Digital Human Production: From Capture to a Real-Time Character

Digital human production combines capture, character modelling, facial rigging, animation and real-time integration into one connected pipeline.

A digital human can be a rendered film character, a real-time presenter, a game actor, a virtual influencer or an interactive AI avatar. Whatever the final use, the visual quality depends on how scanning, modelling, textures, hair, rigging, animation and integration work together.

Treating each stage as a separate handoff often creates mismatched topology, expressions, shaders or performance data. A digital human production plan keeps the character usable from the first capture session through to final delivery.

At Mimic Productions, scanning, character creation, motion capture, facial rigging, animation and real-time integration sit inside one pipeline. This guide walks through each stage and the decisions that shape the result.

Table of Contents

• Digital Human Production Is a Connected Pipeline

• Step 1: Define the Final Use

• Step 2: Capture Appearance and Performance

• Step 3: Build the Character

• Step 4: Rig the Face and Body

• Step 5: Animate and Integrate in Real Time

• Step 6: Validate the Finished Digital Human

• Comparison Table

• Where Digital Humans Are Used

• Frequently Asked Questions

• Conclusion

Digital Human Production Is a Connected Pipeline

A digital human is not one asset made in one step. It is the result of several specialist stages that depend on each other.

Each stage passes something to the next:

• Capture provides shape, surface detail and performance

• Modelling turns that data into clean, usable geometry and textures

• Rigging gives the face and body controls that deform correctly

• Animation brings motion, expression and speech

• Integration places the character in an engine, renderer or interactive system

When these stages are planned together, problems are caught early. When they are treated as separate jobs, small mismatches add up: a face mesh that does not suit the facial rig, textures too heavy for the engine, or capture data that does not fit the final skeleton.

Step 1: Define the Final Use

Start with the platform, camera distance, interaction model and performance needs. A pre-rendered commercial can use heavier geometry and shaders than a real-time kiosk or mobile avatar.

Answer these questions before any capture is booked:

• Will the character be rendered offline or run in real time?

• How close will the camera get to the face?

• Does the character need speech, facial performance and body motion?

• Will it be animated by artists, driven by capture, or controlled live?

• Does it need multiple languages or AI conversation?

• Which engine, renderer or device is the final target?

The answers set the polygon budget, texture resolution, rig complexity and capture method for everything that follows.

Step 2: Capture Appearance and Performance

Capture records two different things: what a person looks like and how they move.

Appearance capture uses 3D body scanning and 3D face scanning to record body shape, face geometry and surface detail. Methods include photogrammetry and structured light, often with a set of facial expressions captured as separate scans.

Performance capture records movement. Body motion capture, facial capture and reference video can be recorded separately or together. Productions may use optical systems, inertial systems or a hybrid setup, as explained in our guide to optical vs inertial motion capture.

Capture choices should match the final rig and animation pipeline. Recording facial expressions that the rig will never use, or body data that does not fit the final skeleton, adds cost without adding quality.

Step 3: Build the Character

Artists clean the scan or sculpt, then build the production character:

• Topology: clean, animation-ready geometry through retopology

• UVs: a planned layout for texturing, covered in UV mapping explained

• Textures and materials: skin, eyes and teeth tuned for the target renderer

• Hair and clothing: grooms and garments built for the final platform

• Real-time budgets: polygon counts, texture sizes and levels of detail

This stage decides whether the character can look convincing and still meet platform limits. Skin realism in particular depends on careful texture and shader work, which our guide to 3D character texturing and skin realism explains in detail.

Step 4: Rig the Face and Body

The body and face receive controls for animation or live tracking. Body and facial rigging usually combines joints for broad movement with blend shapes for detailed facial expression.

Teams test the rig before animation begins:

• Expressions and speech shapes

• Eye movement and blinks

• Shoulders, elbows, wrists and fingers

• Clothing and hair behaviour under motion

Some projects use a standard framework such as MetaHuman for parts of this stage. Others need a fully custom rig, especially for a specific real person or a hero character seen in close-up.

Step 5: Animate and Integrate in Real Time

With the rig approved, the character is animated from keyframes, capture data or live input, then integrated into Unreal Engine, Unity or a rendering pipeline through real-time integration.

For interactive characters, the visual layer must also connect to:

• Voice input and speech recognition

• Dialogue logic or an AI system

• Text-to-speech and lip sync

• Latency controls so responses feel natural

• The final interface, kiosk, app or display

This is the stage where AI avatars come together: a believable face, a responsive voice and a performance that holds up in real time.

Step 6: Validate the Finished Digital Human

Test the character under final lighting, camera distance and performance conditions before release.

Check:

• Facial readability and expression range

• Lip sync and eye focus

• Deformation at the shoulders, hands and neck

• Hair and clothing in motion

• Frame rate and memory on the target device

Keep a clear handover package for future updates: source files, textures, rig documentation and engine project. A well-documented digital human can be updated, relit or reused for years.

Comparison Table

Stage	Main output	Key decision	Common risk
Define final use	Technical brief	Platform, camera distance, interaction	Building for the wrong target
Capture	Scans and performance data	Methods that match the rig	Data that does not fit the pipeline
Build	Mesh, UVs, textures, hair	Detail versus budget	Heavy assets that fail in real time
Rig	Body and facial controls	Joints, blend shapes or hybrid	Poor deformation in close-up
Animate and integrate	Working character in engine	Keyframe, capture or live control	Latency and lip sync issues
Validate	Approved, documented asset	Test conditions and sign-off	Problems found after launch

Where Digital Humans Are Used

The same production pipeline serves very different uses:

• Film and VFX: digital doubles and characters for close-up shots

• Games: hero characters and cinematics

• Advertising: brand ambassadors and campaign characters

• Customer experience: real-time presenters and kiosk hosts

• Training and healthcare: guides that explain clearly and consistently

• Live events and XR: characters performing on stage or in headsets

The pipeline stays the same. What changes is the balance between visual detail, real-time performance and interaction.

Frequently Asked Questions

What is a digital human?

A digital human is a computer-generated person designed for rendered or real-time use. It may copy a real person or be created as an original character.

What does digital human production include?

The scope can include scanning, modelling, textures, hair, clothing, facial and body rigging, motion capture, animation, rendering and real-time integration.

Can a digital human speak in real time?

Yes. A real-time character can connect to speech recognition, a dialogue system, text-to-speech and facial animation, provided the full stack is designed for low-latency interaction.

Is MetaHuman the same as a custom digital human?

MetaHuman is a real-time character framework. A custom digital human may use a different pipeline or be integrated into MetaHuman, depending on identity, quality, rig and platform requirements.

How long does digital human production take?

It depends on realism, the number of expressions and outfits, the rig and the target platform. A simple real-time character is much faster than a photoreal digital double for film.

Do I need to scan a real person?

Only if the character must look like that person. Original characters can be sculpted from concept art, sometimes using scan data as a realism reference.

What is the difference between pre-rendered and real-time digital humans?

Pre-rendered characters are rendered frame by frame offline and can use heavier assets. Real-time characters run live in an engine and must fit strict performance budgets.

What should I prepare before starting a project?

Define the final platform, camera distance, required performance, languages, interaction model and deadline. References for appearance and personality also speed up the process.

Conclusion

Digital human production works best as one connected pipeline rather than a chain of separate jobs.

Define the final use first, capture only what the rig and platform need, build clean geometry and textures, rig the face for the performance required, then test everything under real conditions. Each decision supports the next.

To discuss a project, contact Mimic Productions or explore our 3D character services.

For inquiries, please contact: Press Department, Mimic Productions info@mimicproductions.com