A few days ago, the basically free and openly available Stable Diffusion was extended by an interesting feature: By means of "Stable unCLIP" it is particularly easy to generate variations of an image, which the user can freely specify as input. In short, you send in an image and similar images come out.
Stable unCLIP does not create direct, partial modifications of the original, but rather "dreams up" fundamentally new images inspired by the originals.
This technology thus has the well-known limitations of almost all AI dreaming: it can produce quite amazing results for some images and less impressive results for others. Trial and error continues to lead to success here in a rather uncontrolled manner. In the 
original Github version you can also send a text prompt as embedding to steer the changes in a desired direction.
Stablity.AI, a big backer of Stable Diffusion has started a usable web service, with which you can try out unCLIP immediately in the browser under the name "Stable Diffusion Reimagine". Depending on the motif, you can get more or less interesting and inspiring variations with just one mouse click.
However, with Stable Diffusion Reimagine you can't send a text prompt and you can't use other parameters like change strength or seed to control the "spread of variations" or to develop further "forks" from already good results.
That's why this tool seems more like a web gimmick than a service you'd like to throw money in for on a monthly basis. But for the latter (money) you indirectly support the further development of Stable Diffusion. And then you can install Stable Diffusion Reimagine / unCLIP with a clear conscience on your own computer with a fast GPU. Because the new version Automatic 1111 has of course already integrated it.
;)
;)