D3GS: depth, DINO and diffusion co-guided 3D Gaussian Splatting for sparse-view reconstruction
zhenjun_zhao · x · 2026-09-22
D3GS addresses sparse-view 3D Gaussian Splatting's ambiguous geometry, cross-view inconsistency, and missing details with a three-way co-guided framework:
- Diffusion-based completion plus DPT refinement recovers high-resolution metric depth for robust Gaussian initialization
- DINO-guided view-consistent learning augments Gaussian attributes with structural features
- A diffusion-based Gaussian refinement module injects generative priors via iterative optimization
Experiments on DTU, LLFF, and Mip-NeRF 360 show consistent, substantial gains over strong baselines, with ablations confirming each component's role.
More from Research
- Agora paper: 13 LLM research agents self-organize via Git for 12 days — suchenzang · 2026-09-22
- Rich RL Report Ships With 9B Distilled Model and 7,000 Open RL Environments — tokenbender · 2026-09-22
- New arXiv Paper: Router-Aware Importance Sampling Stabilizes MoE RL Training — tokenbender · 2026-09-22
- Thermofluids professor: OpenAI's Clay problem solution is not physically reproducible — GaryMarcus · 2026-09-22
- Two-stage ESM-2 screening plus structural validation uncovers divergent RNA viruses — bravo_abad · 2026-09-22
- GAE generates video and 3D geometry natively in a shared geometry latent space, code released — yshan2u · 2026-09-22