DiffusionSat ControlNet Texas Housing Super-Resolution Checkpoints

ControlNet (conditional) DiffusionSat checkpoints for the Texas housing super-resolution dataset. For this task, given a high-res and low-res image of a location at time T, and a low-res image of the same location at time T', the model must generate the high-res image of the location at time T'. That is, this model is capable of [caption, metadata, conditioning_images, conditioning_metadata] --> RGB image.

The model generates RGB images at a resolution of 256x256, but uses the 512x512 single-image model as the frozen prior.

This archive contains: 

• checkpoint @ 50k
• optimizer @ 50k
• model config files

Please refer to our Github repo for more details.