AF-NTC paper adds explicit anisotropic filtering to neural texture compression, beats trilinear-BC7 at 25% size
ssh4net · x · 2026-11-11
AMD researchers Shin Fujieda and Takahiro Harada will present AF-NTC at SIGGRAPH Asia 26, bringing explicit anisotropic filtering to neural texture compression.
- Feature-pyramid NTC gets no GPU texture-unit filtering and is trained at fixed UV/LOD, causing uniform blur near the camera
- The decoder is conditioned on the same three bounded values a hardware sampler uses: LOD (minor-axis scale), aspect ratio, and major-axis orientation, driving a tap line and level blend, trained against anisotropically filtered ground truth
- At 25% of a full BC7 mip chain, AF-NTC beats no-AF trilinear-BC7 on all ten material-scene pairs (+1.3 to +7.6 dB) and sits 0.0-3.4 dB below the 4x-larger anisotropic-BC7
More from Research
- DNAForge & MTC Berkeley to host AI × Biology Hackathon in Berkeley on Nov 14 — julia_kiseleva · 2026-11-11
- RoboScript hits 83.4% success on RoboTwin 2.0 with one demo per task, beating trained policies by 20+ points — H-EmbodVis · 2026-11-11
- Alibaba's PhysOPSD uses executable code as privileged context to make video generation physically plausible — AlibabaTongyiLab · 2026-11-11
- mayfer: clean low-rank latents may be the key to continuous online learning — mayfer · 2026-11-11
- PhD student finds lost Mayan city with 6,000+ buildings in old LiDAR dataset — bilawalsidhu · 2026-11-11
- Scaling intelligence: pattern-discovery problems all benefit from test-time scaling — QuanquanGu · 2026-11-11