New and improved!  Now uses the dw pose estimation node to calculate the open pose image internally.

What this workflow does

This workflow creates a costumed image of a character from just a headshot, a background style image, a pose image, and a text prompt to edit.  It created images of me in a suit, and images of a woman in a wedding dress.  

This workflow is almost not fun because it creates a rock solid image in one shot nearly every time.  No more generating hundreds of images and picking the best half dozen to photoshop.  No extra hands, fingers, arms, or feet.  Of course if your job is making people in a specific pose wearing a specific outfit, then you want a tool that just works.

How to use this workflow

Put in a headshot of the person whose face you want to use.  It helps if this headshot is also wearing the final costume.  The top left image is fed into the ipadatpter and the face swap nodes. The bottom left  image is only used by ipadapter.  The bottom center  pose input image is an image that is used to calcuate the open pose image internally.  

Then describe the image output you want in detail, using english sentences in the positive text prompt.

Add the workflow to the queue.  If you get something in the output you don't like, add it to the negative prompt, and adjust the positive prompt to better describe what you wish to see.

Tips about this workflow

Make sure the style images have the same style, and that the prompt is describing the same style for the output.   There are no adjustments to make.

The open pose image is now calculated inside the workflow, so it just takes 2 style images and a pose image. The top left style image is also used to face swap.