Multi-view 2D image learning maps recognized machining features back to 3D part models, improving accuracy for complex and new features.
Lightweight object attributes and 2D floorplans replace dense 3D imagery, cutting bandwidth while preserving realistic shared environments.
Locked image slices, edge matching, and silhouette feedback improve AR alignment of 3D anatomy data without obscuring the patient.
Projected 2D plan views within a 3D building model improve facility inspection and verification without relying on separate displays.
Only objects affected by a changed reflection capture are re-rendered, cutting rendering overhead, power use, and frame freezing.
Multiple image labels are reconciled through shared 3D model regions to improve automated classification accuracy and model segmentation.
Orientation-based distance comparison resolves equal-hit ties in ray tracing, reducing shared-edge holes and color inconsistency.
Dual mesh mapping uses vertex-face and vertex-vertex links to cut soft-body rendering distortion while preserving high graphic mesh precision.
Preserving G-buffer and optical flow in UV space gives frame generation access to hidden-surface data, reducing disocclusion pixel errors.
Compressed oral CT image data enables faster multi-planar reformation while reducing storage and transmission load for 3D model viewing.
Hierarchical acceleration skips irrelevant ray tests, while stored-intersection management helps prevent memory overflow during rendering.
When object removal leaves hollows in 3D models, adjacent structures are extended and compared to restore a coherent spatial model.
Importance-weighted accumulations redistribute samples from clamped pixels while maintaining the desired average count for efficient ray tracing.
Progressive ray-traced rendering lets users capture a 3-D image at a chosen quality level before full refinement finishes.