Okay, so I’ve been messing around with this “stable diffusion” thing, and I wanted to try something a little…different. I had this idea to generate an image of Sable, you know, the WWE wrestler, but like, more muscular. I dunno, it just popped into my head.
First, I fired up my local stable diffusion setup. I’m not using any of those fancy online services, I like to have full control, even if it means my computer sounds like a jet engine sometimes.
Getting Started
- I opened my terminal.
- navigated to stable-diffusion-webui folder.
- Then I typed
Then I waited.It takes a few minutes to initialize all tools.
I opened the browser and typed http://127.0.0.1:7860 to see the UI.
I made sure I had a decent model loaded. I’m using “realisticVision” – I find it works pretty well for, well, realistic-looking stuff. The default one is kinda “meh” for what I wanted.
Crafting the Prompt
This is where the “art” comes in, right? I spent a good chunk of time just tweaking the prompt. Here’s roughly what I ended up with:
“(photorealistic:1.4),sable,wwe wrestler, muscular, strong, gym, sweat, intense, (8k, best quality)”
I added those “photorealistic”, “8k”, and “best quality” tags to, you know, try to get a good-looking image. The “1.4” after “photorealistic” is like, a weight, telling the AI to focus on that aspect more. I played around with different weights, but 1.4 seemed to give the best results.
The negative prompt was pretty simple: “(worst quality, low quality, normal quality:1.3), lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry, ugly, pregnant, vore, duplicate, morbid, mutilated, mutation, deformed, (bad_prompt_version2:0.8)” I basically just wanted to avoid anything that looked messed up or low-res.
Generating and Tweaking
I hit “Generate” and…waited. Stable Diffusion, at least on my setup, isn’t exactly lightning-fast. The first few images were…okay. Sable was there, but she didn’t look quite as buff as I imagined.
I went back to the prompts to add more detail.I added big muscle to see the difference.
I hit “Generate” and…waited. This time it gave me expected results. Sable was there with a big muscle.
So I tweaked the prompt some more, added “bodybuilder” to really emphasize the muscles. I also messed with the “CFG Scale” setting – that kind of controls how closely the AI sticks to your prompt. Too low, and it gets too creative; too high, and things can get weird.
Final Result
After a bunch of iterations, I finally got an image I was happy with. Sable looked like a total beast, all ripped and ready to rumble. It wasn’t perfect – there were a few minor glitches here and there – but overall, it was pretty damn close to what I had in mind.
It’s kinda crazy how you can just type a few words and get these detailed images. I’m definitely going to keep experimenting with this. Maybe next time I’ll try generating, I dunno, a hyperrealistic John Cena fighting a giant robot. The possibilities are endless, really.
