short answer: no.
the way ganbreeder works is (oversimplifying things) taking a square filled with random noise and doing math to it until it resembles an image. the “gan” part of the name stands for generative adversarial network, that’s what you want to read up on if you’re curious about the architecture!
my friend, the person who revamped the software, said it could be possible to allow you to use an image as the random noise, but they haven’t actually experimented with that so it isn’t a functionality we have right now. i’m pretty curious about how that’d look, i’ll nag them a bit about it at some point but they’ve been pretty busy lately.
if you want to add your own images into the mix, you’re looking for generators with img2img capacity, though i don’t really have any to suggest. i’ve used a couple in the past, but they likely no longer work because of old dependecies. they’d need some rewriting / fiddling to work nowadays.
of course at least some of the modern generators, midjourney and the like, do that as well. not really in the spirit of the jam though haha. i’m also not familiar with them actually, so i don’t have any recommendations if you wanna experiment with those.
personally when it comes to mixing up ganbreeder output with my own stuff, i go the route of collage. i think of the images as magazine spreads and just cut out any shape / texture that seems interesting, then recombine that. that’s obviously not the only approach, just what i tend to use since getting very specific visuals out of ganbreeder is rather difficult / counterproductive.
hope my answer helps ^_^