notesum.ai

Published at November 22

Material Anything: Generating Materials for Any 3D Object via Diffusion

cs.CV

cs.GR

Released Date: November 22, 2024

Authors: Xin Huang¹, Tengfei Wang², Ziwei Liu³, Qing Wang¹

Aff.: ¹Northwestern Polytechnical University; ²Shanghai AI Lab; ³S-Lab, Nanyang Technological University

Arxiv: http://arxiv.org/abs/2411.15138v1

[Uncaptioned image]

Method	Type	Input Mesh	FID $\downarrow$	CLIP Score $\uparrow$
Text2Tex [6]	Learning	Texture-less	116.41	30.33
SyncMVD [24]	Learning	Texture-less	118.46	30.66
Paint3D [47]	Learning	Texture-less	153.20	28.40
NvDiffRec [26]	Optimization	Texture-less	103.81	30.14
DreamMat [52]	Optimization	Texture-less	113.34	30.64
Ours	Learning	Texture-less	100.63	31.06
Make-it-Real [12]	Retrieval	Textured	104.38	88.62
Ours	Learning	Textured	101.19	89.70