Digital Twins
Environmental Design: Achieving High-Quality Digital Twin Simulation, Efficiently at Scale
January 27, 2022
· Written by
Karan Harimohan


When it comes to designing high-quality, photorealistic digital twin environments, there are countless design elements to consider – and exacting time constraints to contend with. Too much time spent manually and meticulously simulating every square meter of an expansive environment could potentially result in significant project delays. Or worse, the cumbersome workflow is deemed insufficient to meet important client deadlines, and the project stalls before it ever starts.

Environmental design for digital twins requires an optimal balance of highly detailed data analysis and simulation, with workflow efficiencies that eliminate development complexities.

Our Approach

An effective digital twin for simulation requires both careful accuracy and fidelity to the environment and assets. At Duality, we have the means to automate environmental design workflows in a manner that shrinks overall development and delivery cycles from weeks to days - or even hours - further supporting a frictionless experience to bring any digital twin environment to the metaverse. 

By way of illustration, refer to the following simulated environment:                    

The image above depicts a neighborhood under development in a virtual locale modeled after Pittsburgh – a place that’s near and dear to Duality’s heart! Note the rolling hills depicted here, consistent with Pittsburgh's topography. Here’s where the location-specific reference data starts coming into play.

Step One: Foundation                     

Satellite data is collected and analyzed, yielding the hilly landscape as shown above. From this source data, a height map can be created, and the hill’s dimensions and slope can be approximated. The satellite data is therefore crucial for modeling the terrain.

It’s important to note, however, that there may be instances where high-resolution satellite data isn’t available. In these instances, and in cases where the virtual landscape design is an approximation but not exact duplicate of a real-world reference landscape, some manual landscape ‘sculpting’ may be needed to simulate the desired effect.

Step Two: Foliage and Architecture                    

As a next step, an analysis of local foliage data reveals the types of trees and vegetation native/common to Pittsburgh, which in turn impacts the look of the simulated environment’s natural green space (see above).  

Looking closely at the image above, you’ll note that the style and architecture of the houses simulated here are consistent with Pittsburgh-area design aesthetics for suburban housing. Here again, location-specific reference data was essential for establishing the roof, tile, and brick wall styles, among other factors, that in turn dictate the visual texturing needed to approximate real-world structures as they might look in a Pittsburgh neighborhood.

In all of the above illustrations/examples, real-world reference data played a huge role in informing the overall landscape and structure design. But some level of manual intervention was required in order to implement the aesthetic/texturing in the simulated environment, and this can get cumbersome at scale even with patterning techniques applied for reuse.            

When working with complex and vast terrains, environmental design must be streamlined at every opportunity. In order to maintain efficient development workflows, we must look for opportunities to automate key design processes.

The above image is a great example of how portions of the landscape design can be automated to save valuable time – and this capability is a key differentiator for Duality in keeping with our ‘zero friction’ design ethos. 

When simulating natural formations of rocks and fallen trees, advanced procedural tools can be leveraged in a manner that takes, let’s say, ten random rocks, trees, etc, and automatically ‘scatters’ them around the terrain.

But the end results are hardly random or haphazard.

The height and slope of the simulated terrain directly impacts the location of the item/asset situated upon it. The end location of a virtual rock that’s fallen down a virtual slope will be determined in part by the slope definition and the size/presumed weight of the rock, and this helps to explain why, for example, larger rocks might be found clustered at the base of a slope, with smaller pebbles situated on the slope itself. 

Here’s the important part: If we had attempted to handle this ‘scattering’ process manually across a multi-kilometer environment like the one shown above, it might have taken as long as 60 days. 

With the aforementioned procedural tools, this process can now be done in a matter of hours. We merely program in the height and slope dimensions of the terrain, and the assets (rocks, trees) can be scattered and rotated seemingly at random across huge swaths of the simulated environment – no two square meters of natural landscape will look exactly the same. This is all automated, and the time savings can be dramatic.

Step Three: Metaverse Applications   

How might this simulated environment be used for commercial applications? Imagine a realtor offering virtual neighborhood and/or house tours, or a city planner evaluating locations for commercial development and/or renewable energy plants, to name just a few use cases.

And yet additional factors must still be considered before a virtual digital twin environment is ready for use. Metaverse climate variations and weather conditions must also be accounted for, for example. Damp conditions will affect the way the ground surface appears as it gets wet. The grass may weigh more and therefore lay differently as it gets wetter, with additional implications for lighting and shading. 

At the end of the process, once we have a finalized environmental digital twin, utilizing Duality’s Falcon platform we can simulate some of the climate variations, time of day, etc - all things we couldn't do in the real world. 

By exploiting real-world, location-specific reference data pertaining to terrain, vegetation and architecture attributes, we can achieve greater fidelity and accuracy with our simulated digital twin environments. By leveraging the automated tools and expertise at our disposal, Duality Robotics can quickly and easily scale these design properties across the metaverse.

Karan Harimohan currently works as Senior Environmental Artist at Duality. His past projects within environment/prop art have spanned from modelling, texturing to lighting and level art. Visit his ArtStation profile here.

Learn more about our multidisciplinary team and explore career opportunities at Duality.