Urban Digital Twins: Colorful Smoke shows where the action is

Projekt: “Colorful Smoke” Visual Metaphor for Temporal Occupancy in Urban Digital Twins

Forschende: M.Sc. Olga Shkedova, M.Sc. Jeson Lonappan, Prof. Dr.-Ing. habil. Monika Sester

Projektidee: Develop a 3D city modeling system that fuses heterogeneous data sources into one voxel grid and visualizes moving objects as “Colorful Smoke” - showing busy, empty, or changing spaces without getting lost in complex visual clutter.

Cities are alive, breathing with people, cars, and deliveries that move, pause, and disappear throughout the day. Yet most digital city models remain frozen snapshots that cannot capture this constant dance of space usage. Real-time voxel-based city models offer a powerful alternative to traditional static representations by dividing the urban environment into millions of tiny three-dimensional cubes (voxels), each functioning as a smart pixel that can independently update itself whenever the physical world changes. Models like these make simulations in urban planning possible—what if I were to change something about the current situation? How would that affect, for example, traffic conditions or the urban climate?

The challenge

However, real-time voxel-based city models introduce a significant visualization challenge: when thousands of these dynamic voxels represent moving objects such as vehicles or crowds, they generate visual clutter with numerous small, flickering elements that obscure the very patterns users need to see. 

The idea: Colorful Smoke

This visualization challenge prompted us to propose the "Colorful Smoke" visual metaphor. The concept attempts to represent moving objects as soft, clouds with varying density, transparency, and color, to communicate busy, empty, or changing spaces more intuitively. The rationale behind this approach is that smoke, like fog or clouds, is inherently associated with temporality, leading us to hypothesize that viewers will intuitively interpret these colorful clouds as temporary phenomena, with density indicating occupancy likelihood and color distinguishing object types. We anticipate that this metaphor may reduce visual clutter and offer a more accessible visual language for understanding dynamic urban patterns.

From static models to an urban digital twin

To realize this vision, we first developed a data integration approach that fuses three heterogeneous sources, specifically a classified point cloud obtained by a mobile mapping system (MMS), CityGML building models, and a digital terrain model (DTM), into a unified 3D voxel grid, producing an urban digital twin of a street intersection in Hannover, Germany. The point cloud is classified into 12 semantic categories (buildings, roads, cars, persons, bicycles, etc.), providing both geometric detail and labels to distinguish static elements from dynamic objects. These three sources are integrated within a 10-centimeter voxel grid, preserving spatial accuracy and semantic meaning for every voxel, resulting in a rich data environment that enables real-time updates, semantic queries, and temporal analysis beyond the scope of traditional static 3D city models.

Virtual rays: How to turn moving objects into “Colorful Smoke”

To turn moving objects like people, cars, and bicycles into "Colorful Smoke," we first calculate how likely each 3D voxel is to be occupied, based on how many points (from the MMS) it contains. More points mean a higher chance of occupancy by some object in that voxel. We then smooth these values across neighboring cubes to create soft, gradual transitions between occupied and empty areas, which further enables a natural and continuous smoke-like appearance when rendered using ray marching. This technique shoots virtual rays through the 3D volume of the dynamic object and samples the data along each ray to build the final image, where high-probability areas appear as thick, opaque smoke and low-probability areas as thin, transparent clouds. The smoothed probability field is stored as a 3D texture, and as each ray marches through the volume, it samples this texture at regular intervals; trilinear interpolation is applied at each sample point to smoothly blend between neighboring voxel values, ensuring a fluid smoke appearance of dynamic objects without jagged artifacts. Meanwhile, the static urban scene, including buildings, roads, trees, etc., remains in the voxel representation, allowing users to easily distinguish permanent structures from temporary objects.

Next steps: more colors and user studies

This research introduces a voxel-based urban digital twin framework and the "Colorful Smoke" metaphor for temporal space occupancy visualization. Looking ahead, we will focus on enhancing the metaphor by introducing additional colors, along with a deeper analysis and adjustment of rendering parameters to improve representation and better differentiate high-probability areas from low-probability regions in cases where voxels occlude each other. We will also conduct user studies to evaluate how intuitively people perceive the metaphor, and test the urban digital twin with frequent dynamic updates to assess its performance in real-world changing urban scenes. We believe this is only the beginning of making urban data more accessible and engaging.