Spic·E: Structural Priors in 3D Diffusion Models using Cross-Entity Attention

Supplementary Material

Click below to view our interactive results, comparisons and ablations:

• Our Results
• Comparisons - Semantic Shape Editing
• Comparisons - Text-Conditional Abstraction-to-3D
• Comparisons - 3D Stylization
• Ablations

Ablations

We compare our cross-entity attention mechanism with several baseline methods, detailed in Section 5.2.
As illustrated below, these baseline methods struggle at generating shapes that both conform to the target text prompt and preserve the structure of the guidance shape.