Impact of Ai on Professional Practice and Creative Control in Virtual Production Workflows.

Megan Booth PhD Research resources – Arts University Bournemouth.

Virtual Production
TECH Stack

Physical Production Layer

LED Volume, Cameras, Tracking, Lighting

Real-Time Engine Layer

Unreal Engine, Rendering, Lighting, Physics

Data & Asset Layer

3D Assets, Textures, USD, Photogrammetry, Ai 3D

AI & Automation Layer

Generative AI, Scene Gen, Simulation, ML Tools

Pipline & Orchestration Layer

Sync, Colour, Versioning, Collaboration

1. Capture & Input Layer

(Physical reality → digital signal)

Purpose

This layer is where material reality enters the VP system.
It defines what kinds of reality can be performed, tracked, or reconstructed.

Core Components

  • Camera systems (cinema cameras, witness cams)

  • Motion capture (optical, inertial, markerless)

  • Camera tracking (optical / sensor fusion)

  • LiDAR & photogrammetry

  • Performance capture (body, props, environment)

Notes for framing

  • This is labour-intensive and embodied

  • Decisions here shape who can participate (performers, technicians)

  • AI does not replace this layer — it compresses or reinterprets it

2. Spatial Reconstruction & World Modelling

(Reality → computable space)

Purpose

This layer converts captured data into navigable, renderable worlds.

Tool Classes (not brands)

  • Photogrammetry pipelines

  • NeRF / neural scene reconstruction

  • Gaussian splatting

  • Traditional mesh + texture workflows

  • Hybrid digital twins

Key distinction

MethodStrength
PhotogrammetryMetric accuracy
NeRF / splatsPerceptual realism
Mesh pipelinesEngine control

Cultural implication

This layer redefines realism:

  • from measured geometry → perceived plausibility

  • a major shift in production epistemology

3. Real-Time Engine & Control Layer

(Decision-making core of VP)

Purpose

This is the operational heart of Virtual Production.

Dominant Engine

  • Unreal Engine

What happens here

  • Real-time rendering

  • Virtual camera operation

  • Lighting decisions

  • Set extension

  • Playback & iteration

  • On-set collaboration

Why this layer is central to research

  • It collapses departments (VFX, camera, art, lighting)

  • It moves authorship forward in time (decisions made earlier)

  • It creates new communities of practice (engine-literate creatives)

4. AI Augmentation & Generative Systems

(Acceleration, not replacement)

Purpose

AI modifies speed, access, and scale, not the core VP logic.

Legitimate roles in VP

  • Environment ideation

  • Rapid scene prototyping

  • Asset variation

  • Look-dev acceleration

  • Pre-visualisation

  • Scene remixing

Critical clarification

AI tools are not VP engines.
They sit above or alongside the engine layer.

This distinction protects your thesis from “AI = VP” collapse.

5. Output, Distribution & Post

(VP as pipeline, not endpoint)

https://massive.io/wp-content/uploads/2023/02/Video-Post-Production-Workflow-Diagram.png
 
 

Outputs

  • In-camera VFX footage

  • Broadcast content

  • Feature film plates

  • XR / immersive media

  • Game-engine native experiences

Academic angle

VP blurs production/post-production boundaries,
creating a continuous production loop rather than linear stages.