Coupled Diffusion Sampling
for Training-Free Multi-View Image Editing

Anonymous Submission

Spatial Editing Results

Input video Coarse Edit Ours Liu et al. 2022 Du et al. 2023 SDEdit Per image Image-to-MV GT

Stylization Results

Input video Ours Liu et al. 2022 Du et al. 2023 TEXTure Hunyuan3D Per image Image-to-MV

Relighting Results

Neural Gaffer Relighting

Input video Ours Liu et al. 2022 Du et al. 2023 GT NeRF + Neural Gaffer Per image Image-to-MV GT

IC-Light Relighting

Input video Ours Liu et al. 2022 Du et al. 2023

Illustration: Effects of Coupling Strength

Here we visualize how different coupling strengths affect the output of Stable-Virtual-Camera for spatial editing. We show the original input we would like to edit, and the target GT edit. Note that Stable-Virtual-Camera only sees the input video through the couopling with the 2D model.

Input Strength of 0.0 Strength of 0.01 Strength of 0.025 Strength of 0.05 Strength of 0.1 GT