A speed-threshold trigger keeps the 3D vehicle view stable at low speed, reducing obstacle-driven angle switching, dizziness, and flicker.
Uses 3D scene reconstruction and multiple shadow maps to edit 2D image shadows with less manual work and better real-world consistency.
Collapse multiple independent mesh edges per iteration by restricting neighboring edges, avoiding partition interface issues while preserving simplification history.
A GPU hit map identifies which scene objects need shadow tests in cascaded shadow maps, cutting rendering waste and CPU-GPU traffic.
By recovering per-pixel UV corrections from the Z buffer, this XR reprojection approach cuts latency and memory bandwidth on limited GPUs.
Close-up inspection images are aligned to 3D models with metadata, preserving location context for accurate structure review and annotation.
By filtering scene graph nodes by viewable-region relevance, hit testing stays accurate while cutting processing time and compute load.
Importance sampling traces rays toward high-contribution scene regions, improving bounce light map accuracy with lower real-time rendering cost.
Uses 3D scene representations and multiple shadow maps to edit 2D image shadows more intuitively while preserving real-world scene consistency.
Multiple 3D X-ray views are aligned to the source coordinate system to expand field of view and remove distortion in target presentation.
Parallel raster processing computes shadow edges and distance data at adaptive resolution to cut transmission batches and speed shadow texture generation.
Aggregated visibility tests identify unoccluded environment-map pixels and guide rays toward visible light sources.
Separate view-independent and view-dependent radiance to render diffuse and glossy surfaces under new lighting and viewpoints.
A 3D scanner and tracked input device convert physical tracing into reversible digital annotations on an augmented model.
This case replaces complex 3D modeling with depth and normal maps to improve occlusion accuracy, illumination, and rendering efficiency.
A bounding mesh and curvature-trained neural network accelerate BTF rendering while preserving realism and reducing edge artifacts.