Krea 2 v1: I dusted this one off because I could not get it quite just right with some of the more recent models but Krea 2 came (🥵) to the rescue!I created this so that you don't have to describe the characters as if they were toys, although that's fine if you do, and instead allow to generate the characters as if you were describing real humans. That way any existing prompt templates you may have should be able to be easily reused with it or if you caption a irl image and want it reproduced in this style you don't have to carefully re-craft the prompt to fit it as if it were a scene in the actual show.I also tried to see how multiple characters would work with an adjusted captioning strategy by naming them with unique tagged names but that works only so-so. While this LoRA seems to be able to reproduce most of the primary characters fairly easily, you more than likely will need to be a bit descriptive on their personal traits to get it right, such as brown spiky hair and red shirt for Steve, tan skin with blue shirt for Rick, etc. etc. Or just see what it comes up with on its own. The tags that I used are in the model's trigger word section and in the metadata but generally it's rsthgciaw-steve, rsthgciaw-rick, rsthgciaw-evan, rsthgciaw-chuck. I wouldn't get your hopes up thinking those triggers will do the character specification heavy-lifting, so you'll need to be a bit descriptive to get the characters right, but I think they help a bit. For shits and giggles, I threw in their cat in there just for fun - and it works.Lastly, since its me - you know this def can produce some male-focused spicy results (maybe spicy female results too? I didn't test for that but who knows - it's Krea). Of course the TV series only showed them in their painted clothes on so I had to improvise with some edits where I tried to make it somewhat plausible. Well, it works but their wants do seem to be a bit more realistic than what I made some of those training images so I think that's Krea 2's base model bleeding through but it generally works out fairly well.Enjoy.<hr />Rick &amp; Steve: The Happiest Gay Couple In All The World!<a target="_blank" rel="ugc" href="https://en.wikipedia.org/wiki/Rick_%26_Steve:_The_Happiest_Gay_Couple_in_All_the_World">https://en.wikipedia.org/wiki/Rick_%26_Steve:_The_Happiest_Gay_Couple_in_All_the_World</a>Short lived show in the early 2000's on Logo.. It still holds up today.This one I am trying a few different approaches when it comes to multiple concepts in a single LoRA with a relatively low amount of training images. I added some modified versions to create some flexibility :) ..I call this a version 0 because I was expecting to have to create a base LoRA to generate additional training images to get a decent one going, but turns out even the initial version is still pretty solid.main trigger words: ricknsteve, ricknstevesteve, ricknsteverick, ricknsteveevan, ricknstevechuck(You will need to specify wheelchair in the prompt if you want Chuck to be accurate as depicted in the show)..

Files with SHA256: a3775721447e196426391b4ac979bd10362ac42fed70f9e76aa1bd45d0a93875