“We sometimes take photos (or record a video) so that we can later see what was happening while we were busy taking photos (or recording a video).” ― Mokokoma Mokhonoana
Recently I trained AI with Selfies in Nightcafe, this is a new feature: How to Train AI Image Models Step By Step and next step Fine Tuning - I wanted to test this and, vanity took over, since I tried to make selfies before, I trained my first (and until now only) model with 37 selfie photos. The first few I tested I done with the preset modes, but soon discovered I could use also my weirdest prompts in combination, just add the model first in SDXL 1.0, then add your own prompt and select your format. Now that I know how this works, I will train AI next with my drawings, so it learns my style. I have two more free models I can train, after that it costs 100 points to train additional models. I will have to see how this works with drawings first. The Selfies did work out OK for the most part (there was one fail, which I deleted).
So lets get this going with the very last one I created yesterday
This after I had figured out how to combine the model with my own prompt.
The promt used was this:
<lora:SELFIES:1.0>acrylic pour art - The Inner Life of the Artist Otto Rapp - his Alternate Reality in Bogomil's Universe; macrocosmic reflection, intricately detailed; Hieronimus Bosch, imperial colors, fantasy, International Gothic, oil painting, mysterious, cryptid taxidermy
I also added a seed image of a 'Acrylic Pour Art' creation from earlier. But in my haste, I forgot to activate the negative prompt, thus my hands look a bit weird, only three fingers! Considering that the prompt contained 'cryptid taxidermy', what goes on all around also affected my hands, I guess.
To give you an idea how weird the basic prompt without the selfie model could turn out, here are two examples of it:
But lets get back to the selfies, in random order
GUITAR HERO this one turned out better than expected, combined with a surreal prompt
CYBORG BARTENDER there was cyborg bartender in the prompt, but this looks more like a happy Hippie
TWINS IN SPIRIT I actually wanted to be spiritual twins with this beauty at the bar, but AI created a ghostly twin of myself
STEAMPUNK GUITAR HERO back again playing guitar in a steampunk outer space sort of style, get the shades!
SUNBURN ME IN ACRYLIC POUR AI did it to me with too much sun exposure
NATURE BOY AT THE WATERFALL reminds me of my time at the Multnomah Falls in Oregon near the Columbia River
THE SERIOUS EXECUTIVE IN A SURREAL WORLD a bunch of surrealists from my prompt are going nuts behind me!
GROUNDED ANGEL LOST HIS WINGS not what I expected, nor the sad looking lady behind me
READY FOR BATTLE one of the first tries, with a preset mode for knight
THE WIZARD OF EARTH-SEA also a preset mode, one of the first images generated
SUPERHERO IN RELAXATION MODE also a preset mode - not quite sure what AI was thinking here
WOODSTOCK MEMORIES this was the very first one I tried - a preset mode
CYBORG WARRIOR lots of gadgets, but I am afraid the fufram valve was connected backwards!
JUST PRETENDING lots of blings, the ruler of Bogomil's Universe! - the prompt was about Egyptian Kings
STRAGGLERS
late night still playing, mixing it up like crazy
There were more, but now it is time to give it a rest - thanks for tuning in!
