Sign In

PAYDAY 3 - Joy [ILXL/Pony]

7

60

167

3

Verified:

SafeTensor

Type

LoRA

Stats

24

11

6

Reviews

No reviews yet

Published

Feb 26, 2025

Base Model

Pony

Training

Steps: 1,137
Epochs: 15

Usage Tips

Clip Skip: 1
Strength: 1

Trigger Words

pd3_joy
score_7_up, score_8_up, score_9_up, rating_safe, BREAK pd3_joy, 1girl, solo, gloves, jewelry, green eyes, necklace, trench coat, pants, blue gloves, arms, black hair, vest, shirt, black footwear, short hair, belt, boots

Hash

AutoV2
A4BBB825E9

Hello. This is my Illustrous Lora Model of a Videogame Character Joy from PAYDAY 3.
Trained on my very large dataset of 379 images. Includes her Mask as well.

________________________________________________________________________________
Activation Prompt: ``pd3_joy``
Recommended weight strength: Between 0.8 to 1

Basic prompt: pd3_joy, 1girl, gloves, jewelry, green eyes, necklace, coat, pants, blue gloves, full body, arms, black hair, vest, shirt, black footwear, short hair, belt, boots,

If you want to use the mask, just add the mask keyword to the prompt and remove green eyes from it (or put eyes, green eyes to negatives field), i'd suggest to increase it's intensity to at least 1.2 weights so it looks like this

pd3_joy, 1girl, gloves, jewelry, (mask:1.2), necklace, coat, pants, blue gloves, full body, arms, black hair, vest, shirt, black footwear, short hair, belt, boots,

________________________________________________________________________________

Recommended Settings:

Sampling Method:
Euler A (40 steps), I found it to be the best sampler for this Lora.
DPM++ 2M Karras (45 steps),
DPM++ 3M SDE (63 Steps)

________________________________________________________________________________
HiRes Fix (I always use it after txt2image generation in most of my images):

Denoise values ranging from between 0.48 up to 0.51
Upscaler: 4x-UltraSharp

HiRes Steps: 45 should be fine, I use 63.
________________________________________________________________________________
CFG Scale: Tried it from 4 to 7 which seems to handle the Lora pretty well from my testing with various styles and concepts. Works on both anime/realism.

_______________________________________________________________________________

- Notes from the author:

There may be some small issues regarding the mask generation, i'd strongly suggest to use the inpainting as a workaround.

Also don't let her anywhere near PAYDAY 3 Servers if you value your life....Maintenance

I have tested this particular Lora on three Illustrous based SDXL models, they're awesome:
________________________________________________________________________________
Quintillus - Illustrated "Reality" by nottheninja
https://civitai.com/models/1131256
Hassaku XL (Illustrious) - v1.3 - Style A by Ikena
https://civitai.com/models/140272
Hesperides [Illustrious-V1.0] by bloodsplash
https://civitai.com/models/1246248
________________________________________________________________________________
Thank you for checking out this model page. Have fun generating and have a great day!