A flexible and accurate synthetic data generation platform
The Anyverse Platform
Hyperspectral (pixel-accurate)
render engine
Physics-based sensor simulation
Procedural (API-based) scene generation
Graphical interface for dataset development
Built-in assets library
Scalable cloud architecture
A flexible architecture
Anyverse provides a modular platform that enables efficient scene definition and dataset production. Anyverse™ Studio is a standalone graphical interface application that manages all Anyverse functions, including scenario definition, variability settings, asset behaviors, dataset settings, and inspection. Data is stored in the cloud, and the Anyverse cloud engine is responsible for final scene generation, simulation, and rendering. It produces datasets using a distributed rendering scheme.
Anyverse Studio
Anyverse Studio is the user interface for scene generation that allows users to recreate a diverse synthetic reality in static and dynamic conditions.Whether you need static annotated images or sequences, use Anyverse’s extensive asset library to compose a 3D scene. Apply dynamic behaviors and program the environmental variability you need with Python scripts.
Produce your datasets in the cloud and explore the results in Anyverse Studio, including all the associated ground truth data.
camera SPECS
GUI Scene design
sensor & ISP definition
2D & 3D viewports
3D assets
dynamic behaviorS
weather conditions
LIGHT
CONDITIONS
python
API
endless variability
script based dataset generation
Ground-Truth channels
Anyverse’s datasets include:
color images, raw hyperspectral data, annotations and pixel segmentation.
Dataset metadata:
Annotations, Color, Object Segmentation, Instance Segmentation, Material Segmentation, 3D Position, Radiance, Reflectance, Roughness, Motion Vectors, RAW data (before and after sensor), Depth, Surface normals, Light normals.
Built-in assets library
Anyverse brings a built-in assets library including all kinds of assets and objects ready to populate any scenario for a variety of use cases. All of them are classified (including materials) and categorized for easy and intuitive management.
Hyperspectral render engine
Anyverse implements a pure spectral path-tracing engine that computes the spectral radiance of every light beam, simulating lights, cameras and materials with physical accuracy. This allows for a precise simulation of the amount of light reaching the camera sensor to generate a final image containing full spectral information per pixel.
custom lens
model
hyperspectral
(256 bands sampling)
motion blur
Rolling shutter
complex environments (sky and WEATHER)
high bit-depth image output
raw sensor data
photometric accuracy
Anyverse simulates energy as an electromagnetic wave across the rendering pipeline to its final digital values.
Light sources, including the sky and sun, are modeled through their characteristic spectrum profile that depends on the type of light source and temperature (LED, incandescent, etc.). Materials are also physically modeled using BSDF functions.
Sensor simulation
Accurate modeling of sensors is essential to guarantee a high degree of generalization and reduce the domain gap.
Anyverse combines a physics description of lights and materials in a 3D scene with a detailed hyperspectral simulation of sensor intrinsics. The sensor pipeline calculates energy’s propagation per wavelength, considering many subtle physics effects such as the conversion of photons into voltage and RAW digital values before producing a final image.
A variety of sensors and lenses in visible and non-visible (near infrared) bands are supported.
Anyverse performs optical simulation using advanced path-tracing technology, capturing effects such as extreme optics distortion, lens shading, lens blurring (depth of field), complex assembly of multiple lenses, and more.
pixel size
exposure time
fill factor
well capacity
noise
qe curves
gamma
analog offset/gain
infrared
filter
low-pass
filter
color filter array
conversion gain
white
balance
rgb to xyz matrix
xyz to device matrix
pixel vignetting
FREE EBOOK
Anyverse camera sensor simulation
Learn everything you need to know about our physically-based sensor simulation.
LiDAR
Anyverse performs LiDAR simulation using its core ray tracing engine, which accurately tracks the interaction of beams emitted from the LiDAR sensor with different objects and materials in the scene. Mechanical components involved in the emission and reception of laser beams are approximated by different functions to match the scanning pattern of the physical LiDAR sensors.
Anyverse can combine camera and LiDAR simulation simultaneously. This means that for every color image a corresponding LiDAR point cloud can be produced. This enables sensor fusion algorithms to use combined LiDAR and camera sensor inputs.
Different parameters can be configured to simulate a specific LiDAR, including spinning, solid-state, and flash LiDAR.
- Range
- Points per second
- Horizontal & vertical field of view
- Angular resolution horizontal & vertical
- Rotation rate
- Accuracy
- Beam divergence
The point cloud data can be generated in different file formats. For every 3D point, rich scene information is supplied, such as position, distance, object ID, material ID, 3D bounding boxes, speed (in the case of dynamic objects), etc.
Start generating hyperspectral synthetic data
for every stage of your advanced perception system development cycle with Anyverse™.
A flexible and accurate synthetic data generation platform
Craft the data you need for your perception system in minutes. Design scenarios for your use case with endless variations. Generate your datasets in the cloud.
THE ANYVERSE PLATFORM
Anyverse™ is a scalable platform to design, train, validate, or fine-tune your perception system. It provides unparalleled computing power to generate all the data you need in a fraction of the time and cost compared with real-world data workflows.
Hyperspectral render engine
Physics-based sensor simulation
Procedural (API-based) scene generation
Built-in assets library
Graphical interface for dataset dev
Scalable cloud architecture
A FLEXIBLE ARCHITECTURE
Modular design that enables efficient scene definition and dataset production.
ANYVERSE STUDIO
Anyverse™ Studio is a standalone graphical interface application that manages all Anyverse functions, including scenario definition, variability settings, asset behaviors, dataset settings, and inspection.
Ground-Truth channels
Anyverse’s datasets include:
color images, raw hyperspectral data, annotations and pixel segmentation.
Dataset metadata:
Annotations, Color, Object Segmentation, Instance Segmentation, Material Segmentation, 3D Position, Radiance, Reflectance, Roughness, Motion Vectors, RAW data (before and after sensor), Depth, Surface normals, Light normals.
Built-in assets library
Anyverse brings a built-in assets library including all kinds of assets and objects ready to populate any scenario for a variety of use cases. All of them are classified (including materials) and categorized for easy and intuitive management.
HYPERSPECTRAL RENDER ENGINE
Anyverse™ hyperspectral render engine implements a pure spectral path-tracing engine that computes the spectral radiance of every light beam interacting with materials in the scene, simulating light transport at close physical level.
SENSOR SIMULATION
Using the hyperspectral data provided by the render, Anyverse can simulate all the physics happening at the sensor.
LiDAR
Anyverse simulates LiDAR accurately, including spinning, solid-state, and flash LiDAR, being able to combine camera and LiDAR simulation simultaneously. This enables sensor fusion algorithms to use combined LiDAR and camera sensor inputs.
Point cloud data can be generated in different file formats. For every 3D point, rich scene information is supplied, such as position, distance, object ID, material ID, 3D bounding boxes, speed (in the case of dynamic objects), etc.