Anatomically accurate, fully labelled synthetic palm-print and palm-vein imagery — generated from a body atlas, with the vascular network modelled as Bézier curves rendered 0.4 mm beneath the skin.
MESH · VASCULAR SPLINES · —0.4mm
Hand, palm-print and vein datasets are small, demographically narrow, and wrapped in consent, PII and biometric-privacy law. Labelling vessels by hand is slow and inconsistent. Models trained on what little exists don't generalise across sensors, skin tones or capture conditions.
Every sample ships with perfect ground truth and full control over the identity and the capture — at any scale.
Each hand carries its own palm-line signature, vein topology, geometry, skin tone, age and even scars or tattoos — no two alike.
Arteries and veins are built as Bézier + spline networks and rendered as true subcutaneous structure, not painted-on texture.
Re-capture the same identity under any lighting, pose, rotation, perspective, occlusion or sensor profile you need.
Vessel masks, hand geometry and identity IDs are generated with the image — zero manual annotation, zero label noise.
No real people, no consent forms, no PII. GDPR / BIPA / CCPA-safe to store, share and ship in a product.
High-resolution 1888×2248 output, delivered in standard formats and folder conventions your pipeline already expects.
The same synthetic subject, re-rendered across acquisition conditions — the exact within-identity variation biometric matching and anti-spoofing models need, and can never harvest at scale.
Every sample traces back to an anatomical model, so the vessels are structurally plausible and the labels are exact by construction.
A parametric hand is drawn from an online body atlas, giving realistic geometry, proportions and surface detail as the base.
Arteries and veins are laid out as Bézier curves and splines and embedded 0.4 mm below the skin surface — a unique network per identity.
The identity is rendered under configurable light, pose and sensor conditions, with vessel masks and metadata emitted alongside every frame.
Distinct identities and capture conditions — each image paired with exact vessel and geometry labels.
Train and benchmark authentication models with balanced identities and controlled within-subject variation.
Generate paired genuine and adversarial capture conditions for presentation- attack detection.
Pre-train vein-finder and venipuncture-guidance models on structurally accurate subcutaneous vasculature.
Backfill under-represented ages, skin tones and sensors to close coverage gaps and de-bias existing datasets.
Reproducible, shareable datasets with exact ground truth — no ethics-board bottleneck.
Simulate acquisition conditions before hardware exists to de-risk scanner and optics design.
Tell us the identities, conditions and labels you need. We'll send a free sample pack and a quote for the full generation run.