It's deja vu all over again!
I accidentally released this model a few days early and had to unpublish it but it's back with a new name and the same... everything else.
This model would literally not have been possible to make without the help of @Ziperto. Not just because of his advice and friendship but also because the model I fine-tuned was an unpublished version of Mature Citron using config settings I got from him. I don't really have the words to properly express my gratitude.
If you want some help prompting for Platelier check out this guide!
Image Settings
Sampler: DPM++ 2M Karras / Euler A/ Restart
Steps: 25–50 (base), 30–50+ for Hi-Res Fix
CFG: 3–7
Clip Skip 2
This is my most ambitious model so far, I used over 1200 images to fine-tune a beautiful painterly style while keeping the model's versatility intact. I love playing around with artist and style tags so I had to make sure this model was capable of handling whatever I could throw at it. I included a lot of different styles in the showcase images using only tags and no LoRAs. I did test with multiple LoRAs and the model worked well with all of them.
As I mentioned this model is based on a Mature checkpoint but I wouldn't call it Mature itself. It's capable of that style with prompting with words like "mature, adult, aged_up." These tags will also change the body type to be curvier.
There was a price to pay for the style, I don't think the prompt adherence of this model is quite as good as my last 2 models. It's still solid though and any problems I've had I was able to fix with tag weights and parenthesis around the important tags.
Description
FAQ
Comments (9)
One thing that is never mentioned in the comments/reviews: is your ability to improve the quality upon each of your models or updates. Crispy clear images, intricate backgrounds, stunning visuals, highly detailed eyes, to name a few features! All these, without baking in any loras or boosters instead only hard work, with your blood, sweat and tears. Such a shame, your models are hidden gems!
Thank you so much! I really appreciate the kind words and all the help!
What is the purpose of including quality tags 5 times in the positive prompt in your showcase images? E.g. “masterpiece”. I would think it was enough to include “masterpiece” once to get the desired effect.
No purpose, it's just copy and paste errors from splicing different prompts together. They don't hurt anything so I don't really worry about cleaning them up
Redundancy can be good to achieve a good picture, worst case scenario it does nothing, best case it adds a little variation to a particular setting, giving you the perfect image to display/post!
a how work for furry?
I'm not sure. I wouldn't think it would be very good at that without using loras, I didn't have any furry images in the training data I used for the finetune.
First anime based checkpoint I found that obeys to certain dynamic pose instructions in the prompt, so I didn't need to use ControlNet for that. Awesome!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.









