Does what it says on the tin.. Freshly generated for your viewing pleisure.

These turned out great!!! So very hot and messy!

These are amazing, seeing as Bing throws lots of dogs around these days.

If you're getting 1 image for every 3 dogs, you're still getting images and your prompt is on the money. That's the way to do it with Bing right now IMO.

1:3 to 1:5 is a sweet spot with current filters. I can engineer a prompt that will do 1:1 with multiple returns but the results are dull and lifeless.

I'll tend to run a prompt to exhaustion for a couple of reasons.
1) The variability, I might get 1 image worth keeping for every 5
2) My longer term intention is build a large fine tuned model and I want to avoid copyright images. A larger dataset I can sort through is going to be stronger longer term.
3) I will tweak very small elements of a working prompt. Camera angle, poses, even just the colour of the substance. Again, the variability of the image will enrich a larger finetune.
4) I will also generally scale back the prompt to increase variability. Removing fine detail in the particular outfit will greatly increase success rates. Let the AI do the work and it will also give more variability to images.

These are amazing.

I wish there was some way this could be replicated to underwear, or Bra & Knickers etc.
As though they had spared their clothes before the dunking in the slime .