top of page

Synthetic training data for Vision AI, generated on your own hardware.

Chameleon is Synthera's synthetic data platform. Build scenes, generate annotated images and video at scale, and ship to production without sending a single byte off-site.

PLATFORM CAPABILITIES

Everything you need to take a dataset from idea to production-ready.

Drag-and-drop scene builder

Compose scenes with assets, lighting, and actors. No ML engineer required, domain experts ship from day one.

Hybrid 3D + GenAI rendering

3D fidelity where it matters; generative variation where it doesn't. Zero hallucinations on geometry or labels.

Digital humans built in

Diverse, posable, animatable people, the heart of the human-activity stack other generalists don't have.

Scripting API & automations

Drive the simulator. Sweep parameters, batch-render scenarios, and integrate into your CI.

Pixel-perfect annotations

Bounding boxes, segmentation, keypoints, depth, optical flow, exported in custom formats.

Self-hosted by default

Runs on your hardware. Air-gapped, IP-safe, and ready for the accounts SaaS competitors can't reach.

TECH SPECS

The technical bits.

Designed to drop into existing vision-AI training pipelines.

ANNOTATIONS

2D

Bounding boxes, oriented boxes

3D

3D boxes, depth, surface normals

SEGMENTATION

Semantic, instance

HUMAN

Keypoints, skeletal pose, gaze, gesture

EXPORT FORMATS

STANDARD

PNG, JSON, CSV

CUSTOM

HNF, HMF, HDF

PLATFORM

OS

Windows, Linux (Ubuntu 22.04+)

GPU

8 GB+

CPU

2 GHZ

RAM

16GB

Why not give Chameleon a try?

Chameleon™ lets your team build synthetic training datasets using your own domain knowledge. TRy our cloud version below.

bottom of page