ControlNet — Generative AI glossary — LiliDi
A conditioning network that locks a generated image to a structural reference (pose, depth, edges).
By LiliDi Editorial
A conditioning network that locks a generated image to a structural reference (pose, depth, edges). ControlNet is a family of conditioning networks attached to diffusion models that let you constrain output by pose skeleton, depth map, edge map or scribble. It is the standard tool for replicating a specific composition or character pose across many variations.