notesum.ai

Published at December 4

SGSST: Scaling Gaussian Splatting StyleTransfer

cs.CV

cs.GR

eess.IV

Released Date: December 4, 2024

Authors: Bruno Galerne¹, Jianling Wang¹, Lara Raad², Jean-Michel Morel³

Aff.: ¹Institut Denis Poisson, Université d'Orléans, Université de Tours, CNRS; ²Instituto de Ingeniería Electrica, Facultad de Ingeniería, Universidad de la Republica; ³City University of Hong Kong

Arxiv: http://arxiv.org/pdf/2412.03371v1

Method	Transfer quality	Short-range consistency		Long-range consistency
	Gram $\downarrow$	LPIPS $\downarrow$	RMSE $\downarrow$	LPIPS $\downarrow$	RMSE $\downarrow$
SGSST	$\mathbf{2.59\mathrm{e}8}$	$\mathbf{0.030}$	$0.032$	$0.055$	$0.063$
StyleGaussian	$4.61\mathrm{e}8$	$0.033$	$\mathbf{0.029}$	$\mathbf{0.050}$	$\mathbf{0.056}$
ARF	$5.77\mathrm{e}8$	$0.040$	$0.037$	$0.072$	$0.066$