Search papers, labs, and topics across Lattice.
This paper introduces {\alpha}Depth, a novel layered representation that addresses the challenges of accurately modeling soft boundaries in stereo conversion, such as hair and defocus blur. By employing a Circular Alpha Representation (CAR), the method shifts from global target extraction to local boundary decomposition, enabling efficient scene-level inference without the need for user intervention. Extensive evaluations show that {\alpha}Depth outperforms existing techniques, significantly reducing background bleeding and structural distortions at soft boundaries.
Achieving state-of-the-art stereo conversion, {\alpha}Depth eliminates background bleeding and structural distortions at soft boundaries, redefining how we approach complex scenes.
Accurately modeling soft boundaries, e.g., hair and defocus blur, is a fundamental challenge in stereo conversion due to the ambiguous blending of foreground and background. Existing depth models primarily predict single-layer depth, leading to ambiguity in depth correspondence at soft boundaries. While matting techniques can capture opacity for layered modeling, they often struggle in complex scenes with multiple targets and usually require user intervention. This paper introduces {\alpha}Depth, a layered representation that decomposes soft boundaries for high-fidelity stereo conversion. Specifically, we first resolve mixed color and depth ambiguity by estimating layered color and depth values at soft boundaries. Considering complex multi-target scenes, we design a Circular Alpha Representation (CAR) that shifts the paradigm from global target extraction to local boundary decomposition. Unlike prior matting methods restricted to a single foreground/background, CAR enables efficient scene-level inference without manual guidance. Extensive evaluations demonstrate that {\alpha}Depth achieves state-of-the-art performance in stereo conversion, eliminating background bleeding and structural distortions at soft boundaries.
