prompt that genertes a low variety

For some reason, when I strip down your prompt to "baby cat, heavy strokes, dripping paint" and I use the style menu on the right to select Concept Art and Fantasy, I am able to get a little more variety with the types of cats. The style of dripping paint is a different issue. 

Regarding giving feedback, at the moment we don't have an option for all 4, but if you just rate one of the images and give us some reasoning in the text field, we'll see it.

Hey thank you for getting back to me.

I should have explained the problems.

This promot tests the models ability to generate images that are in a drip paint style while exploring it with different cat's.

Now I'm finding thst the model isn't able to generate a variation of cat's and there all in the same almost whater colour style.

This points to a lack of training data in the area of cats and drip paint.

I tried changing the promot to fit firefly but the cats stay more or les the same. An embedding may help but as of now that not a feature i can test.

There was no button to rapirt all generations on one prompt so i hope this was a good way to give this feed back

Thank you so much for the good work on firefly!

Thank you for the feedback. I'm not sure what exactly you are trying to produce, but I would suggest playing around with the prompts more and substituting different words. We use Adobe Stock captions to train our models, if that helps you discover the types of words that may work well with Firefly.