0nc8C_anima-base-v1_0

Anima model：anima-base-v1.0.safetensors

VAE：qwen_image_vae.safetensors

TE：qwen_3_06b_base.safetensors

res set:
bs8 1024² 19,041images

0nc8C_nai-anime-v2

Stable Diffusion Checkpoint：nai-anime-v2.ckpt（CRC32: B16D490B）

res set:
bs12 1024² 19,041images

0nc8C_hdm-xut-340M-1024px

Denoising Network & VAE：hdm-xut-340M-1024px-note.safetensors

TE：Qwen/Qwen3-0.6B

res set:
bs20 1024² 19,107images

0nc7P_anima-preview3

Anima model：anima-preview3-base.safetensors

VAE：qwen_image_vae.safetensors

TE：qwen_3_06b_base.safetensors

res set:
bs20 1024² 12,600images(effective 3,150)

Bias, Risks, and Limitations

This LoRA was trained entirely on a synthetic dataset. Because the dataset is synthetic, the model may sometimes generalize better to characters/concepts that were not explicitly included in the training set than to the specific characters that were included. This is not guaranteed and can vary significantly depending on the base model, prompt, sampler, CFG, resolution, and LoRA weight.

For transparency, the following is, to the best of my knowledge, the complete list of named characters/concepts explicitly included in the synthetic training dataset:

jabami_yumeko
citlali_(genshin_impact)
reze_(chainsaw_man)
hirume_of_heavenly_incense
rita_rossweisse
iono_(pokemon)
label_(nikke)
vivian_banshee
justia_(brown_dust)
xilonen_(genshin_impact)
mari_(blue_archive)
perlica_(arknights)
bowsette
princess_king_boo
sanhua_(wuthering_waves)
monika_weisswind
klaudia_valentz
nishikigi_chisato
nivora_(wuthering_waves)
sparxie_(honkai:_star_rail)
sparkle_(honkai:_star_rail)
aria_(zenless_zone_zero)
rossi_(arknights)
kanna_(blue_archive)
kanna_(swimsuit)_(blue_archive)
friedrich_der_grosse_(azur_lane)
lilith_(fate)
alexandrina_sebastiane
asuna_(blue_archive)
mash_kyrielight
mash_kyrielight_(dangerous_beast)
ako_(blue_archive)
mash_kyrielight_(dangerous_beast)_(cosplay)
tangtang_(arknights)
meika_hime
st._louis_(luxurious_wheels)_(azur_lane)
sigrika_(wuthering_waves)
cissia_(zenless_zone_zero)
denia_(wuthering_waves)
gilberta_(arknights)
angelina_(arknights)
ardelia_(arknights)
eyjafjalla_(arknights)
chisa_(wuthering_waves)
hiyuki_(wuthering_waves)
agnes_tachyon_(umamusume)
nahida_(genshin_impact)
ganyu_(genshin_impact)
silver_wolf_(honkai:_star_rail)
silver_wolf_(lv.999)_(honkai:_star_rail)
furina_(genshin_impact)
mimori_(blue_archive)
linnea_(genshin_impact)
ukinami_yuzuha
ukinami_yuzuha_(tanuki_in_broad_daylight)
keqing_(genshin_impact)
mono_monet
zhuang_fangyi_(arknights)
maruzensky_(umamusume)
maruzensky_(heavenly_maiden_of_divine_speed)_(umamusume)
vivlos_(umamusume)
promeia_(zenless_zone_zero)
fluorite_(arknights)
mint_(neverness_to_everness)
illustrious_(azur_lane)
lucilla_(wuthering_waves)
serapeum_(honkai_impact)
klukai_(girls'_frontline_2)
hk416_(girls'_frontline)
hotori_(neverness_to_everness)
escoffier_(genshin_impact)
cartethyia_(wuthering_waves)
kirara_(genshin_impact)
fleurdelys_(wuthering_waves)
lacrimosa_(neverness_to_everness)
evanescia_(honkai:_star_rail)
remielle_dan
lucy_(cyberpunk)
sister_gigant
kokuriko_(blue_archive)
nicole_(genshin_impact)
anchorage_(azur_lane)
anchorage_(sweet_evening_milk)_(azur_lane)
implacable_(azur_lane)
jane_doe_(zenless_zone_zero)
elegg_(nikke)
suisui_(wuthering_waves)
chang_wu_(azur_lane)
chang_wu_(in_full_glory_spring's_branch_blooms)_(azur_lane)
kanan_(kanan-sama)
velina_airgid
tsaritsa_(genshin_impact)
yamashiro_(azur_lane)
orphie_magnusson
cinderella_(nikke)
shinku_(neverness_to_everness)
lupa_(wuthering_waves)
kansas_(azur_lane)
kansas_(late_night_leisure)_(azur_lane)
female_rover_(wuthering_waves)
jean_(genshin_impact)
alsace_(azur_lane)
norma_hollowell
sweety_(zenless_zone_zero)
yangyang_(wuthering_waves)
yangyang_(xuanling)_(wuthering_waves)
yae_sakura
yae_sakura_(dream_raiment)
belle_(zenless_zone_zero)
nanally_(neverness_to_everness)
taihou_(azur_lane)
zhezhi_(wuthering_waves)
musashi_(azur_lane)
musashi_(the_purple_rabbit's_marksmanship)_(azur_lane)
odette_(genshin_impact)
xianyun_(genshin_impact)
marciana_(nikke)
marciana_(marine_study)_(nikke)
alice_thymefield
lynae_(wuthering_waves)
lynae_(peppermint)_(wuthering_waves)
yoimiya_(genshin_impact)
chisa_(peach_parfait)_(wuthering_waves)
sakurako_(blue_archive)
anis_(nikke)
yixuan_(zenless_zone_zero)
plana_(blue_archive)
kochou_shinobu
nicole_demara
haiqing_(astral_party)
buling_(wuthering_waves)
gouden_leeuw_(azur_lane)
seed_(zenless_zone_zero)
togawa_sakiko
elysia_(honkai_impact)
lauma_(genshin_impact)
mavuika_(genshin_impact)
hakuhou_(azur_lane)
kanroji_mitsuri
rory_mercury
lion_(azur_lane)
anby_demara_(silver_soldier)
anby_demara
nagato_(azur_lane)
lucia_elowen
galbrena_(wuthering_waves)
lumine_(genshin_impact)
signora_(genshin_impact)
shinano_(azur_lane)
dehya_(genshin_impact)
eden_(honkai_impact)
jahoda_(genshin_impact)
hatsune_miku
azuma_(azur_lane)
hanako_(blue_archive)
belfast_(azur_lane)
izuna_(blue_archive)
blanc_(nikke)
matikanetannhauser_(umamusume)
silence_suzuka_(umamusume)
wakamo_(blue_archive)
sirius_(azur_lane)
bronya_zaychik_(silverwing:_n-ex)
the_dahlia_(honkai:_star_rail)
rio_(blue_archive)
mimosa_vermillion
yukikaze_panettone
yelan_(genshin_impact)
jinhsi_(wuthering_waves)
yorktown_(azur_lane)
prinz_eugen_(azur_lane)
griseo
kikyou_(blue_archive)
hindenburg_(azur_lane)
aegir_(azur_lane)
bay_(nikke)
cipher_(honkai:_star_rail)
clorinde_(genshin_impact)
dusk_(arknights)
gentildonna_(umamusume)
hasumi_(blue_archive)
ju_fufu
lynette_(genshin_impact)
mejiro_mcqueen_(umamusume)
nagasaki_soyo
rukkhadevata_(genshin_impact)
shiroko_terror_(blue_archive)
skirk_(genshin_impact)

This list is intended to be exhaustive with respect to the named characters/concepts explicitly included in the synthetic dataset. It is not a list of every visual feature, style, composition, or abstract concept that the model may have learned, and it does not imply that the model will reproduce only these names or reproduce them accurately.

These names are provided for disclosure purposes only. They do not imply any affiliation with, endorsement by, or authorization from the owners of these characters, names, or trademarks. All rights remain with their respective owners. This LoRA is an unofficial/experimental model and is not an official product.

Synthetic data can still contain biases, artifacts, and stylistic inconsistencies inherited from the generation process. The model may underrepresent or overrepresent certain features, and it may produce unintended or inaccurate outputs. Do not use it to impersonate real people, misrepresent identities, or generate illegal or harmful content. Users are responsible for how they use the model and for complying with applicable laws, platform policies, and the rights of third parties.

License & Usage Notice

This LoRA is a derivative work of three base models. Different sections apply depending on which base model you use it with. Read the section relevant to your case.

Quick Reference

For Anima-Base-v1.0 by CircleStone Labs, commercial use of the LoRA or derivatives is not allowed; commercial use of images (Outputs) is allowed; the model is also subject to the NVIDIA Open Model License.

For NovelAI Diffusion Anime V2, commercial use of the LoRA or derivatives is not allowed because the dual license applies and the stricter NonCommercial term controls; commercial use of images is ambiguous and should be treated as restricted for safety; the key condition is the dual license of CreativeML Open RAIL-M plus CC BY-NC-SA 4.0.

For HDM-xut-340M-Anime, commercial use of the LoRA or derivatives is not allowed until development completes; commercial use of images is not clearly separated and should be treated as restricted; the key condition is CC BY-NC-SA 4.0, explicitly until development finishes.

Section 1 — Base Model: Anima-Base-v1.0 (CircleStone Labs)

This LoRA is a derivative work based on Anima-Base-v1.0 by CircleStone Labs and is governed by the following terms.

Key Points

Non-Commercial Restriction on the Model. The Anima-Base-v1.0 model and its derivatives, including this LoRA, are licensed under the CircleStone Labs Non-Commercial License v1.0. Commercial use of the model itself is strictly prohibited. You cannot host this model or its derivatives on a platform and charge for access or generation, use it behind a paid API, or integrate it directly into a paid product such as a game or software.

Commercial Use of Outputs Is Allowed. You may use images generated with this model (the “Outputs”) for commercial purposes, including art for visual novels, video games, or anime; concept art, illustrations, or assets for commercial projects; and images in advertisements. The license expressly distinguishes between the Model (which is restricted) and the Outputs (which are not).

Derivative Works and Distribution. This LoRA is a Derivative Work under the license. The license explicitly clarifies that adapters such as LoRAs are Derivative Works. CircleStone Labs has no “rugpull” clause and cannot arbitrarily terminate the license or demand takedowns of compliant derivatives. Commercial licensees, such as platforms, may be granted rights to use derivatives like this LoRA.

Upstream License Inheritance. Anima-Base-v1.0 is itself a Derivative Model of Cosmos-Predict2-2B-Text2Image and remains subject to the relevant terms of the NVIDIA Open Model License Agreement.

Legal Compliance. Users are responsible for ensuring their use complies with applicable law. The model creator and licensor are not liable for misuse.

License Links

CircleStone Labs Non-Commercial License v1.0: https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md

NVIDIA Open Model License Agreement (Upstream): https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/

Base Model: Anima-Base-v1.0 on Hugging Face (circlestone-labs/Anima-Base-v1.0-Diffusers)

Section 2 — Base Model: NovelAI Diffusion Anime V2

This section applies when this LoRA is used with NovelAI Diffusion Anime V2 (Licensor: Anlatan LLC).

License Structure: Dual License, Both Apply Simultaneously

NovelAI Diffusion Anime V2 is based on Stable Diffusion 1.5 and distributed under a dual license: CreativeML Open RAIL-M dated August 22, 2022, and CC BY-NC-SA 4.0 (Attribution-NonCommercial-ShareAlike 4.0 International).

The model card states: “This means that the terms of both licenses apply at the same time.” You must comply with both. Where they conflict, the more restrictive term controls in practice for lawful use.

Key Points for This LoRA

The LoRA itself is NonCommercial when derived from this base. CC BY-NC-SA 4.0 defines “Adapted Material” broadly to include material derived from or based on the Licensed Material and modified in a way that requires permission under copyright. A LoRA trained on or derived from NovelAI Diffusion Anime V2 weights falls within that definition. The CC BY-NC-SA 4.0 grant is limited to NonCommercial purposes only: you may reproduce and Share Adapted Material “for NonCommercial purposes only.”

CreativeML Open RAIL-M is more permissive on model commercialization, but because both licenses apply, the NonCommercial restriction in CC BY-NC-SA 4.0 governs the LoRA as Adapted Material.

As a result, you cannot use this LoRA with the NovelAI V2 base for commercial model hosting, paid API access, or integration into a commercial product.

Attribution and ShareAlike obligations apply. If you share this LoRA or any Adapted Material based on NovelAI V2, CC BY-NC-SA 4.0 Section 3 requires attribution and ShareAlike. For attribution, retain creator identification, copyright notice, license notice, warranty disclaimer, and a URI or hyperlink to the licensed material. For ShareAlike, if you share Adapted Material, the license you apply must be a Creative Commons license with the same License Elements (Attribution, NonCommercial, ShareAlike), this version or later, or a BY-NC-SA Compatible License.

CreativeML Open RAIL-M use-based restrictions also apply. The RAIL-M Attachment A Use Restrictions remain enforceable conditions. They prohibit uses such as violating applicable law; exploiting or harming minors; generating or disseminating verifiably false information with intent to harm; generating or disseminating personal identifiable information usable to harm an individual; defamation or harassment; fully automated decision-making that adversely impacts legal rights or creates or modifies a binding, enforceable obligation; discrimination against or harm to individuals or groups based on online or offline social behavior or known or predicted personal or personality characteristics; exploiting vulnerabilities of a specific group based on age, social, physical, or mental characteristics to materially distort behavior in a way that causes or is likely to cause physical or psychological harm; discrimination against individuals or groups based on legally protected characteristics or categories; providing medical advice or medical results interpretation; and generating or disseminating information for administration of justice, law enforcement, immigration, or asylum processes, such as predicting criminal or fraudulent behavior through text profiling, causal relationships between assertions, or indiscriminate and arbitrarily targeted use.

Outputs (generated images) are the main ambiguity in the dual-license structure. CreativeML Open RAIL-M treats Output as separately licensed from the model and expressly permits commercial use of generated images. Section 6 states that the licensor “claims no rights in the Output You generate using the Model,” and its guidance is generally read as allowing commercial use of generated images. CC BY-NC-SA 4.0 grants rights only for NonCommercial purposes and applies to the Licensed Material and Adapted Material — the model weights and derivatives — not necessarily to the informational content of a generated image in isolation.

Because both licenses apply at the same time, and because NovelAI’s model card does not separately carve out an unconditional commercial-output permission for V2, the conservative position is to treat commercial use of images generated with the NovelAI V2 base as restricted, or at minimum as requiring independent legal analysis. If you intend commercial output use, consult the licensor or qualified counsel before relying on the RAIL-M output clause alone.

Distribution obligations apply. If you redistribute this LoRA with NovelAI V2, CreativeML Open RAIL-M Section 4 requires you to include the RAIL-M use-based restrictions (Attachment A) as an enforceable provision in any legal agreement governing use or distribution; give recipients a copy of the RAIL-M license; mark modified files with prominent change notices; and retain copyright, patent, trademark, and attribution notices. CC BY-NC-SA 4.0 Section 3(a) also requires attribution and a link to the license, and Section 3(b) imposes ShareAlike on Adapted Material.

License Links

NovelAI Diffusion Anime V2 full license text (CreativeML Open RAIL-M + CC BY-NC-SA 4.0): available on the NovelAI Hugging Face repository

CreativeML Open RAIL-M: see the license text embedded in the NovelAI model card

CC BY-NC-SA 4.0: https://creativecommons.org/licenses/by-nc-sa/4.0/

Section 3 — Base Model: HDM-xut-340M-Anime

This section applies when this LoRA is used with HDM-xut-340M-Anime by KBlueLeaf.

License: CC BY-NC-SA 4.0 (Development-Stage Restriction)

The HDM-xut-340M-Anime repository states: “This project is still under development, therefore all the models, source code, text, documents or any media in this project are licensed under CC-BY-NC-SA 4.0 until the finish of development.”

The Hugging Face metadata lists the license as cc, and the project’s GitHub and ComfyUI pages confirm the CC-BY-NC-SA 4.0 designation during development.

Key Points

NonCommercial restriction applies to the model and derivatives. Under CC BY-NC-SA 4.0, you may not use the Licensed Material or Adapted Material for commercial purposes. A LoRA derived from or trained on HDM-xut-340M-Anime weights is Adapted Material. The LoRA, when used with this base, is NonCommercial.

This means no paid hosting of the LoRA plus HDM base combination, no paid API access using this combination, and no integration into a commercial product where the model itself is a monetized component.

Attribution is required. If you share the LoRA or any Adapted Material based on HDM-xut-340M-Anime, you must identify the creator(s) of the Licensed Material; provide a copyright notice; include a notice referring to CC BY-NC-SA 4.0 and the warranty disclaimer; provide a URI or hyperlink to the Licensed Material where reasonably practicable; and indicate if you modified the material and retain indications of previous modifications.

ShareAlike applies to Adapted Material. If you share Adapted Material, including a LoRA distributed for use with HDM-xut-340M-Anime, the license you apply must be a Creative Commons license with the same License Elements (Attribution, NonCommercial, ShareAlike), this version or later, or a BY-NC-SA Compatible License.

Outputs (generated images) are treated differently here. CC BY-NC-SA 4.0 frames the NonCommercial restriction around the Licensed Material — the model and its adaptations — not around the informational content of individual generated images as a separate category. Unlike CreativeML Open RAIL-M, it has no explicit “Output is separately licensed and may be used commercially” clause. Because the HDM project’s licensing notice does not separately address output commercialization, treat commercial use of HDM-generated images as not clearly permitted during the development phase. If commercial output use matters to you, contact the HDM maintainer for clarification.

The development-stage qualifier matters. The CC BY-NC-SA 4.0 restriction is expressly tied to the development phase. The licensor may change the license when development finishes. Check the HDM-xut-340M-Anime repository for the current license status before relying on any assumption about future terms.

License Links

HDM-xut-340M-Anime on Hugging Face: https://huggingface.co/KBlueLeaf/HDM-xut-340M-anime

HDM project on GitHub: https://github.com/KohakuBlueleaf/HDM

CC BY-NC-SA 4.0 full text: https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode

Cross-Cutting Summary for Users

You can, subject to Section 1 conditions, use this LoRA with Anima-Base-v1.0 for personal or research purposes; use images generated with the Anima-Base-v1.0 base commercially, subject to NVIDIA upstream terms and applicable law; and share this LoRA with Anima-Base-v1.0, provided you comply with Section 1’s notice and license-inclusion requirements.

You cannot use this LoRA as a model component in any commercial service or product with any of the three base models; treat the NovelAI V2 or HDM-xut-340M-Anime combinations as commercially usable for the model itself; or assume that commercial output rights for the NovelAI V2 or HDM combinations are as clearly established as they are for the Anima-Base-v1.0 plus CircleStone license combination.

You must comply with applicable law; comply with the use-based restrictions in CreativeML Open RAIL-M Attachment A when using the NovelAI V2 base; provide attribution and ShareAlike licensing when sharing derivatives based on NovelAI V2 or HDM-xut-340M-Anime; and retain license notices and include copies of the relevant licenses when redistributing.

Disclaimer: This notice summarizes licensing terms for informational purposes only and is not legal advice. The licenses themselves govern; where this summary and a license text differ, the license text controls. If you intend to use this LoRA or its outputs commercially, consult qualified legal counsel and, where appropriate, the upstream licensors directly.

Files with SHA256: b7ad554a6741bfa79ccd240616b4474614d7479de2646730cd9e4b5bc0f2b9aa