Real-Time Dynamic Bokeh Rendering With Efficient Look-Up Table Sampling.

Area of Science:

• Computer Graphics
• Image Processing
• Real-time Rendering

Background:

• Defocus blur rendering is crucial for photorealism in computer graphics.
• Existing methods often struggle with aliasing and computational cost for complex bokeh effects.

Purpose of the Study:

• To develop a real-time bokeh rendering technique that enhances photorealism.
• To address undersampling artifacts and computational demands in current rendering approaches.

Main Methods:

• A novel technique that splats pre-computed sprites, incorporating dynamic visibility and intrinsic appearance at runtime.
• Combines dense visibility sampling of highlights via rasterization with sparse sampling for regular objects.
• Dynamically transforms intrinsic appearance using a precomputed 2D texture lookup table encoding radial aberrations.

Main Results:

• Achieves complex bokeh effects in real time without undersampling artifacts.
• Significantly improves the photorealism of defocus-blur rendering.
• Efficiently handles dynamic visibilities and intrinsic appearances.

Conclusions:

• The proposed method offers a computationally efficient and effective solution for real-time photorealistic bokeh rendering.
• It advances the state-of-the-art in computer graphics by enabling complex, artifact-free defocus effects.