ECCV 2026: DF3DV-1K dataset and DI²FIX benchmark for distractor-free novel view synthesis
rsasaki0109 · x · 2026-09-15
Researchers from UTS, University of Sydney and NYCU released DF3DV-1K, a large-scale dataset and benchmark for distractor-free novel view synthesis, accepted to ECCV 2026, along with the DI²FIX codebase and checkpoints. Newly added DF3DV-Extra scenes double the dataset size with 1,000+ additional scenes. Dataset, benchmark and leaderboard are open-sourced on GitHub.
More from Multimodal
- AI-Generated Dark Fantasy Musical Film 'Her Dark Emergence' Showcases Graphic Visuals — Th3dzon33 · 2026-09-15
- AI Toolkit adds LoRA training support for music model YuE2-3B — ostrisai · 2026-09-15
- HeyGen's open-source HyperFrames hits 50k GitHub stars: agents edit video with HTML — HeyGen · 2026-09-15
- GPT-6-powered Spawn builds a custom Blender cheetah with procedural fur — majidmanzarpour · 2026-09-15
- All-AI thriller short "All Aboard": 9 generated blocks, one color per character, prompts public — tsi_org · 2026-09-15
- Intangible's Motion Paths turns a drawn line into 3D camera choreography — jnack · 2026-09-15