This workflow replaces Stage C of Stable Cascade with SD 1.5.

Some reasons for this:

1. SD 1.5 is really fast compared to Stage C Stable Cascade
2. You can add your favorite Lora as stack
3. Embeddings support is better than Cascade
4. Supports NSFW indirectly
5. I wanted to see if it can be done

Some of you will wonder why not just use SD1.5 itself.

Well you can, but, I found Stage B really polishes the image well and bring out details and does not hog Memory on low VRAM machines when dealing with huge image sizes. (I could create images upto 4k on my 4050 6gb easily)

PS: I have included Kohya Deepshrink support if you intend to start off from really huge image sizes to avoid duplication of subjects.

Feel Free to modify or share!