SPAGHETTI: Editing Implicit Shapes Through Part Aware Generation

Hertz, Amir; Perel, Or; Giryes, Raja; Sorkine‐Hornung, Olga; Cohen–Or, Daniel

doi:10.48550/arxiv.2201.13168

Cited by 1 publication

(3 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Next, we compare the shape generation capability of our method with four state-of-the-art methods: IM-GAN [Chen and Zhang 2019], Voxel-GAN [Kleineberg et al 2020], Point-Diff [Luo and Hu 2021], and SPAGHETTI [Hertz et al 2022]. To our best knowledge, ours is the first work that generates implicit shape representations in frequency domain and considers coarse and detail coefficients to enhance the generation of structures and fine details.…”

Section: Comparison With Other Methodsmentioning

confidence: 99%

“…Some works attempt to use the reconstruction task to first learn a latent embedding [Chen and Zhang 2019;Ibing et al 2021;Mescheder et al 2019] then generate new shapes by decoding codes sampled from the learned latent space. Recently, [Hertz et al 2022] learn a latent space with a Gaussian-mixture-based autodecoder for shape generation and manipulation. Though these approaches ensure a simple training process, the generated shapes have limited diversity restricted by the pre-trained shape space.…”

Section: Related Workmentioning

confidence: 99%

“…Following the data split in [Chen and Zhang 2019], we use only the training split to supervise our network training. Also, similar to [Hertz et al 2022;Luo and Hu 2021], we train a single model for generating shapes of each category in the ShapeNet dataset [Chang et al 2015].…”

Section: Implementation Detailsmentioning

confidence: 99%

See 2 more Smart Citations

Neural Wavelet-domain Diffusion for 3D Shape Generation

Hui¹,

Li²,

Hu³

et al. 2022

Preprint

View full text Add to dashboard Cite

This paper presents a new approach for 3D shape generation, enabling direct generative modeling on a continuous implicit representation in wavelet domain. Specifically, we propose a compact wavelet representation with a pair of coarse and detail coefficient volumes to implicitly represent 3D shapes via truncated signed distance functions and multi-scale biorthogonal wavelets, and formulate a pair of neural networks: a generator based on the diffusion model to produce diverse shapes in the form of coarse coefficient volumes; and a detail predictor to further produce compatible detail coefficient volumes for enriching the generated shapes with fine structures and details. Both quantitative and qualitative experimental results manifest the superiority of our approach in generating diverse and high-quality shapes with complex topology and structures, clean surfaces, and fine details, exceeding the 3D generation capabilities of the state-of-the-art models. CCS CONCEPTS• Computing methodologies → Shape analysis; Neural networks; Mesh models.

show abstract

Section: Comparison With Other Methodsmentioning

confidence: 99%