WorldMesh: Text-to-Navigable Multi-Room 3D Scenes Accepted by ECCV2026
angelaqdai · x · 2026-07-05
The WorldMesh paper has been accepted by ECCV 2026, with its code open-sourced on the same day. Led by @mschneider456, the method generates navigable multi-room 3D scenes from text prompts by using a mesh skeleton to condition the image diffusion model, balancing global consistency with photo-level details. It represents a text-to-3D research paper and open-source project.
More from Multimodal
- Dense instance segmentation via Muse Wonder + SAM 3.1 on Meta's Model API — nikhilaravi · 2026-10-26
- LichtFeld Studio v0.5.4.3 ships community-built SteadySplats, POPSpa and Voxelization — janusch_patas · 2026-10-26
- One prompt to a 2-3 min narrated short: hands-on test of Antigravity video generation — AI_Andrew · 2026-10-26
- Creator makes 10,564-frame animated short entirely with Gemini toolchain — AI_Andrew · 2026-10-26
- FLUX.3 Klein step-distilled to 4-8 steps with CFG baked in — Fit-Structure-6489 · 2026-10-26
- ostris pre-training "Primal VAE": 1-channel latents match modern 8x VAEs via patching — ostrisai · 2026-10-26