modele/Arcane Style LoRA - offset

Arcane Style LoRA - offset

Lykon

1/26/2026

5:43:57 PM

| Discussion|

Powiązane słowa kluczowe i tagi

Młoda kobieta o czerwonych oczach i średniej długości czarnych włosach, nosząca duży czarny kapelusz czarownicy z złotą lamówką, czarną pelerynę, chokera oraz czerwoną sukienkę z odkrytymi ramionami w pomieszczeniu przy ciepłym oświetleniu.

Zalecane podpowiedzi

arcane style, 1girl, bangs, looking at viewer, solo

Zalecane negatywne podpowiedzi

(painting by bad-artist-anime:0.9), (painting by bad-artist:0.9), text, error, blurry, cropped, watermark, worst quality, jpeg artifacts, low quality, normal quality, signature, username, artist name, (worst quality, low quality:1.4), bad anatomy

Zalecane parametry

samplers

DPM++ SDE Karras, Euler a

steps

20 - 28

cfg

7 - 9

resolution

512x704, 920x1376, 896x1344, 512x768

other models

AnyLoRA_bakedVae_fp16_NOTpruned (b775b13509), AnimePastelDream_Soft_noVae_fp16 (6739b459f3)

Zalecane parametry wysokiej rozdzielczości

upscaler

R-ESRGAN 0.4 denoise, R-ESRGAN 4x+, R-ESRGAN AnimeVideo, 4x_foolhardy_Remacri

upscale

1.8

steps

denoising strength

0.06 - 0.45

Wskazówki

Use Arcane Style LoRA with AnyLoRA around 1 weight for offset or 0.65 for original.

Highres fix using R-ESRGAN 0.4 denoise improves output quality.

This LoRA is style based, not character specific, but can approximate characters with accurate prompting.

In auto1111, copy the LoRA file to stable-diffusion-webui/models/lora and adjust the weight from default :1 as needed.

Sponsorzy twórcy

Making models can be expensive. Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕

Arcane Style LoRA

Making models can be expensive. Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕

Greatest show of 2021, time to bring this style to 2023 Stable Diffusion with LoRA.

This should be used with AnyLoRA (that's neutral enough) at around 1 weight for the offset version, 0.65 weight for the original one (with highres fix R-ESRGAN 0.4 denoise for better results). Trigger is arcane style but I noticed this often works even without it.

Allow me to remind this is not a character LoRA (I didn't tag characters in the dataset). Still ,as you can see, it can get pretty close to the original characters with accurate prompting. This should also work with most of my character LoRA's (used Ahri and Lucy to test).

Please enjoy.

How to use LoRA's in auto1111: