Stable Part Diffusion 4D: Multi-View RGB and Kinematic Parts Video Generation
bidirectional diffusion fusioncontinuous rgb-like imagescontrastive part consistency lossdual-branch diffusion modelharmonic skinning weightskinematic part videoskinematicparts20klatents vaemulti-view rgbpart segmentation mapsrigged objectsspatial alignmentspatial color encodingstable part diffusion 4dtemporal alignment
We present Stable Part Diffusion 4D (SP4D), a framework for generating paired RGB and kinematic part videos from monocular inputs. Unlike conventional part segmentation methods that rely on appearance-based semantic cues, SP4D learns to produce kinematic parts