Compressed facade displacement maps.

Area of Science:

• Computer Graphics
• Geometric Modeling
• Real-time Rendering

Background:

• Rendering massive urban models is computationally intensive.
• Memory transfer bottlenecks limit the performance of existing methods.
• Detail generation often requires large amounts of data.

Purpose of the Study:

• To develop a memory-efficient and fast rendering approach for massive urban models.
• To enable direct rendering from a compressed representation, bypassing memory transfer limitations.
• To improve the speed and scalability of urban visualization.

Main Methods:

• Rendering crude building outlines as polygons.
• Generating surface details via ray-tracing displacement maps in the fragment shader.
• Developing a compression algorithm for displacement maps enabling selective, fast access.

Main Results:

• Achieved an 85x compression factor for massive urban models.
• Outperformed basic geometry-based renderers by 40x to 80x in rendering speed.
• Demonstrated a prototype implementation of the proposed rendering technique.

Conclusions:

• The proposed method effectively addresses memory bottlenecks in urban model rendering.
• Direct rendering from compressed displacement maps offers significant performance improvements.
• This approach enables faster and more efficient visualization of large-scale urban environments.