Views
No views yet
ModelSamplingDiscrete node into your workflow. Set sampling to v_prediction, zsnr to true..yaml config file into the model folder. The .yaml file must have the exact same name as the model file, only with the .yaml extension instead of .safetensors. Set Noise schedule for sampling in settings to Zero Terminal SNR.score_9). You can use any of these styles:
score_X, tag1, tag2, tag3, ...score_X, [your description here]score_X, [description], tag1, tag2, ...score_1score_1, lowres, artistic error, scan artifacts, jpeg artifacts, multiple views, too many watermarks, negative space, blank pagescore_1, score_2, score_3, lowres, artistic error, film grain, scan artifacts, jpeg artifacts, chromatic aberration, dithering, halftone, screentones, multiple views, logo, too many watermarks, negative space, blank pagemimic_scale to match the CFG scale and set both minimum scales to the same lower value. I use Half Cosine Up for both modes.1import torch
2from diffusers import DiffusionPipeline
3
4pipe = DiffusionPipeline.from_pretrained(
5 "Koolchh/AnimeBoysZeroXL",
6 torch_dtype=torch.float16,
7 use_safetensors=True,
8 variant="fp16"
9)
10pipe.to("cuda")
11
12prompt = "score_9, 1boy, male focus, shirt, solo, looking at viewer, smile, black hair, brown eyes, short hair"
13negative_prompt = "score_1"
14
15image = pipe(
16 prompt=prompt,
17 negative_prompt=negative_prompt,
18 width=1024,
19 height=1024,
20 guidance_scale=5,
21 num_inference_steps=28
22).images[0]score_X, where X is a range from 1 to 9.
score_9 represents the highest aesthetic quality based on my personal preferences.| tag | rating |
|---|---|
rating:general | general |
rating:sensitive | sensitive |
rating:questionable | questionable |
rating:explicit | explicit |
year YYYY (ranging from 2005 to 2025) to target specific era styles.