AI
GeoVerse: World-Consistent Novel View Synthesis in Geometric Latent Space
Publication: Hugging Face Daily Papers
Author: Not available in collected metadata
Source published: 27/09/2026, 20:00:00 UTC
Briefing created: 29/09/2026, 14:51:26 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 29/09/2026, 07:18:25 UTC.
Briefings
GeoVerse is a framework for world-consistent novel view synthesis that combines pretrained 3D foundation models with video generative models. It operates within the geometric latent space of Wan2.2 VACE, using a ControlNet-style adapter to inject appearance priors from video data. A global spatial memory enforces cross-view coherence by reprojecting guidance to a shared scene representation. Experiments show improved visual quality and geometric consistency, achieving a 2.23 dB higher PSNR on DL3DV and 32.4% lower ATE on Mip-NeRF360 compared to GLD.
Key points
- GeoVerse synthesizes world-consistent novel views by performing generation within the geometric latent space of a pretrained 3D foundation model.
- The framework injects appearance priors from a video generative model into the geometric latent diffusion model via a ControlNet-style adapter.
Context and limitations
Results are limited to specific datasets including DL3DV and Mip-NeRF360, with no generalizability to other domains or unseen regions.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 29/09/2026, 14:51:26 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeich7q7yb3oybhexehqu37lefiffiwnqa3o57i22r7ohbdpz2y3vyq