So my thoughts are probably rather disjointed here, as this started out as just a reply to a question someone else posed elsewhere, but now it's really got me thinking & second guessing things.....
I originally intended to write an in-depth guide on how to recreate specific faces from dreams & imagination, that can't simply be trained via conventional data sets.... as obviously none would exist yet...
Original response I made follows;
I think the coolest thing about AI images, is recreating things in your mindseye, & making them reality. I <3 Sci-fi & Dark Souls like dark fantasy, but I mostly like recreating things from my dreams, & imagination. Recreating characters/scenes that already exist (real or fictional)...lets say from something like Star Wars, is just way less interesting to me.
I think people mostly equate face swaps (ReActor, etc) mostly as celeb, or fictional fan scenes/mixing live-action characters, etc....but you can also use it to create & swap faces that don't exist anywhere yet. This is what interests me about it.
"ReActor composite image face swap (artificially blended together custom character models)."
I'm intending to get around to creating a guide detailing how to recreate exact faces from dreams & imagination; that don't actually exist anywhere else yet, so can't just be trained conventionally like real faces from existing photos.
.....I dunno....Perhaps no one is really interested in this except me lol (I kind of naturally assumed others would be, too; but now I'm starting to think that may not actually be the case in a really big way lol). Perhaps other people are happy with just generic attractive faces in image generations, rather than exact artistic recreations from their imagination, & haven't given it much thought otherwise. But this is what has interested me since the first day I actually got into AI image generations. I used to recreate/sketch scenes from my dreams when I was younger via more conventional artistic means, but it's too time consuming.
I can see things in my head literally like a lucid dream though (If I'm relaxed enough, actually seemingly identical to convetional dreaming when I'm fully asleep). When I was taking Ritalin, it was often unclear to me whether I was even dreaming or not whilst trying to get to sleep (or even capable of technically sleeping from having that much of it still in my system).
Apparently this sort of vivd imagination & mental picturing is way more common if you have ADHD/OCD, but I gather not so common to that extent for most people out there (which I still find hard to fully believe). From what I gather, approx. 10% of the general population can't do this to any capacity whatsoever (Research Aphantasia). Whilst the remaining majority can only summon vague images of things in their head that lack such extreme detail, or from what I've heard others describe, something more akin to "semi-vague images from imagination covered in a smoke like haze like a mirage".
Anyway, I'm intending to try to get around to making a guide on this (unless there's just no interest....which is kind of making me second guess the whole thing now lol).
Given this would be quite a time consuming & detailed guide for me to create.... is this something that would illicit any actual interest ? Because I'm intending to write a ton of fluff 10min quick read basic guides I think would be helpful, but this was going to be a comprehensive guide probably spanning over a good month of back & forth until it's done....
Now I'm really starting to think I may have too much tunnel-vision in pursuing this particular idea.
"this is what has interested me since the first day I actually got into AI image generations.", others would be interested in that sentiment & share it, or no? (I think I'll probably need to sleep a day to get clarity on this, & rethink it, then come back to it).