Perhaps the benchmark for AI image generators is
www.midjourney.com but you only get about 20 goes before you have to sign up to a paid plan. But
their showcase gallery is a great source of inspiration both for images and the phrases you use to prompt image creation (it gives you the prompt used to create an image if you hover your mouse over the image in the gallery).
Microsoft's Bing (
www.bing.com/images/create) is not as good and not as "smart"/"creative", which means that you have to spoon-feed it a bit more by giving it more detailed prompts to stop it doing things you don't want. The results still won't be as good as Midjourney but the big advantage is that effectively you have unlimited free goes, the only real "cost" is the time it takes to process so it's best done as something in the background whilst you're doing something else. So Bing is a good option for exploring what's possible and how the process of prompting works.
In general the way these things work is you say what you want to be depicted, and then the style in which you want them so for instance
Left :
beer glasses oil painting, brush strokes
Right :
beer glasses pencil sketch
View attachment 821614View attachment 821620
Perhaps the most helpful thing is naming particular artists, and there's a real art to choosing the right one, which relies on more knowledge of art history than the average person has! Which is why it's good to find out about them through other people's prompts, as with Beatrix Potter above.
Left :
beer glasses in the style of Andy Warhol
Right :
beer glasses in the style of van Gogh
View attachment 821618View attachment 821615
And it doesn't have to be just artists, you can do styles as well
Left:
beer glasses, futuristic digital art
Right:
beer glasses, 8-bit
View attachment 821621View attachment 821629
You can also specify specific filmstock, although the differences between films are not as obvious on Bing compared to Midjourney
Left :
beer glasses, Agfacolor
Right :
beer glasses, Kodachrome
View attachment 821624View attachment 821626
So in that example above, the left hand picture of the dog was on the Midjourney showcase, so I took the prompt (it's easiest to do Ctrl-U and search for it in the underlying code of the webpage) and used it as the basis of a prompt in Bing, but found myself having to edit it quite a bit to make up for Bing being a more clunky tool.
And then I edited it in an image-editing program to remove the bottom bit of the image as it wasn't very interesting. Also you can see that Bing leaves a small watermark in the bottom left corner, so you want to remove that if you're using the image for an actual purpose.
and have no idea of the scale of effort needed...
