Downloads · 30 days
161
71% of all-time downloads
sleephashira/rei-character-noobai-xl-vpred-lora
rei-character-noobai-xl-vpred-lora is a text-to-image model from sleephashira. Use it when you need an image from a text prompt. It is set up for diffusers. The card lists the license as other.
This SDXL character LoRA is an unofficial release with no affiliation with or endorsement by any authors, artists, publishers, rights holders, or upstream model authors. It is provided only for personal, research, and…
Downloads · 30 days
161
71% of all-time downloads
All-time downloads
227
Public
Repo size
462 MB
Likes
0
Public
Click a slice to open those files.
.safetensors456 MB · 99%
From the Hugging Face model README
This SDXL character LoRA is an unofficial release with no affiliation with or endorsement by any authors, artists, publishers, rights holders, or upstream model authors. It is provided only for personal, research, and educational use.

sleephashira/rei-character-noobai-xl-vpred-lorareichar_noobai_vpred.safetensors456487828 bytese7eddcd155cf34ea53c85160f8cdc2aa2e1902d087abab113965beb0915196c2Laxhar/noobai-XL-Vpred-1.0reicharThe adapter was trained on 28 synthetic images of Rei, an original project character. Checkpoint step 1800 was selected for this release.
The 28 images were generated and restyled with the copyrighted-derived
multi3_lora.safetensors adapter. That adapter was trained on 600 monochrome
Bleach panels, 600 official-color Bleach panels, and 600 JoJo panels. Those
source panels are copyrighted and were not licensed by their copyright holders
for that training. Synthetic generation does not erase this provenance or
establish rights in any inherited visual material.
NoobAI XL V-Pred requires Euler, prediction_type="v_prediction",
rescale_betas_zero_snr=True, guidance_rescale=0.7, and CFG 4-5. Other
samplers or epsilon-prediction settings do not provide the intended inference
behavior. Put reichar near the start of the prompt and describe clothing,
camera, expression, and scene explicitly. A tested starting point is 28 steps,
CFG 5, and character LoRA scale 0.7 at approximately 1024x1024 total pixel area.
from diffusers import EulerDiscreteScheduler
pipe.scheduler = EulerDiscreteScheduler.from_config(
pipe.scheduler.config,
prediction_type="v_prediction",
rescale_betas_zero_snr=True,
)
pipe.load_lora_weights(
"sleephashira/rei-character-noobai-xl-vpred-lora",
weight_name="reichar_noobai_vpred.safetensors",
)
pipe.fuse_lora(lora_scale=0.7)
image = pipe(
"reichar, 1boy, black hair, grey eyes, grey trench coat, portrait",
width=832,
height=1216,
num_inference_steps=28,
guidance_scale=5.0,
guidance_rescale=0.7,
).images[0]
Results vary by prompt, seed, resolution, and adapter strength. The small synthetic dataset can overfit poses, clothing, facial traits, or backgrounds and can produce malformed anatomy or inconsistent character details.
Rei is an original project character, and this adapter was trained on 28
synthetic images rather than directly extracted manga panels. However, those
images were generated and restyled using the multi3 adapter, which was trained
on copyrighted, unlicensed Bleach and JoJo panels. No copyright holder licensed
or endorsed that source training, synthetic dataset, or this release.
Permitted scope: personal, research, and educational use only.
Prohibited scope: commercial or monetized use, advertising, paid services, resale, and any other revenue-generating use are prohibited.
license: other identifies non-standard model terms. These terms are a custom
use restriction, not an OSI-approved open-source license. No rights are granted
to any source artwork, characters, names, trademarks, datasets, synthetic
training images, or other third-party material. The statement that Rei is an
original project character does not grant rights to the downloader and does not
remove rights or claims that may attach to the generation lineage. Downloading
the weights does not supply permission from any rights holder.
The downloader and user are solely responsible for determining legality, obtaining all necessary permissions, complying with local law, and following all applicable platform and upstream terms. The software and artifact are provided as-is, without warranties of any kind. To the maximum extent permitted by law, the maintainer excludes liability for use, outputs, claims, damages, or other consequences arising from the artifact.
This disclaimer and the custom restriction do not override applicable law or upstream licenses. All applicable upstream base-model terms also govern. Review the NoobAI XL V-Pred 1.0 model card and terms, the upstream Illustrious XL model card, its terms of use, and the linked Fair AI Public License 1.0-SD before downloading or using this adapter.