DATASET BREAKDOWN
FOUR CORE DATASETS
RAW CR2 Images
Growing daily
01
High-resolution RAW CR2 images captured with professional-grade equipment, providing uncompressed source material for maximum flexibility in post-processing and analysis.
RAW 3D Scans from 6700 individuals
Growing daily
02
Complete RAW 3D scans of unique individuals, each captured with 11 distinct expressions, offering comprehensive facial performance data for realistic character creation.
Retopologised Neutral Faces
Growing daily
03
Production-ready neutral face meshes with optimized topology, perfect for animation and real-time rendering in games and interactive applications.
Complete Retopologised Faces
Growing daily
04
Fully retopologised faces including eyes, teeth, and hair systems, ready for immediate integration into production pipelines with full anatomical detail.
UNIFIED HUMAN GEOMETRY
TOPOLOGY CONSISTENCY
ONE CONSISTENT TOPOLOGY. DATASET 3 / 4
Unified Topology 01
By maintaining identical topology, UV layouts, and vertex correspondence across all heads, the dataset enables seamless morphing, rig reuse, and expression transfer for VFX and real-time character pipelines, while simultaneously providing clean, noise-free geometry for AI and machine-learning workflows. This consistency allows models to learn true facial variation rather than mesh artefacts, supports large-scale synthetic identity generation, and ensures that data from multiple datasets can be combined with ease.
SYNTHETIC MODELS
The potential number of unique, identity-safe synthetic human heads generated from our unified topology and modular facial components.
FACIAL EXPRESSIONS
EXPRESSION CAPTURE
Each individual is captured across a fixed set of 11 core facial expressions, designed to span neutral states, functional movements, and primary emotional extremes. DATASET 2
Expression Set 01
The expression set includes neutral, eye and mouth articulation, brow movement, and key emotional states such as happiness, sadness, anger, surprise, fear, and disgust.
All expressions are processed and exported with 16384 x 16384 px textures and low and high polygon models 200k and 7 million polygons respectively.
CONSISTENT UVS
TEXTURES
All textures are captured and processed to align with a shared mesh topology and UV layout, allowing textures to transfer seamlessly across identities. DATASET 3 / 4
Unified UV Space 01
All textures are authored in a single, consistent UV layout, allowing them to transfer cleanly between heads while preserving fine skin detail and anatomical alignment
Interchangeable Identity 02
The same texture can be applied across multiple geometries — and multiple textures across the same geometry — enabling rapid variation without re-baking or reprocessing.
Synthetic Appearance Generation 03
By blending and recombining textures from multiple individuals, we generate new, identity-safe skin appearances that remain fully compatible with the underlying geometry.
SYNTHETIC DATA GENERATION
PHOTOREALISTIC SYNTHETIC IMAGERY AT SCALE
Using our unified 3D human dataset, we generate billions of images across controlled viewpoints, expressions, and lighting conditions purpose-built for AI training and evaluation.
FACIAL EXPRESSIONS
EXPRESSION TRANSFER
Using a delta-based transfer technique, facial expressions can be applied consistently across every head in the dataset. DATASET 3 / 4
FACS 01
Because all models share identical topology and vertex correspondence, any expression can be transferred to any identity with minimal manual adjustment.
This enables scalable, FACS-compatible facial variation for animation, simulation, and machine learning workflows without additional capture.
Photogrammetry
Each subject is captured simultaneously with over 70 cameras, providing full geometric coverage, high facial detail, and consistent alignment across all scans. This setup enables high-volume capture with no compromise in data quality.
The lighting environment is tightly controlled and synchronised, eliminating motion artefacts and lighting variation. All cameras fire simultaneously under identical, flat illumination, producing clean, shadow-free imagery optimised for reconstruction and machine-learning workflows. This consistency is essential for reliable geometry, texture extraction, and synthetic image generation.
All capture took place in a purpose-built public studio designed for ethical, paid scanning at scale. The studio operated for over a year, enabling steady data collection with clear consent and reliable processes. This setup allowed the dataset to grow over time without compromising quality.
CAPTURE AT SCALE
We have been at the forefront of photogrammetry capture for over 15 years
Rig 01
Lighting 02
Studio 03
ART-LED CAPTURE
BUILT BY CHARACTER ARTISTS
With over 18 years of experience creating digital characters for AAA games, animation, and film, Ten24 has worked with studios of all sizes, from global technology companies to independent developers. We also operate 3D Scan Store, supplying production-ready digital assets since 2012, with over one million models distributed worldwide. This long-term experience informs how we capture, process, and deliver scan data.
PRODUCTION SOLUTIONS
OUR SERVICES
Casting 01
Cast from our ever growing library of over 6500 individual head scans.
Custom Topology and Metahumans 02
We can create custom character or metahumans to your specification using our library as a basis for custom characters
Synthetic Humans 03
Our team of artists can create unique, synthetic individuals by blending multiple scans to your specification
NOT SCRAPED
ETHICAL DATA CAPTURE
All data is captured directly from informed, paid participants, never scraped or third-party sourced. The dataset is anonymised, GDPR-compliant, and licensed for long-term commercial use.
Paid, Opt in, Participation
Every individual in the dataset has participated voluntarily and has been fairly compensated for their time. All participants are fully informed about how their data will be captured, processed, and used, with explicit consent provided prior to scanning.
Directly Captured, Never Scraped
All data in the dataset is captured directly by our team under controlled studio conditions. Every scan comes from a real participant, captured specifically for this dataset with informed consent. No scraped or third-party data is used.
GDPR Compliant by Design
The dataset is developed and managed in compliance with GDPR requirements. Ten24 acts as the data controller, with governance processes covering consent, purpose, data minimisation, and rights. Personal identities are anonymised, and no identifying information is shared.
Responsible Commercial Use
All data is licensed for responsible commercial use, including games, film, advertising, and AI research. Explicit restrictions prohibit use in sexually explicit content, surveillance, biometric identification, or other high-risk or unethical applications.
CLIENTS
TRUSTED BY INDUSTRY
We have worked with organisations of all sizes, from global enterprises to specialist studios. These long-standing relationships are built on consistency, discretion, and the ability to deliver scan data that holds up in real production environments.