What this workflow does

This workflow adds an external VAE on top of the basic text-to-image workflow ( https://openart.ai/workflows/openart/basic-sd15-workflow/lkOtNJ2UexVd6vK0kYhd). The VAE model is responsible of converting the latent image into the pixel space. The process is "lossy" and a good VAE model is essential to get good quality images.

How to use this workflow

This one workflow uses a default good VAE but there are more. Check out the description on Huggingface or CivitAI if the author suggests a specific VAE.