DEV Community: Mimic Minds

How Hologram Studios Build Human Holograms

Mimic Minds — Wed, 24 Jun 2026 16:34:12 +0000

From capture and rigging to projection geometry and show control, this guide breaks down the technical pipeline behind a convincing holographic performance.

The most convincing holographic performances look deceptively simple. A human figure appears onstage, addresses an audience, moves with believable weight, and seems to occupy the same space as a live performer.
Behind that moment is a chain of disciplines: rights clearance, 3D capture, character modeling, facial rigging, body motion capture, hair and cloth work, lighting, rendering, projection geometry, audio, networking, and show control.

For developers and technical artists, the interesting question is not whether the result looks futuristic. It is how the complete system preserves identity, timing, depth, and performance intent from the first scan to the final pixel.

First, what are holograms?

In strict optical terms, holography records information about a light wavefront and reconstructs it through interference and diffraction. Dennis Gabor received the 1971 Nobel Prize in Physics for the invention and development of the holographic method. His work describes a process in which object light and a reference beam create an interference pattern that can later reconstruct an image with depth.

In entertainment and experiential production, the word is used more broadly. Many displays described as holographic use reflection, projection, transparent media, LED surfaces, or volumetric techniques rather than classical wavefront holography.

That distinction is useful, not pedantic. It affects the asset format, camera setup, viewing zone, latency target, lighting plan, and whether the content can react in real time. A detailed introduction to what are holograms is a useful starting point, but a production team must still identify the exact display technology before building the character.

A believable holographic figure begins as a coordinated character, performance, rendering, and display pipeline.

The word “hologram” covers several display systems

A production brief may say “hologram” while referring to one of several very different systems.

True optical holography

Classical holography records and reconstructs light through interference and diffraction. Modern computer-generated holography calculates wavefront patterns for spatial light modulators or related optical systems. This is the technically strict use of the term.

The engineering challenge is optical. The system must reconstruct light with the right phase and direction so the viewer receives depth cues from the intended viewing zone.

Stage projection and reflection illusions

Large entertainment productions often create a human figure using a bright rendered image, carefully controlled black levels, and a transparent or reflective surface positioned relative to the audience. The content may be pre-rendered video or a live engine output.

The engineering challenge is compositional. Camera perspective, stage geometry, background darkness, brightness, reflection angle, and performer blocking have to work together. The audience is not viewing a free-floating object from every position. It is viewing a highly controlled illusion from a planned zone.

Volumetric and light-field displays

Volumetric systems generate visible samples within a physical volume, while light-field displays deliver different rays or views in different directions. These approaches can support more natural parallax, but they introduce constraints around resolution, capture format, compute, viewing area, and display size.

Before production begins, identify which display method the project actually uses.

A studio should therefore ask one question before discussing character quality: What does the display need from the content?

Why human holograms are difficult

A floating logo or abstract product can tolerate simplification. A human face cannot.

Viewers are highly sensitive to facial identity, eye direction, speech timing, posture, weight shifts, and emotional coherence. A few frames of lip-sync drift may be more damaging than a moderate reduction in texture resolution. A perfectly modeled face can still feel wrong if the smile shape, blink rhythm, or head motion does not match the performance.

The term human holograms therefore describes more than a display effect. It implies a digital-human pipeline capable of preserving a recognizable person or creating a believable new performer.

That pipeline has to solve four linked problems:

Identity: Does the model preserve the face, proportions, age, and defining traits?
Deformation: Does the face and body move correctly from every required angle?
Performance: Do speech, gesture, expression, and timing feel intentional?
Display translation: Does the final image survive the brightness, contrast, perspective, and viewing constraints of the venue?

The scan-to-stage production pipeline

1. Define rights, intent, and display constraints

Before scanning or modeling, define who controls the likeness, which performances are authorized, where the asset may appear, and how long the rights remain valid.

The technical brief should also capture:
• display type and physical dimensions
• expected viewing distance and audience zone
• frame rate, resolution, and color format
• pre-rendered or interactive behavior
• live performer interaction
• network and compute limits
• fallback behavior if the system fails

A beautiful character built for the wrong camera and viewing geometry is not production-ready.

2. Capture or construct the digital human

A real person may be captured using photogrammetry, facial scanning, structured light, calibrated photography, or a combination of these methods. Historical or unavailable subjects may require reference-based reconstruction from photographs, footage, casts, and expert review.

The resulting geometry is usually too dense or inconsistent for animation. Artists retopologize it into a mesh with deformation-friendly edge flow, create UVs, and build texture sets for skin color, roughness, normal detail, displacement, eyes, teeth, and clothing.

3. Build facial and body rigs

The body rig needs stable shoulders, hips, spine, hands, and feet. The facial rig needs enough control to reproduce speech and identity-specific expressions without turning the face into a generic puppet.

Common facial approaches include:
• blend shapes or pose-space deformations
• joint-based controls
• muscle-inspired solvers
• scan-derived expression shapes
• hybrid rigs that combine several methods

The rig should be tested with the target performance range early. A rig that looks correct in a neutral pose may fail under a wide smile, compressed lips, cheek movement, or rapid speech.

4. Capture and direct the performance

Body motion capture provides timing, weight, and gesture. Facial capture records expressions, mouth movement, head motion, and sometimes eye behavior. Voice may be recorded simultaneously or replaced later.

Raw capture is not the finish line. It needs cleanup, retargeting, contact correction, timing adjustments, and artistic direction. The performance must also fit the display. A subtle gesture that works in a close-up may disappear when the figure is ten meters from the audience.

5. Simulate hair and clothing

Hair and fabric communicate scale and motion. They also create expensive simulation and rendering problems.

A pre-rendered show can use higher-cost strand hair and cloth simulation. A real-time system may need groom simplification, cards, bone-driven secondary motion, reduced collision, or baked caches. The correct method depends on whether the character must react live.

6. Light for the destination, not only the render

A holographic stage image often needs deeper blacks, carefully shaped rim light, and stronger silhouette separation than a normal film render. Brightness and contrast have to survive the display medium and ambient venue lighting.

The team should test on the actual hardware as early as possible. A standard monitor cannot reveal all projection artifacts, viewing-angle limits, reflected light, ghosting, or black-level problems.

7. Adapt perspective and framing

The virtual camera must correspond to the physical stage and audience. A mismatch can make the figure appear tilted, stretched, floating too high, or disconnected from the floor.

For a fixed-view illusion, lock the camera to the intended audience perspective. For multi-view displays, prepare the scene and capture volume for the required parallax range.

8. Integrate show control

The final character must synchronize with audio, lighting, stage automation, live performers, video servers, or interactive systems. Timecode may be sufficient for a pre-rendered piece. Real-time experiences need events, state management, monitoring, and cancellation behavior.

The pipeline has multiple handoffs. Identity and timing can be lost at any one of them.

Pre-rendered or real-time?

The choice changes almost every downstream decision.

Use pre-rendered content when:

• the performance is fixed
• maximum image quality matters more than interaction
• the venue requires deterministic playback
• the show has tightly rehearsed timing
• compute or networking is limited

Pre-rendering allows heavier simulation, more samples, complex compositing, and careful frame-by-frame review.

Use real-time content when:

• the character must respond to an operator or audience
• dialogue changes between performances
• the system reacts to sensors or live data
• the character shares a stage with unscripted participants
• personalization is central to the experience

Real-time work trades some visual headroom for responsiveness. Asset budgets, shader complexity, hair, cloth, lighting, and animation systems must fit a stable frame budget.

A hybrid design is often best. Render the visually expensive sequences in advance, then switch to a real-time character for interactive sections.

Synchronizing a real-time holographic character

A live digital performer is a distributed system. Audio, dialogue, facial animation, gestures, rendering, stage cues, and display output may run in different processes or machines.

The safest design uses a shared clock, unique response IDs, and explicit cancellation.

Here is a simplified event model:
type HologramEvent = | { type: "speech.started"; responseId: string; atMs: number } | { type: "audio.chunk"; responseId: string; atMs: number; url: string } | { type: "viseme"; responseId: string; atMs: number; value: string } | { type: "gesture"; responseId: string; atMs: number; name: string } | { type: "lighting.cue"; responseId: string; atMs: number; cue: number } | { type: "response.cancel"; responseId: string; atMs: number } | { type: "response.completed"; responseId: string; atMs: number };
Every component should reject events whose responseId is no longer active. That prevents late audio, lip-sync, or gestures from a previous turn from reaching the stage.

A cancellation path might look like this:
function cancelActivePerformance(responseId: string): void { audioPlayer.stop(responseId); faceController.clear(responseId); gestureController.blendToIdle(responseId, 120); showControl.cancelCues(responseId); renderer.setListeningState(); }
The important point is architectural: stopping the language response is not enough. The audio, face, body, lights, and stage cues associated with that response must stop together.

Common failure modes

Where hologram studios add value

Specialized hologram studios sit between character production, VFX, animation, display engineering, and live show delivery. Their value is not simply creating a 3D person. It is translating that person into the visual and operational constraints of a particular holographic system.

That includes decisions such as:
• whether digital face replacement is preferable to a full digital body
• how much facial detail the final display can preserve
• which hair and cloth methods will survive the frame budget
• how to light for transparency and black levels
• how to calibrate camera perspective to the venue
• how to monitor a live performance and recover from failure

Mimic Productions, for example, describes a workflow that combines 3D scanning, facial rigging, animation, motion capture, digital face replacement, and holographic production for entertainment, exhibitions, and brand experiences.

For readers who want a wider overview of applications, types, and limitations, this guide to holograms covers entertainment, museums, corporate events, retail, education, medical visualization, and industrial use.

Rights, disclosure, and digital resurrection

Reconstructing a living or deceased person is not only a technical project. It raises questions about consent, estate approval, performer rights, voice rights, cultural context, and the limits of the authorized performance.

A responsible production should document:
• who authorized the likeness and performance
• which source materials may be used
• whether synthetic voice or dialogue is permitted
• how the character will be labeled to audiences
• which edits require additional approval
• where and for how long the asset may appear
• how the files and training data will be secured

The more lifelike the result, the more important the disclosure. Audience members should be able to understand whether they are seeing archival footage, a digital recreation, a live actor with face replacement, or an interactive synthetic performance.

A practical preflight checklist

Before final delivery, test the complete experience in venue-like conditions.

Character

• facial identity holds across the required expression range
• eyes, teeth, tongue, and lips survive close views
• body proportions and floor contact are believable
• hair and clothing remain stable for the complete show

Display

• camera and perspective match the physical setup
• black levels and edge treatment are clean
• the audience viewing zone is documented
• ambient lighting does not destroy the illusion
• color and brightness are tested on final hardware

Performance

• dialogue, lips, gesture, and body timing stay synchronized
• transitions into and out of the figure are rehearsed
• live performers know the figure’s apparent position
• interruption and operator controls are tested

Reliability

• playback or engine health is monitored
• a fallback state exists for network or model failure
• the show can recover without restarting the venue system
• logs identify which cue, response, or asset failed

Rights and safety

• likeness, voice, music, and source-material rights are cleared
• the audience receives appropriate disclosure
• interactive dialogue has a defined scope
• a human operator can intervene when needed

Final thoughts

The phrase “human hologram” makes the result sound like a display trick. In practice, the display is only the final layer.

A convincing performance depends on an end-to-end system that protects identity, deformation, timing, visual quality, and rights across many handoffs. The best projects begin with the display constraints, build the digital human for those constraints, and test the complete experience early.

For developers, the main lesson is familiar: optimize the system, not only the component. A stronger face model cannot fix bad time synchronization. A faster dialogue model cannot fix poor stage calibration. A brighter projector cannot fix an unconvincing performance.
When character craft, real-time engineering, and show control are designed together, the technology becomes less visible. What remains is the intended effect, a believable presence sharing the room with the audience.

About Mimic Productions: Mimic Productions is a Berlin-based 3D studio specializing in character creation, scanning, motion capture, animation, real-time assets, and digital-human production.

Designing Production-Grade Avatars for AI Agents, Games, Business, and Robotics

Mimic Minds — Wed, 27 May 2026 11:46:28 +0000

Most avatars are still treated as visual assets. Someone asks for a face, a body, a voice, or a character style, then the asset is handed over to a product team.

That approach works for prototypes. It does not work for production.

A production-grade avatar is closer to an interface system than a static character. It has to communicate intent, match the product context, respect technical limits, and behave consistently across long sessions. The same core discipline can support a business avatar, an AI agent avatar, production-grade avatars for gaming, or robot avatars, but each context changes the design requirements.

This article breaks down how to think about avatars as systems: visual identity, runtime behavior, integration logic, performance constraints, and the failure modes that appear when the avatar is only treated as a cosmetic layer.

The avatar is not the product skin, it is part of the interface

When users interact with an avatar, they are not only reacting to its appearance. They are reading signals:

• Does this character represent the brand or product accurately?
• Does its expression match the situation?
• Does it feel responsive or delayed?
• Does the voice match the face?
• Does it know when to answer, pause, escalate, or stay quiet?

Those questions are interface questions. A strong avatar pipeline therefore needs both art direction and system design.

A simple production model looks like this:

Product goal

-> Persona specification
-> Visual identity
-> Voice and tone model
-> Rig and animation grammar
-> Runtime integration
-> Safety and escalation logic
-> QA across real user scenarios

The visual layer matters, but it is only one part of the system. A polished render will not save an avatar that interrupts at the wrong time, gives unsupported answers, or breaks character after three user turns.

A shared pipeline, different deployment contexts

Business avatars, AI agent avatars, gaming avatars, and robot avatars can share a similar production backbone:

Concept design

-> Character modeling
-> Texturing and shading
-> Rigging
-> Animation states
-> Voice or speech layer
-> Runtime export
-> Integration testing

The difference is where the pressure appears.

A business avatar is judged on trust, brand consistency, and clarity. An AI agent avatar is judged on conversation quality, memory boundaries, and escalation behavior. A game avatar is judged on deformation, responsiveness, engine performance, and customization. A robot avatar is judged on safety, legibility, and whether non-technical users can understand what the machine is doing.

The asset may look similar in a portfolio image. In production, these are four different interface problems.

Business avatars: brand presence as a reusable interface

A business avatar usually represents a company, product, or expert role. It may appear in onboarding flows, product demos, support journeys, retail screens, internal training, or customer education.

The mistake is to treat it like a generic presenter. A useful business avatar needs a defined operating range:

• What topics can it explain?
• What tone should it use with a new user versus an enterprise buyer?
• When should it hand off to a human?
• Which expressions are allowed for serious, technical, or sensitive topics?
• How does it remain recognizable across web, mobile, video, and kiosk contexts?

For a developer or product team, the important point is consistency. The avatar should not become a different character every time it appears in a new channel. That requires a persona brief, approved language patterns, expression rules, voice direction, and a deployment plan.

In a production system, the business avatar can sit on top of multiple back-end services. The front end may show the same character while the request routes to documentation search, CRM data, human support, or a scripted onboarding flow.

User request
-> Channel context
-> Intent classification
-> Approved knowledge source
-> Response policy
-> Avatar delivery layer

The avatar is the visible layer, but the trust comes from the routing and policy behind it.

AI agent avatars: conversational systems need behavior, not just a face

Adding a face to an AI agent changes user expectations. A text chatbot can feel like a utility. A visual agent feels more like a representative. That means mistakes carry more weight.

A practical AI agent avatar stack often looks like this:

User input
-> ASR or text input
-> Intent router
-> Context and memory layer
-> Safety and escalation policy
-> Response planner
-> TTS or text output
-> Facial animation and gesture layer
-> Avatar runtime

The key layer is not the face. It is the policy layer.

The system needs to know what the avatar is allowed to say, what it should never imply, when it should ask for clarification, and when it should escalate. Without that layer, the avatar may appear confident even when the system is uncertain.

That is dangerous from a UX perspective. A realistic face can increase perceived authority. The system should not visually overstate certainty.

Good AI avatar design includes constraints such as:

• Clear uncertainty behavior, for example, "I do not have enough information to answer that."
• Human handoff paths for support, sales, medical, financial, or operational issues.
• Memory controls that define what can be stored, retrieved, and forgotten.
• Tone boundaries that prevent the agent from drifting away from the brand persona.
• Animation states for listening, thinking, speaking, waiting, and escalation.

The avatar should make the AI easier to understand, not harder to question.

Gaming avatars: the character must survive the engine

Game avatars are different because users do not only watch them. Users inhabit them.

That creates a stricter production standard. The character must hold up during movement, camera changes, customization, lighting variation, combat, emotes, network conditions, and repeated close-up views.

For game teams, the technical checklist usually includes:

• Clean topology that deforms well around the face, shoulders, elbows, hands, hips, and knees.
• LOD versions that preserve identity while reducing cost at distance.
• Texture sets designed for the target platform, not only for beauty renders.
• Rig specifications that match the animation system.
• Blend shape or joint-based facial controls for expressions and speech.
• Customization rules that do not break silhouette, clipping, or performance.
• Engine validation in the target environment, not only in DCC tools.

A character that looks excellent in a turntable can still fail in gameplay. The avatar needs to be tested where players will actually see it.

Art direction

-> Game mesh
-> Rig and deformation test
-> Animation state test
-> LOD test
-> Customization test
-> Engine lighting test
-> Runtime performance test

For player-facing systems, avatar quality is also tied to identity. Skins, cosmetics, expressions, and body variants are not only visual features. They are part of the player's relationship with the game world.

Robot avatars: the face is the translation layer

In robotics, the avatar often becomes the communication layer between a machine and a human.

The robot may handle movement, navigation, sensing, or task execution. The avatar explains what is happening in a form the user can read quickly. This matters in public spaces, healthcare, retail, industrial operations, and consumer devices.

Robot avatars need a different kind of clarity:

• What is the robot doing right now?
• What will it do next?
• Does it need permission?
• Is it blocked, waiting, navigating, or escalating?
• Can a non-technical user understand the system state?

The avatar should not hide machine limitations. It should expose them in a human-readable way.

A useful robot avatar can simplify interaction by translating internal states into visible behavior:

Robot state: path blocked
Avatar behavior: pause, look toward obstacle, explain the issue, ask user to clear path

Robot state: low confidence
Avatar behavior: reduce certainty, ask for confirmation, offer handoff or retry

Robot state: task complete
Avatar behavior: confirm completion, summarize next step, return to idle state
For robotics, the avatar is not only emotional polish. It is an operational interface.

What changes by avatar type

Avatar type Primary role Main technical pressure Common failure mode
Business avatar Brand representative Persona consistency, channel integration, approved messaging Looks polished but feels generic or off-brand
AI agent avatar Visual interface for an AI system Policy layer, memory rules, escalation logic, latency Appears too confident when the system is uncertain
Gaming avatar Playable or interactive character Topology, rigging, LODs, textures, engine performance Looks good in renders but breaks during gameplay
Robot avatar Human-readable machine interface State communication, safety cues, plain-language output Hides system limitations instead of explaining them

The lesson is that avatar quality cannot be judged from a still image alone. The real test is behavior under deployment conditions.

Practical checklist for production teams

Before building or buying an avatar system, answer these questions:

• What user problem does the avatar solve?
• What should the avatar never do?
• What is the avatar allowed to know, remember, and say?
• What runtime environment will it appear in?
• What are the latency constraints?
• What happens when the system is uncertain?
• What is the handoff path to a human, script, or safer workflow?
• How will the avatar be tested after deployment?
• Which metrics will define success: task completion, retention, support resolution, trust, conversion, or engagement?

These answers should shape the character brief before modeling begins.

Common implementation mistakes

The most common mistake is starting with visual fidelity and solving behavior later. That usually leads to a character that looks impressive in marketing material but feels disconnected in actual use.

Other mistakes include:

• Building one avatar for every channel without testing context-specific behavior.
• Letting an AI agent speak without a clear policy and escalation layer.
• Treating game avatars as cinematic assets instead of runtime assets.
• Giving a robot avatar too much personality when users mainly need state clarity.
• Ignoring accessibility, especially captions, plain-language output, contrast, and nonverbal alternatives.
• Using facial expressions that imply certainty, empathy, or urgency when the underlying system cannot support that signal.

A production avatar should never communicate more confidence than the system actually has.

A better way to evaluate avatar quality

Instead of asking only "Does it look real?" or "Does it look cool?", teams should ask:

Does it behave correctly in context?
Does it stay in character over time?
Does it degrade gracefully when the system is uncertain?
Does it work within platform constraints?
Does it make the product easier to understand?

That evaluation applies across business, AI, games, and robotics.

A high-quality avatar is not just a face. It is a directed character connected to a real system. It has visual design, behavior rules, technical constraints, safety logic, and a role in the product experience.
That is the difference between an avatar that looks good in a demo and an avatar that works in production.

Closing thought

The future of avatars is not one universal character that appears everywhere. It is a production discipline that adapts character design to context.

A business avatar needs trust. An AI agent avatar needs policy and clarity. A game avatar needs performance and identity. A robot avatar needs legibility and safety.

The same craft can support all four, but only when the avatar is treated as part of the system from the beginning.

Coach Avatar Technology: How 3D Face Scanning Is Transforming Digital Experiences

Mimic Minds — Fri, 10 Apr 2026 16:34:31 +0000

Introduction

The digital landscape is evolving quickly, with new technologies reshaping how people interact, learn, and engage online. One of the most exciting developments is the rise of the Coach Avatar, which enables personalized and interactive digital guidance.
At the same time, innovations like 3D face scanning and 3D body scanning are making avatars more realistic and immersive, enhancing user experiences across industries.

What Is a Coach Avatar?

A Coach Avatar is a digital character designed to guide, train, or assist users in various environments, such as education, fitness, or gaming. These avatars provide personalized support and improve user engagement.
With the integration of AI, coach avatars are becoming smarter, more responsive, and capable of delivering real-time feedback.

The Role of AI Game Avatar in Digital Interaction

Interactive Gaming Experience

An **AI Game avatar **enhances gameplay by offering dynamic and responsive interactions.

Personalized User Engagement

Players can connect more deeply with games when avatars adapt to their behavior and preferences.

Expanding Applications

Beyond gaming, AI Game avatar technology is also being used in training simulations and virtual environments.

How 3D Face Scanning Enhances Avatars

3D face scanning captures detailed facial features and converts them into accurate digital models.
This technology allows avatars to closely resemble real individuals, improving personalization and realism in digital experiences.

The Importance of 3D Body Scanning

3D body scanning complements facial scanning by capturing full-body details, creating a complete and lifelike avatar.
This enables more realistic movement, better customization, and enhanced applications in gaming, fitness, and virtual simulations.

Business Applications of Avatar Technology

Virtual coaching and training programs
Gaming and interactive entertainment
Personalized learning platforms
Digital fitness and wellness solutions A Coach Avatar is becoming a valuable tool for delivering engaging and scalable digital experiences.

About Mimic Minds

Mimic Minds specializes in developing advanced Coach Avatar solutions that enhance user engagement and personalization. Their expertise also includes creating intelligent AI Game avatar systems that improve interaction and digital experiences. By combining AI with creative design, Mimic Minds enables scalable and effective virtual solutions.

About Mimic Productions

Mimic Productions focuses on delivering high-quality 3D face scanning and 3D body scanning solutions. Their technology enables the creation of highly realistic digital avatars used in gaming, film, and virtual production. By combining precision with innovation, Mimic Productions delivers immersive and lifelike digital experiences.

FAQs

What is a Coach Avatar?

A Coach Avatar is a digital character designed to guide and assist users in training, learning, or interactive environments.

What is an AI Game avatar?

An AI Game avatar is a digital character that interacts dynamically with players in gaming or virtual environments.

How does 3D face scanning work?

3D face scanning captures facial details and converts them into accurate digital models.

Why is 3D body scanning important?

3D body scanning creates full-body digital representations, enhancing realism and customization.

Conclusion

The rise of the Coach Avatar, supported by technologies like 3D face scanning and 3D body scanning, is transforming how users interact with digital platforms. With the integration of AI Game avatar technology, businesses can create more engaging, personalized, and immersive experiences. Adopting these innovations is essential for staying ahead in today’s digital world.