Some projects can change scope, while others are overshadowed by more optimized approaches. However, many of these nodes continue to work and provide functionality well after they are abandoned by their developers.

SeargeSDXL’s XL prompt encoding and sampling nodes are some of the finest that I find the be both the most efficient and optimized way to chain together Sable Diffusion XL workflows.

To install this extension, I would recommend downloading the .bat file from the GitHub and following the instructions. It is not advisable to use the extension manger for this.

The prompt encoder for XL is great because it gives the proper context to the prompt as you conduction the latent space during diffusion. You can separate the various positive and negative text inputs, but I find this node set up works well.

If you notice the default values for base/targe/refiner height/width as well as the “ascore” values are all here and easily modifiable. Once you have your Base and Refiner clip plugged in you can minimize the node for a cleaner workspace.

The “Sampler v3” is a highly efficient sampler node that will handle both the Base and Refiner of XL and properly diffuse according to your settings.

I like to create a “Primitive” node for the seed so I can use it for other samplers I have in my workflows at the same time to share a common seed value. This also works with the Sampler Settings node seen above.

The “Prompt text Combine” node allows you to chain together sections of text and works well for assigning file names with folder directories and adding LoRA activation text to prompts.

This image should contain the node workflow I used above.