We use the threestudio implementation for all the baselines. A fixed configuration is used for each method (including ours) without prompt-wise hyperparameter tuning. The results here are without cherry picking.
Dreamfusion-IF
Magic3D-IF-SD
Text2Mesh-IF
ProlificDreamer
Ours
Pikachu with hat
animal skull pile
saber from fate stay night, 3D, girl, anime
a DSLR photo of a squirrel playing guitar
A statue of angel, blender
A product photo of a toy tank
TRUMP figure
Handpainted watercolor windmill, hand-painted
girl riding wolf, cute, cartoon, blender
baby yoda in the style of Mormookiee
@article{shi2023MVDream,
author = {Shi, Yichun and Wang, Peng and Ye, Jianglong and Mai, Long and Li, Kejie and Yang, Xiao},
title = {MVDream: Multi-view Diffusion for 3D Generation},
journal = {arXiv:2308.16512},
year = {2023},
}