LUMIERE-Q
Collection
LUMIERE-Q, the first version: A high-quality anime model built on SDXL. β’ 2 items β’ Updated β’ 1
How to use BackGwa/LUMIERE-Q with Diffusers:
pip install -U diffusers transformers accelerate
import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("BackGwa/LUMIERE-Q", dtype=torch.bfloat16, device_map="cuda")
prompt = "(1girl),aris (blue archive),blue archive,solo,long hair,dim lighting,cute,looking at viewer,masterpiece,best quality,highres,absurdres,newest"
image = pipe(prompt).images[0]import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("BackGwa/LUMIERE-Q", dtype=torch.bfloat16, device_map="cuda")
prompt = "(1girl),aris (blue archive),blue archive,solo,long hair,dim lighting,cute,looking at viewer,masterpiece,best quality,highres,absurdres,newest"
image = pipe(prompt).images[0]LUMIERE is the spiritual successor to Stelra, upgraded from SD 1.5 to SDXL 1.0, making it one of the most outstanding models capable of generating high-quality anime-style images.
It has been merged with frequently used community models along with its own high-quality LoRA models through LUMIERE merging solution.
LUMIERE uses prompting techniques and methods commonly employed with the most widely used SDXL models. Please refer to the prompting techniques below.
(GENDER), CHARACTER NAME, SERIES, STYLE, SITUATION OR OTHER TAGS, QUALITY TAGS
masterpiece, high score, great score, absurdres
lowres, bad anatomy, bad hands, text, error, missing finger, extra digits, fewer digits, cropped, worst quality, low quality, low score, bad score, average score, signature, watermark, username, blurry
Sampling method : Euler a, DPM++ 3M SDE
Sampling steps : 20 - 40
CFG : 5 - 6


LUMIERE was merged and trained using the following structure.
Iβd appreciate your support and interest! coff.ee/backgwa
Base model
stabilityai/stable-diffusion-xl-base-1.0