How Tencent’s L4 3D Landmark Models Transform Modern Mapping

As demand for precise navigation and local‑service maps grows, maps have evolved from 2D to 3D, with L4 landmark building models playing a crucial role by providing intuitive, expressive, and exploratory experiences.

#01 L4 Landmark Definition and Application Scenarios

L1–L3 models are automatically generated at varying detail levels, while L4 models are manually designed and reconstructed landmarks that help users recognize their location, enhance immersion, and add fun.

L4 models are used not only in mobile maps but also in car‑navigation, autonomous driving, and custom map products, selecting iconic regional buildings and reproducing them with strict standards for realism and aesthetics.

#02 Tencent L4 Model Personality

Tencent L4 Building’s Small Personality

When creating L4 models, Tencent defines a clear style direction that balances realism with design expression, resulting in a distinctive visual identity.

1. Design abstraction based on reality

Each building is meticulously reconstructed: dimensions, proportions, spatial composition, and surface features are derived from architectural data, ensuring high design quality while faithfully preserving the real structure.

Wuhan Hankou Railway Station

Chongqing Raffles City / Wuhan Yellow Crane Tower

2. Good adaptability

The models are optimized for multiple platforms; thanks to efficient optimization and batch adaptation, they render well on mobile maps, car‑navigation (Tencent Smart Driving), WeChat mini‑programs, and other custom environments.

3. Wide coverage, many types

Tencent’s L4 models now cover over a hundred cities, ranging from large stations, airports, convention centers, and commercial complexes to small pagodas and signboards, encompassing diverse architectural styles while following detailed production standards to keep the collection orderly.

4. Light as a feather

To ensure smooth loading in extreme environments, Tencent developed an automated batch reduction pipeline that cuts polygon count by 30‑50% while preserving visual quality. Texture standards and high utilization allow rich detail even at very low resolutions and file sizes.

3D landmark model reduction 50%

UV layout for Level of Detail

#03 Model Capability Support

1. Architectural design thinking

Using architectural design principles, Tencent avoids superficial surface focus and instead reconstructs the true form of buildings, progressively breaking down structures from large to small to ensure accurate shape representation.

2. Model process design

Detailed standards govern every production step; after thousands of models, a strict yet reasonable system and rich techniques for modeling and texturing have been established, boosting texture utilization and allowing four‑fold size reduction while improving clarity.

3. Programmatic technical standards

Batch polygon reduction: adopting game‑industry wiring optimization and a self‑developed tool, reducing face count by 30‑50%.

Texture batch compression: fine‑grained compression across color space, channels, and resolution, preserving rendering quality while cutting sampling cost.

Texture batch adaptation: one‑click adjustment of texture specifications for car‑navigation, mobile, and custom map environments, enabling a single asset to serve multiple platforms.

4. Continuous pursuit of better experience

Design style evolves from simple monochrome gradients to richer textures; polygon counts are constantly pushed to new limits, and ongoing exploration of design methods, engine capabilities, and application scenarios ensures ever‑improving 3D map quality.

Replay of “Tencent 3D Landmark Model Design”.