ControlNet: How to Steer Stable Diffusion Without Breaking It

6 Просмотры
Издатель
ControlNet lets you add pose, edge, depth, and segmentation control to pretrained text-to-image diffusion models like Stable Diffusion. It works by locking the original model, training a copy, and connecting them with zero convolutions so the pretrained backbone stays intact even with small datasets.

Категория
AI Картинки
Выполните вход или зарегистрируйтесь для отправки комментария.
Комментариев нет.