You asked how image creators should earn, and 240+ of you answered. Here's what we heard and what we're shipping: New & Upcoming discovery feeds (live), remix galleries and sticker placement (coming), and real new uses for Blue Buzz.

How image creators earn: what we heard, what we're building

We're a Day-0 launch partner for MiniMax H3, a next-gen open-weights multimodal video model - and it's already in the generator. No waitlist. Open weights land in a few days.

MiniMax H3 is live: open-weights video, generate now

Krea 2 LoRA training is 50% off, and we're running a Krea 2 Training Contest with 1.8 million Buzz in prizes across four LoRA categories.

⚡ Krea 2 Training Contest + 50% Off Training

This is a style model that does art like game cg artist 3ly. The artist specializes in demon girls, and textured hair. Tends to work best at 0.6. This model was trained on NAI anime-full-pruned and generated in Anylora (v1 v2), Cute Yuki Special (v2.5, v4.0), based68 (v5). Made with the Kohya tool for AI Training through google colab (v1-v4), and local kohya (v5). Any suggestions to improve or recommended things for the data set would be appreciated. To get the best effect try prompting: full body, standing, looking at viewer

Trigger word optional if effect isn't strong enough. First full lora I made on my new computer using local Kohya. Trained on anime final full pruned and generated in based 68. added a few new work to the dataset. Since this training was very different than the original please comment on if you think version 4 or 5 was done better. Probably will go back and update several lora to this new style that is no longer restricted on training time.

v5.0

OKAY this dataset is of 276 different images. this has picked examples from promotional art, artbook references, individual pieces, and game cg's from the earliest editions to the ones still in development. This removed some of accidentally included images from earlier sets includes more diversity and isn't permanently locked onto full body shots switching from cowboy/upper body/ full body without as much difficulty. Only problem I can say is that it doesn't do from behind and some other angles too well, but that is because this artist simply doesn't do shots from other perspectives. This was trained to the best my computer can handle of 6 repeats on 10 epochs if I ever upgrade I'll update this to the 10 by 10 it deserves, but it does an amazing job as it is.

v4.0

This was a minor patch up is pretty much the exact same as version 2 but I added 15 cropped and cleaned images to help make upper body/cowboy shots. Version 2 I do think Is a terrific model for full body shots and I use it to show off clothing lora all the time, but it has trouble with other types of images. This patch hopefully improves other types a little bit. I have plans to make a full version 3 eventually I did find a lot more content to use, but it will take a while for me to go through and choose the best images. I'll probably go back and increase resolution of smaller in game cg images and pics (since they are tiny) as well, but that will take a while to. The lora itself is named 3ly3 (for file organization purposes), but this isn't a major update to be proud of just a minor patch for now.

v2.5

Cut down to 156 images by removing similar images that only varied in expressions, did some retagging to focus more on clothes, and added simple backgrounds to the standing portraits (to make background prompting easier). Which made it a little easier to control and easier to prompt. A weakness I would say is this artist draws almost everything straight on so there is very little alternate angles (a bummer since I like trying to spin characters around) To get the best effect try prompting: full body, standing, looking at viewer Credits: UBW by FallenIncursio

v2.0

Made on 180 images of mainly full bodies. Some notes: Most images were the standing portraits from games. This means this was trained with very little backgrounds. As is has a very strong effect of the artist which is good, but it seems to not be as consistent as I would like. Once I skim through the dataset again, I'll make another version of this. This is the first version so, Here are some notes for myself: Go through and double check all tags from dataset Remove similar images from training to cut down on step count

v1.0

3LY Style

physical violence

weapon violence

wide hips

revealing clothes

downblouse

convenient censoring

pg-13

corpses

suggestive

oral invitation

pg13

sexy

huge breasts

thick thighs

sexual situations

male nudity

disturbing

male swimwear or underwear

female swimwear or underwear

partial nudity

undressed

female nudity

breasts out

exposed female nipple

breast out

lingerie

male underwear

hair over breasts

female swimwear

gigantic breasts

no panties

graphic violence or gore

covered nipples

huge butt

strapless leotard

sitting on face

emaciated bodies

one breast out

female underwear

nude

nsfw

graphic male nudity

adult toys

illustrated explicit nudity

nudity

graphic female nudity

hentai

futanari

porn

sexual intent

genitals

peeing

vore

oral

sexual activity

anal

blowjob

dildo riding

incest

hanging

hate symbols

nazi party

white supremacy

diapers

scat

self injury

hate speech

urine

extremist

child on child

latex clothing

swimwear

bukkake

fellatio

cumshot

implied fellatio

eat_cum

cumdrip

cum in pussy

cum on face

after fellatio

cum on hair

cum on body

cum on tongue

cum on hands

cum in mouth

triple fellatio

autofellatio

fucked silly

cum on pussy

Writers