SPAGHETTI

Neural implicit fields are quickly emerging as an attractive representation for learning based techniques. However, adopting them 3D shape modeling and editing is challenging. We introduce a method E diting I mplicit S hapes T hrough P art A ware G enera ion, permuted in short SPAGHETTI. Our architecture allows manipulation of shapes by means transforming, interpolating combining segments together, without requiring explicit part supervision. SPAGHETTI disentangles into extrinsic intrinsic geometric information. This characteristic enables generative framework with part-level control. The capabilities demonstrated using interactive graphical interface, where users can directly edit neural shapes. code, user interface demo pre-trained models available at github.com/amirhertz/spaghetti.