3D Modelling

This is the  journey of the 3D Modelling System, a visionary project that sought to blend the physical and digital worlds into one. Through the lenses of multiple cameras, humans and their environments are reborn into virtual 3D representations, setting new standards in digital innovation and immersive experiences.

The Client: At the forefront of digital modelling, tasked with the ambitious goal of capturing real-world movements and surroundings to create lifelike virtual avatars and settings.

The Challenge: The complexity of accurately rendering 3D human models from live camera feeds, ensuring natural movement and interaction within both outdoor and indoor virtual environments, required a novel approach beyond traditional modelling techniques.

The development team embarked on creating a Python-based desktop application, distinguished by its seamless user experience and expandable framework. This solution was designed to not only capture and process live imagery but also to seamlessly integrate with future technological advancements, such as LIDAR and laser camera data.

• 3D Movement: Crafting lifelike 3D human movements required overcoming the inherent limitations of digital modeling, leading to the development of proprietary solutions for smoother animations.

• Scarcity: The scarcity of open-source 3D models compatible with Python necessitated complicated workarounds, including the creation of custom converters.

• Balancing: Balancing the high demands of detailed 3D graphics with the computational intensity of real-time deep learning posed significant technical challenges.

• Indoor Modelling: Leveraging depth-sensing camera technology, detailed architectural models were constructed. Digital humans, generated through sophisticated deep learning techniques, were then navigated within these structures.

• Real-Time Processing: Powered by the robust capabilities of an NVIDIA Tesla GPU, the system managed simultaneous video streams, maintaining flawless real-time processing and visualization.

• Outdoor Modelling: By applying advanced computer vision algorithms, the team was able to isolate and map human figures from live feeds, subsequently integrating these models into a dynamic 3D world.