Image Generation TRIZ Cases
- Dynamic Splat Streaming for Viewport-Based Bandwidth Control
This case streams high-fidelity splats to important view regions and lower-fidelity data elsewhere to reduce bandwidth and compute.
- Visual Reordering of Partial Vector Objects Without Geometry Damage
Automatically reorder partial vector objects while preserving original geometry.
- Cache-Based 3D Scene Acceleration Structures for Dynamic Ray Tracing
Hierarchical volumetric caches reduce unnecessary intersection tests while supporting real-time acceleration-structure updates.
- Dynamic Guidance for Complete, Low-Artifact 3D Object Scanning
Image segmentation and audio, visual, and haptic cues guide camera distance, orientation, and speed for more complete 3D scans.
- Dot-Based Image Generation for Precise Object Placement
Abstract prompts and rigid camera perspectives hinder image editing; structured dots enable precise object placement and modification.
- Primitive Tiling with Boundary Testing for Lower-Power Graphics Processing
This graphics case uses primitive boundary intersections to limit precise tile tests, reducing floating-point work, time, and power.
- 3D Model Tree Generation for Visibility-Based Low-Delay Streaming
Quality and visibility values select 3D model tree nodes, reducing transfer size and loading delays for XR rendering.
- Voxel Generation with Hash Tables for Point Cloud Efficiency
This case maps point locations to voxel indications in a hash table, reducing memory checks while preserving accurate representations.
- Ray-Tracing Child-Node Ordering for Faster BVH Traversal
This graphics processor stores ordering strategies per parent node to reduce ray-intersection testing time during ray tracing.
- Learned Mesh Ray Tracing for Real-Time 3D Rendering
This case combines learned meshes, BVH traversal, and shader processing to support real-time 3D rendering with view-dependent colors.
- Optimal 3D Model Trees for Quality-Aware XR Streaming
This case builds an optimal 3D model tree that streams visible, high-detail regions first, reducing data transfer and waiting time.
- Photometric Tangent Map Creation with Low-Frequency Normal Correction
This case separates mesh and photometric normal detail, correcting low-frequency variation before tangent map generation.
- Volumetric Image Rendering with Trailing Light for Smooth Interaction
Trailing light contributions reduce re-rendering burden while smoothing volumetric images.
- Bit-Encoded Hierarchical Spatial Trees for Dynamic 3D Object Queries
This case uses interleaved coordinate codes and node arrays to reduce memory and computational load in dynamic 3D environments.
- 3D Object Model Storage in ISO BMFF with Adaptive LOD Retrieval
ISO BMFF stores multiple 3D model LODs so XR devices can match detail to network bandwidth and processing capacity.
- Infinitely Scalable Point Clouds from 3D Mesh Models Without Zoom Distortion
This case maps enhancement values onto densely generated polygon points, preserving detail and reducing zoom distortion in point clouds.
- NeRF Graphics Rendering with a GPU-NPU Pipeline for API Compatibility
This case pairs GPU rendering and NeRF training with NPU inference to preserve graphics API compatibility and accelerate new-view images.
- Mixed-Reality Scenario Reconstruction via Multi-Device Point Cloud Alignment
Point cloud alignment and coordinate scaling reduce calibration errors when multiple MR devices reconstruct one scenario.
- Mixed Reality Occlusion Filtering for Visible-Region Video Quality
This case masks virtual-object occlusions before exposure, white-balance, and focus analysis, improving visible-region video quality.