3D Shape Synthesis via Content–Style Revealing Priors

Remil, Oussama; Xie, Qian; Chen, Honghua; Wang, Jun

doi:10.1016/j.cad.2019.05.017

Cited by 2 publications

(3 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We generate surface patches by growing from each center until the geodesic radius reaches a threshold r = 0.07 of the shape's bounding box diagonal. Here the radius r is decided based on experience from early patch‐based works [HLK*17, RXCW19], which should be local but large enough to provide meaningful perceptual information. Motivated by the style‐learning works [LKS15,HLK*17], we use a feature collection that contains both low‐level geometric information and high‐level perceptual information.…”

Section: Methodsmentioning

confidence: 99%

“…Lee et al [LVJ05] proposed handcrafted mesh saliency via local geometric cues to measure the regional visual importance of 3D meshes, and Song et al were proposed to extract the cross-category style-defining local elements. Moreover, Remil et al [RXCW19] unsupervisedly learned the content-revealing patches and style-revealing patches from a set of shapes in the same category. Our work also learns the local perceptual information from the global dataset.…”

Section: Local Perceptual Analysis Of 3d Shapesmentioning

confidence: 99%

“…As local‐based perceptual datasets are sometimes difficult to obtain, style co‐analysis supervised by global style labels [HLK*17] or semi‐supervisions like global triplet comparisons [YZX*18] were proposed to extract the cross‐category style‐defining local elements. Moreover, Remil et al [RXCW19] unsupervisedly learned the content‐revealing patches and style‐revealing patches from a set of shapes in the same category. Our work also learns the local perceptual information from the global dataset.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Learning 3D Shape Aesthetics Globally and Locally

Chen

Lau

2022

Computer Graphics Forum

View full text Add to dashboard Cite

There exist previous works in computing the visual aesthetics of 3D shapes “globally”, where the term global means that shape aesthetics data are collected for whole 3D shapes and then used to compute the aesthetics of whole 3D shapes. In this paper, we introduce a novel method that takes such “global” shape aesthetics data, and learn both a “global” shape aesthetics measure that computes aesthetics scores for whole 3D shapes, and a “local” shape aesthetics measure that computes to what extent a local region on the 3D shape surface contributes to the whole shape's aesthetics. These aesthetics measures are learned, and hence do not consider existing handcrafted notions of what makes a 3D shape aesthetic. We take a dataset of global pairwise shape aesthetics, where humans compares between pairs of shapes and say which shape from each pair is more aesthetic. Our solution proposes a point‐based neural network that takes a 3D shape represented by surface patches as input and jointly outputs its global aesthetics score and a local aesthetics map. To build connections between global and local aesthetics, we embed the global and local features into the same latent space and then output scores with the weights‐shared aesthetics predictors. Furthermore, we designed three loss functions to supervise the training jointly. We demonstrate the shape aesthetics results globally and locally to show that our framework can make good global aesthetics predictions while the predicted aesthetics maps are consistent with human perception. In addition, we present several applications enabled by our local aesthetics metric.

show abstract