The author proposes a novel end-to-end 3D-aware image generation and editing model that disentangles appearance features from shape features, incorporating multi-modal conditions for flexible tasks. The approach outperforms alternative methods in both qualitative and quantitative aspects.
Proposing a novel approach for 3D-aware image generation and editing, focusing on disentanglement of shape and appearance features, multi-modal control, and superior performance.
Proposing a novel end-to-end 3D-aware image generation and editing model with multi-modal conditions.
Innovative disentanglement strategy for 3D-aware image generation and editing with multi-modal conditions.