Strangely enough, my attempts at creating WAM images are relatively consistent. I have a pretty good idea of what will get through the filters and what won't, and I can get very specific with the clothes worn and poses used.

It's when I try to do non-WAM pictures of people that I consistently get the dog.

I love clothes and fashion in general and want to create inspiration images for the stories I write. I also like using descriptions I find in books to create pictures of what those characters might look like. A lot of these images get blocked by the dog, even the really generic ones, and I have no idea why.

At first, I thought it was the styles or fabrics, like necklines, leggings, heels, leather, satin, etc. But I can get super specific with colors, fabrics, styles, etc in my WAM images with only an occasional dog (my WAM stuff is pretty PG for the most part).

My most recent attempt was a sort of superhero pose with a woman in black crouching on a rooftop overlooking the city, like Catwoman or Dark Angel (Jessica Alba's TV show), with her hair blowing in the wind. My most recent prompt was:

"photo of a woman wearing a sleeveless black top and long black gloves and black leather leggings and high-heeled black leather boots crouching on a rooftop overlooking the city streets, long blonde hair waving dramatically in the wind"

I've done so many variations of that, using synonyms, similar clothing items, generic clothing, etc. I even stripped the prompt down to "photo of a woman in black on a rooftop." That stripped-down prompt got the dog twice, and only generated one image the other two times.

I heard somewhere that your prompts are censored much more strictly when you repeatedly get the dog, so I retreated back to a WAM prompt I've had success with: a princess washing a car in a ballgown in the rain, super specific with the design, color, and fabric of her attire, including references to wet hair and wet clothing. All four images generated, no issues. Tried it again, three successes, each with changes to the outfit's description. Went back to the woman in black: just the dog.

But I can't get basic fashion advertising-type images to generate. This is driving me nuts. Even the prompts that include specific names get blocked less often than these attempts at a woman in black. I just don't get it.

Any tips, advice, or possible ways to generate this image in my head would be greatly appreciated. My brain won't let it go.

It's been mentioned here before, and it is true, that if you add extra details not pertaining to the main thing you want, it will 'dilute' your prompts and distract the filters. I was doing outdoor mud pictures, and by adding 'trees in the distance with a bird flying past' it got me far fewer dogs. A few non-related details can help a lot.

Thank you! I added details about the moon and sky, and started getting 1-3 images per attempt. And while wetlook wasn't my intent with these, I did add it after I got some images I liked. This one was one of my favorites of the wet ones.

It's crazy that simple images of a fully dressed woman just standing there doing nothing get blocked by the filters. That's gotta drive novice AI users nuts.