MV-Adapter
MV-Adapter enhances pre-trained T2I models for efficient multi-view image generation, retaining strong priors and supporting 3D geometry modeling.
English
English
What is MV-Adapter?
MV-Adapter is a novel adapter-based solution for multi-view image generation that enhances pre-trained text-to-image models without altering their structure, enabling efficient training and retention of prior knowledge. It supports 3D geometry modeling and integrates camera parameters and geometric information for versatile applications.
