RealMe Turns Sparse 720p Multi-View Captures into 2K Volumetric Video with Video Diffusion
zju3dv · hf · 2026-10-22
Zhejiang University's CAD&CG lab and 4DV.ai present RealMe, which enhances sparse, low-resolution multi-view captures into high-resolution, 4D-consistent volumetric video. It uses a video diffusion model with layout-aware attention and frequency-domain latent supervision, upscaling 720p sparse views to 2K volumetric output.
More from Multimodal
- "X-ray vision" AI image trend: game characters with visible skeletons, prompt included — creatoroff · 2026-10-22
- Midjourney haute couture alien prompt shared with full sref code stack — creatoroff · 2026-10-22
- More Midjourney thinking mode tests: cinematic monster-portrait compositions — creatoroff · 2026-10-22
- Solarized Oceanic Enigma: a deep-sea neon prompt template — LudovicCreator · 2026-10-22
- Higgsfield launches Content Factory to turn any idea into viral reels — CurieuxExplorer · 2026-10-22
- Midjourney creator shares tribal-portrait sref style code with full v6.1 parameters — michaelrabone · 2026-10-22