RISE-Editing uses neural 3D point clouds, with associated neural features, to model and editing radiance field. Specifically, we first propose a novel rotation-invariant neural point field ...
Abstract: With the advent of generative models and vision-language pre-training, significant improvement has been made in text-driven face manipulation. The text embedding can be used as target ...
ScenePainter: We propose ScenePainter, which aims to generate semantically consistent yet visually diverse 3D view sequences starting from a single view. Pipeline of ScenePainter. We propose a ...
When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback