T2I-Adapter is an injection-based SD control, an additional model that can sit “in front” of a Stable Diffusion model to help control the outputs. See their Overview on Github.
T2I-Adapter is similar to ControlNet, allowing you to define a depth-map, outline, etc for your Stable Diffusion model’s output.
Unique Features
T2I seems to do a lot of the same things as ControlNet, and I haven’t yet explored the differences.
One unique feature for now is allowing for a style image as input (not currently supported by ControlNet). This makes T2I pivotal for doing Style Transfer From Image In Stable Diffusion.
| Sketch | Style image | Result |
|---|---|---|
 |  |  |
Using T2I
To use T2I, you can download models from GitHub - TencentARC/T2I-Adapter: T2I-Adapter
The ControlNet WebUI Extension (which is just a UI, not the actual ControlNet models) now supports T2I models as well.