Not sure what people complaining about the new model is on about, sure it’s quite a bit slower but so much more accurate, it rewards descriptive descriptions, and even less descriptive tends to turn out good imo.
Nah, he’s pretty bad… I have a rpg group and it was set up in 2028, so, many characters had social media and stuff, 8 players, 5 male characters, 3 female, all the male characters now can’t have an specific hair type, and ALL the females no matter how you change the prompt, they come up with fake surgery lips and generic e-girl faces, casual photo, professional photo and cinematic are unable to create variants, at all.
I get that it is learning, people who enjoy this model are probably just creating signs and anime, and even then, I’ve seen a lot of people complain about Anime filters
I’m pro new AI model because it generally tries to follow your description but I can agree that this insanity with lips contouring ruining it. I’ve tried negative and positive prompts with hundreds of “lip contouring”, “lip surgery”, “natural lips”, “thin lips” etc, the same for brows. It’s ugly in real life and it’s ugly when AI tries to copy it, but unfortunately I couldn’t get rid of it completely. The only hope is that it’s still learning and we’ll get the variety with time. And if someone found a solution, please make a post about it.
I’ve had the opposite going on, i feel i can very much specify what i want a character to look like, if i don’t want surgical lips, i’ll just tell the AI to generate her with natural lips, but this model is way more detail oriented, it wants a detailed description of what you want to see, if you supply that, it’s really nice, only big downside is that it’s slow.
I respect your opinion, the chat is chaotic enough as it is, but some models, I swear, they dont change, I’ve used about 15+ prompts for lips, nothing, it feels like the same images that Midjourney discord bot generated, I’m glad you like it, but for me, its a waiting game, either the old model return as a option, or, wait for the AI to learn some variations
I really don’t get why people are complaining the only bad thing i can say its that its slow and the images look different from what i made before .but i kept using the same anime style because the others looked bad the more western styles would always have messed up fingers and toes .with this one that happens less but if i use the Anime like styles now they all like different so that can be both a good or bad thing depending on your preference . but the old one couldn’t even do basic things like drawing people shaking hands it would always be messed up no matter what i did
Try cel shading cel shaded artwork with bold line work
