Adventures in AI on Christmas Day

 

So, I’ve been playing around a bit, trying to figure out how I could get Google Gemini to maintain the same look of a character throughout a series of pictures, because up until now Gemini keeps changing the way a character looks, and that's really frustrating. It does seem there is a way and this is my first attempt. I’ll preface this by saying I’m not very good with computers. I don’t have Chloe’s programming skills, so this is me probably doing everything the hard way.

 

As an experiment I took the first picture I had randomly generated as “Emma’ (the one at the top of this post) and, because I was quite happy with it as a ‘look’ (her hair is cut short at the beginning of chapter one), I used it as a ‘model’ for any picture with “Emma” in it.

 

Basically, in each subsequent picture prompt, I uploaded the original picture and defined that Gemini should stick with the same appearance, which it seems to have done. Note, I haven’t tried to stabilise any other characters in the picture – it’s really just an experiment to see if I could do it first with just “Emma”. But if I could, then in theory I can do the same with multiple characters. These then are the pictures I generated that way, and I think this time around the appearance of “Emma” has been stabilised, which suggests I can do that going forwards for any character I’m happy with. See what you think.

 

Of course it was only after generating all the images based on the original that I realised I should have changed the original picture first to make the slave collar a bit narrower. Oh well… bit of a blonde moment. ;) Gemini is still very restrictive when it comes to content, so it has limited applications for me, but it's there for when I'm really desperate for pictures to go with stories!