Hipoly 3D Model LoRA - CivArchive (CivitAI Archive)

Hipoly 3D Model LoRA - v1.0

NSFW

(日本語は後半にあります)

High-poly 3D Model LoRA

This is a LoRA trained on high-polygon 3D model images.

It can provide clean, high-resolution skin and hair materials, as well as detailed clothing elements.

For those interested, I have compiled the technical insights gained during the training of ver.2 in the following article.

https://note.com/takumi__ncr/n/n21016c358ea5

Changes in 2.0

Increased the number of training images
Increased the training resolution (ver.1: 768, ver.2: 896)
Revised the tagging
- Removed the trigger word as well
Reviewed the training parameters

Improvements in 2.0

It can reproduce a more 3D-like texture and stereoscopi effect than ver.1
LoRA can be applied without a trigger word
The style can be controlled using 3d and realistic tags
- When added to Positive Prompt, it enhances the 3D feel
- When added to Negative Prompt, it adds details such as clothing while maintaining the model's art style
The issue of eroding the model's art style with tags such as intricate, detailed has been alleviated

Additional information

The effect of LoRA was confirmed using the following my custom merged models:

High-poly 3D Model LoRA

ハイポリゴン 3D モデル画像を学習した LoRA です。

3D らしいクリーンで高精細な肌・髪の質感や、衣服のディテールを与えることができます。

ver.2 の学習時に得られた技術的知見をまとめましたので、興味がある方は次の記事をご覧ください。

https://note.com/takumi__ncr/n/n2fb9d265ffa9

2.0 での変更点

教師画像を増やしました
学習解像度を向上しました (ver.1: 768, ver.2: 896)
タグ付けを見直しました
- トリガーワードも削除しました
学習パラメータを見直しました

2.0 での改善点

ver.1 よりも 3D らしいテクスチャ感、奥行き感が再現できます
トリガーワードなしで LoRA を適用することが可能です
3d, realistic タグを使用することで絵柄のコントロールが可能です
- Positive Prompt に入れた場合、より 3D 感を強化します
- Negative Prompt に入れた場合、モデルの絵柄を保ちつつ、衣装などのディテールを追加することができます
intricate, detailed などのタグによってモデルの絵柄を侵食してしまう問題が緩和されました

その他

LoRA の動作確認は以下の自作マージモデルを使用して行いました。

Description

FAQ

Comments (26)

ritcher1Feb 14, 2023· 3 reactions

CivitAI

Which models do you recommend with this great LORA ?

NeoClassicalRibbon

Author

Feb 14, 2023· 1 reaction

My favorite combination is with 7th_anime_v3_C and AbyssOrangeMix2 for an anime style or with doll774 and dosmix for a realistic style.

ritcher1Feb 14, 2023

Thank you !

semaforFeb 23, 2023

do you have the link for doll774?

mihainFeb 24, 2023

@semafor search on Hugging Face

ritcher1Feb 24, 2023

@semafor https://huggingface.co/doll774/doll774/tree/main

@NeoClassicalRibbon

This one, right ?

https://huggingface.co/doll774/doll774/resolve/main/doll774.ckpt

bbbus9901Feb 14, 2023

CivitAI

this lora is amanzing !but i can not Reprint you pic what VAE you use?

NeoClassicalRibbon

Author

Feb 15, 2023

I accidentaly included pictures that used a LoRA in an experimental stage.

If you'd like to reproduce the same image, please refer to the images which have a prompt "<lora:hiqcg_body_768_epoch-000005:0.5>".

Lute_HendeFeb 14, 2023· 6 reactions

CivitAI

"<lora:hiqcg_face-epoch-000004:0.4>"

Do you have a lora for faces too? :)

NeoClassicalRibbon

Author

Feb 15, 2023· 3 reactions

I accidentaly included pictures that used a LoRA in an experimental stage.

At first I tried to train the separate LoRAs for each face and body.

However, I finally decided to train train a single LoRA using a full body image.

I'm sorry to confuse you.

anaiFeb 15, 2023

CivitAI

Can you tell me which model to use?

Thanks！

NeoClassicalRibbon

Author

Feb 15, 2023

If you prefer a realistic style, why not try the doll774 model.

anaiFeb 15, 2023

Thank you very much！

YozoRaAruFeb 16, 2023

CivitAI

Could you please tell me where I can find the doll774 model? I tried searching on Civitai, but I couldn't find it. Thank you very much！

NeoClassicalRibbon

Author

Feb 16, 2023· 4 reactions

Here you are! https://huggingface.co/doll774/doll774/blob/main/doll774.ckpt

YozoRaAruFeb 16, 2023

@NeoClassicalRibbon Thank you !

wilson01234Feb 17, 2023

CivitAI

what is your base model?

NeoClassicalRibbon

Author

Feb 19, 2023

nai-sfw is.

iamddtlaFeb 20, 2023

CivitAI

There is <lora:hiqcg_face-epoch-000004:0.4> in your prompt word, where can I download this lora?

NeoClassicalRibbon

Author

Feb 20, 2023

I updated the model explanation.

See Notices.

iamddtlaFeb 24, 2023

Thanks.

shekelbergFeb 22, 2023

CivitAI

what else are you using to get those fantasy dress/armor designs?

bryan_beatthebushMar 17, 2023· 6 reactions

CivitAI

can you only indicate LoRA WITHOUT the trigger keyword hiqcgbody?? this is redundancy.

AkitachiApr 5, 2023

CivitAI

This Lora tends to distort the face of the charactor. I used exactly the same parameters in your sample PNG but got a weird face. Which VAE did you use? Except for the data in PNG (like seed, size. etc), is there any adjustment you used during generating the sample images?

ordinarycitizent5095Apr 7, 2023

"beautifuleyeslikeness_purefacelikeness" Lora seems to help with faces, in some cases.

AkitachiApr 20, 2023

@ordinarycitizentwelve518 Thanks. Although that lora did not give me the same image as the samples here, it gave me an image as good as the samples

LORA

SD 1.5

by NeoClassicalRibbon

Download (Beta) View on CivitAI