Learning Detail Transfer based on Geometric Features (bibtex)
by Sema Berkiten, Maciej Halber, Justin Solomon, Chongyang Ma, Hao Li, Szymon Rusinkiewicz
Abstract:
Abstract The visual richness of computer graphics applications is frequently limited by the difficulty of obtaining high-quality, detailed 3D models. This paper proposes a method for realistically transferring details (specifically, displacement maps) from existing high-quality 3D models to simple shapes that may be created with easy-to-learn modeling tools. Our key insight is to use metric learning to find a combination of geometric features that successfully predicts detail-map similarities on the source mesh; we use the learned feature combination to drive the detail transfer. The latter uses a variant of multi-resolution non-parametric texture synthesis, augmented by a high-frequency detail transfer step in texture space. We demonstrate that our technique can successfully transfer details among a variety of shapes including furniture and clothing.
Reference:
Learning Detail Transfer based on Geometric Features (Sema Berkiten, Maciej Halber, Justin Solomon, Chongyang Ma, Hao Li, Szymon Rusinkiewicz), In Computer Graphics Forum, volume 36.
Bibtex Entry:
@article{doi:10.1111/cgf.13132,	
	author = {Berkiten, Sema and Halber, Maciej and Solomon, Justin and Ma, Chongyang and Li, Hao and Rusinkiewicz, Szymon},
	title = {Learning Detail Transfer based on Geometric Features},
	journal = {Computer Graphics Forum},
	volume = {36},
	number = {2},
	pages = {361-373},
	doi = {10.1111/cgf.13132},
	url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/cgf.13132},
	eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1111/cgf.13132},
	abstract = {Abstract The visual richness of computer graphics applications is frequently limited by the difficulty of obtaining high-quality, detailed 3D models. This paper proposes a method for realistically transferring details (specifically, displacement maps) from existing high-quality 3D models to simple shapes that may be created with easy-to-learn modeling tools. Our key insight is to use metric learning to find a combination of geometric features that successfully predicts detail-map similarities on the source mesh; we use the learned feature combination to drive the detail transfer. The latter uses a variant of multi-resolution non-parametric texture synthesis, augmented by a high-frequency detail transfer step in texture space. We demonstrate that our technique can successfully transfer details among a variety of shapes including furniture and clothing.}
}
Powered by bibtexbrowser