Sim-to-real image generation engine for physical AI model training

Physics-accurate synthetic data generation and end-to-end optical pipeline validation for Physical AI systems

Summary :

3DOptix offers PhotonCore, a photonics simulation platform that generates physics-accurate, multi-modal synthetic datasets for training and validating AI perception systems, and links illumination, scene, optics, and sensor/ISP modeling in a single end-to-end workflow. We are seeking European research institutes, universities, and industrial partners to co-develop project proposals applying this technology to sensor/AI validation challenges in sectors such as automotive, robotics, space, defense, or industrial inspection.

Description of cooperation:

3DOptix's PhotonCore extends traditional optical design workflows into large-scale, physics-based data generation for Physical AI. It enables statistical performance analysis, edge-case discovery, and physically-based object augmentation for training and validating AI models, while also providing true end-to-end validation of full imaging/sensing pipelinesת from illumination design through scene interaction, optical propagation, and sensor/ISP evaluationת within one workflow. A natural-language interaction layer supports fast, iterative prototyping, lowering the technical barrier for exploration and design.

We are looking to join or form a consortium under Horizon Europe or a national/EU funding mechanism, to apply this capability to a concrete sensing or AI-validation challenge where synthetic data generation and full-pipeline optical validation can reduce cost, risk, or time-to-certification.

Advantages and innovative aspects:

Physics-accurate simulation (not purely statistical/generative) grounds synthetic data in real optical behavior, improving sim-to-real transfer.

Single connected workflow spanning illumination, scene, optics, and sensor/ISP removes the need to stitch together multiple disconnected tools.

Natural-language interface accelerates iteration cycles for non-specialist users, widening the pool of potential consortium contributors (SMEs, application-domain experts without deep optics expertise).

Applicable across multiple verticals (automotive, robotics, space, defense, industrial inspection, medical imaging), making the technology a flexible core for varied consortium themes.

Stage of development: Already on the market / available for adaptation to specific use cases.

Type of partner sought:

Research institutes or universities with expertise in AI perception, sensor systems, or robotics

Industrial partners (OEMs, Tier-1s, sensor/camera manufacturers) with a concrete validation or dataset-generation need

National metrology or defense research organizations for domain-specific edge-case data

Consortium coordinators experienced in Horizon Europe or Chips JU proposal submissions

Role of the partner sought:

Partners would contribute domain-specific validation challenges, real-world data/scenarios for benchmarking, application expertise, and (where relevant) consortium coordination experience, while 3DOptix contributes the PhotonCore platform and photonics/optical simulation expertise.

Keywords: photonics, physical AI, synthetic data generation, sensor validation, optical simulation, edge-case discovery, end-to-end pipeline validation, digital twin, autonomous systems