asymmetry and surface detail. | Depends on the artist’s reference interpretation and time; can reproduce a likeness but is not constrained to one person. | Uses the scan as a likeness and detail source while allowing approved design changes. |
| Design control | Limited until artists reshape the captured result; the source reflects the subject’s anatomy. | Direct control over proportions, stylization, exaggeration, and creature design. | Artists retain control over the final proportions and silhouette after capture. |
| Surface detail | Efficiently records pores, wrinkles, scars, and other small features under suitable capture conditions. | Can be sculpted or textured deliberately, but detail authoring takes time. | Combines captured macro and skin detail with authored corrections. |
| Topology and deformation | Requires cleanup, retopology, landmark alignment, and correction of occluded areas. | Topology can be designed around animation from the start. | Retains scan fidelity while rebuilding the mesh for predictable deformation. |
| Performance nuance | Can preserve an actor’s distinctive movement when paired with performance capture, followed by solving and cleanup. | Animators have precise control over every pose and timing decision. | Captured clips and hand-authored poses can be blended through motion graphs or other systems. |
| Stylization and altered proportions | Requires substantial redesign if the final character departs from the scanned subject. | Well suited to stylized characters, creatures, and intentional exaggeration. | Useful when realism is desired in materials but the design needs nonliteral proportions. |
| Scale and iteration | Scan libraries can seed many faces, but high-resolution processing and cleanup become a production task. | Each face is designed individually, giving control but increasing manual labor. | Procedural tools can accelerate variation while artists supervise the result. |
| Runtime suitability | Raw capture is usually too dense or inconsistent for direct real-time use and must be optimized. | Models can be built to the target budget from the outset. | Uses optimized game meshes and textures derived from richer captured sources. |
When scanning is the better starting point
- A recognizable performer matters. A scan captures the facial relationships that audiences associate with a real actor more efficiently than building every feature from scratch.
- The project emphasizes photorealism. Skin microdetail and natural asymmetry provide a convincing base for close-up cameras.
- You need many realistic background faces. A library of scans can provide variation, especially when automated landmarking and reusable components are available.
- Actor-specific performance is important. Facial capture can preserve subtle timing and movement that an animator would otherwise need to interpret manually.
When hand modeling is the better choice
- The character is stylized or non-human. Scanning a real face adds little value when the design intentionally breaks human proportions.
- Gameplay readability outranks literal realism. Enlarged features, clearer brows, or stronger mouth shapes may communicate emotion better at distance.
- The design is still changing. A hand-built model can be reshaped quickly around new concept decisions without repeatedly processing capture data.
- Precise authored emotion is the priority. Hand-sculpted poses give animators exact control over holds, transitions, and scripted beats.
- The target hardware is constrained. A deliberately economical mesh and texture set can be planned around the runtime budget instead of reduced from a dense source later.
Why studios usually combine the methods
The choice is rarely a contest between a scanner and an individual artist. Production teams divide the work according to what each method does best. Capture supplies measurable likeness, natural irregularity, and a starting point for realistic materials. Artists supply design intent, clean topology, corrected anatomy, animation controls, and performance timing.
A practical hybrid sequence is: scan the actor, approve the likeness, reshape the head to match the character brief, rebuild the facial topology, project and clean textures, create a pose or FACS library, solve performance data, hand-correct difficult expressions, and optimize the final assets. The scan may remain as a high-resolution source while the game uses several lower-detail representations.
Common mistakes and how production teams avoid them
Treating the scan as a finished character
Symptom: the head looks impressive in a viewer but collapses when it talks or blinks. Fix: schedule retopology, landmark alignment, corrective shapes, rigging, and animation as planned production stages rather than optional polish.
Preserving likeness at the expense of the character design
Symptom: the actor is recognizable, but the result no longer resembles the approved fictional character. Fix: compare the scan against the design brief early and let artists adjust proportions, planes, and features before materials and animation are finalized.
Free tools Windows power users keep installed
One-click scans. No signup required.
Assuming capture records hidden anatomy
Symptom: lips, eyelids, ears, or the inside of the mouth deform unpredictably. Fix: inspect occluded regions, repair them manually, and build corrective poses for extreme expressions.
Using dense data directly at runtime
Symptom: frame rate, memory use, or animation evaluation becomes unacceptable. Fix: create optimized topology, texture resolutions, level-of-detail meshes, and compressed animation representations for the target platform.
Expecting a solver to replace animators
Symptom: captured dialogue is technically synchronized but emotionally flat, with drifting eyes or weak lip shapes. Fix: combine solved performance with hand-authored corrections, eye and breathing layers, look-at behavior, and scripted emotional timing.
Scaling a scan library without consistent standards
Symptom: crowd heads differ in landmarks, topology, texture space, or expression controls, making them difficult to swap. Fix: standardize landmark definitions, polygon wrapping, naming, texture conventions, and validation before adding more scans.
A decision framework for a new game project
- Define the identity requirement. Decide whether players must recognize a real performer, a fictional design, or only a readable archetype.
- Set the visual target. Establish camera distance, lighting conditions, platform limits, and whether close-up photorealism is required.
- List the animation needs. Identify dialogue coverage, languages, emotional states, player-driven expressions, cinematics, and crowd behavior.
- Choose the source strategy. Select scanning for likeness and surface detail, hand modeling for control and stylization, or both for a hybrid result.
- Prototype the hardest expression. Test speech, blinking, squinting, jaw motion, and an extreme emotion before processing an entire cast.
- Budget the conversion work. Include cleanup, retopology, textures, rigs, pose libraries, solver correction, optimization, and quality assurance—not only capture-day costs.
- Validate in the game context. Review the face under gameplay lighting, at intended camera distances, during real dialogue, and alongside the body and hair assets.
Bottom line
Face scanning wins at starting from a convincing real likeness and detailed skin; hand-created modeling wins at intentional design, stylization, topology, and authored gameplay expression. Neither method removes the need for facial animators. The most reliable production pattern is to scan or photograph when that improves fidelity, then have artists redesign, clean, rig, animate, and optimize the result until it serves the character and the game.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
