1).
Feed-forward framework unifies 4D reconstruction and generation from video latents
SplatFlux decodes observed or generated video latents into explicit static-dynamic 4D Gaussians without per-scene optimization.
Chinese Tech
Runmin Zhang · Binglin Ren · Zhu Yu · Fudong Wang · Jiangke Lin · Yichun Shentu · +3 more
Zhejiang University · Ant Group
Research Digest··3 min read
SplatFlux is a feed-forward model that takes video latent sequences -- either from a VAE-encoded real video or from a video diffusion model -- and outputs a 4D Gaussian splat representation of the dynamic scene.
§