A Fast and Realistic Bloom Rendering Method for Large Scale 3D Scene

As a ubiquitous phenomenon in nature, bloom plays an important role in many 3D realistic rendering fields, such as CG games, films and scientific visualization. With the employment of bloom, the quality of synthetic image would be improved and the representation of abstract engineering data could be refined further. However, rendering bloom is always a concern due to its heavy computational overhead. Considering the irregular geometric shapes and spatial occlusion, the issue becomes more complicated. To facilitate bloom rendering in the practical scenario, the paper described a convolution-based image-space rendering algorithm, which exploited the ability of rendering to texture(RTT) and frame buffer object, to generate, transform and process the bloom textures in the multi-pass rendering style. The introduced method offered a high-quality and effective way both for nature objects rendering and abstract data-set visualization. The paper also implemented the introduced approach in aerospace application and demonstrated its feasibility in large scale 3D scene.