modelle/Style Squeeze [LoRA] - v2

Style Squeeze [LoRA] - v2

MachineCode

10/15/2025

10:01:32 AM

| Discussion|

Empfohlene Parameter

samplers

DPM++ 2M SDE Karras, Euler a, DPM++ 2M Karras

steps

20 - 35

cfg

4 - 9

clip skip

resolution

512x768, 512x960, 512x640

other models

darksun_v40 (48a4cb67d2), toonyouJP_alpha1 (022bfe5fa2), MCBS_v2 (faebcebd4c), icbinpICantBelieveIts_afterburn (65cd001daf), aniflatmixAnimeFlatColorStyle_v20 (b0d25db787), babes_20 (65cf944caa)

Empfohlene Hires (Hochauflösungs-) Parameter

upscaler

R-ESRGAN General 4xV3, R-ESRGAN 4x+

upscale

1.5

steps

10 - 20

denoising strength

0.4 - 0.6

Tipps

This lora helps primarily with color, composition, clothing and other things that make the image look 'cool'.

All sample images are generated with a weight of 1, but you can go a little overboard if you want.

It's not super good with generating text (or text-like shapes) at the moment.

Versions-Highlights

More images, balanced outputs

Ersteller-Sponsoren

NOTE: The SDXL version has been moved into its own listing. Click Here

UPDATE:

v3: Larger dataset, better training

v2: More varied dataset used for training

Squeeze every OUNCE of style and coolness from your prompts!

This lora helps primarily with color, composition, clothing and other things that make the image look 'cool'. All the sample images are generated with a weight of 1, but you can go a little overboard if you want.

NOTE: It's not super good with generating text (or text-like shapes) at the moment, something I hope to correct in the future.