Locon trained on Arkeus art. Good for making cartoon pictures

Locon trained on Arkeus art. Good for making cartoon pictures

This is an experimental LoHA, mainly to improve eyeliner / eyeshadow makeup.

Original intention was to train an e-girl makeup concept, but ended up not enough e-girl likeness.

Overall the LoHA can improve the prompts such as "eyeshadow", "eyeliner", "eyelashes", "makeup"... See xyz plot image for comparison (last image).

I suggest paring this with Colored Lipstick - LoHa.

Images were generated using my chilloutNebula_v2.

Recommend weight: 0.6 - 0.8, please use LoRA Block Weight to minimize the effect on face shape. I have attached my presets on the second version (the zip file). Paste the content into your LoRA Block Weight extension preset box.

This is an experimental LoHA, mainly to improve eyeliner / eyeshadow makeup.

Original intention was to train an e-girl makeup concept, but ended up not enough e-girl likeness.

Overall the LoHA can improve the prompts such as "eyeshadow", "eyeliner", "eyelashes", "makeup"... See xyz plot image for comparison (last image).

I suggest paring this with Colored Lipstick - LoHa.

Images were generated using my chilloutNebula_v2.

Recommend weight: 0.6 - 0.8, please use LoRA Block Weight to minimize the effect on face shape. I have attached my presets on the second version (the zip file). Paste the content into your LoRA Block Weight extension preset box.

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi
If you're looking for a >>LoRA Making Tutorial<<

GYMTASTIC!

The Summer body is built in Winter. You have to work out.

But it's hard,isn't it?

You need the motivation, discipline, a training program, a diet plan, training equipment, probably a bunch of gym buddies too!

Don't worry, I'll help, I'll give you a LoRA to make a training waifu aswell!

What do you think? With your training waifu (or husbando, I believe this can produce good results for men aswell) you are well-equipped to start your fitness journey towards a better you!

Download this LoRA and get sweaty!

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi
If you're looking for a >>LoRA Making Tutorial<<

GYMTASTIC!

The Summer body is built in Winter. You have to work out.

But it's hard,isn't it?

You need the motivation, discipline, a training program, a diet plan, training equipment, probably a bunch of gym buddies too!

Don't worry, I'll help, I'll give you a LoRA to make a training waifu aswell!

What do you think? With your training waifu (or husbando, I believe this can produce good results for men aswell) you are well-equipped to start your fitness journey towards a better you!

Download this LoRA and get sweaty!

⚠warning

This is "LoHa"! ! ! ! Not LoRA! ! ! !

Doesn't work with sd-webui-additional-networks! ! ! !

Please read the notes displayed on Civitai! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

Oh LoRA version? it's over here

LoRA Version

👍Model

Some adjustments are required, but all fit well.

1. AOM

2. 7th layer

3. HeavenOrangeMix

4. kawaiDiffusionSD15_v4Nsight

5. linneamix

6. paleyMIX
   

7. Animelike Series (no longer exists in Civitai) [DELETED]

8. sita7taker

Trigger Word

hammann_\(azur_lane\),1girl

The words are the same as LoRA and LoHa, so I will use the same explanation!

Basically, the higher the percentage, the stronger the influence.

"solo" and "white_background" are also 100% triggers, but you should use them depending on the situation. Because "white_background" is strange even though the background is the ocean, isn't it?

QA

Q. What is version Lion_VerD?

he name of the scheduler used for training🦁

"D" was only created fourth. -> Ver.A,B,C,D

Users don't need to worry about it.

Q. Is it necessary to divide into DB, LoRA, and LoHa?

Since the number of learning is very small, it was necessary to make up for it by branching 😭

Also, LoRA and LoHa are not perfect. DB+LoRA is a good option for those who don't know "LoRA Block Weight". Of course, for advanced users, LoHa may be enough!

Q. Ret...

No Retrofit Hammann! ! It's not in the dataset!

Q. Where is the Retrofit version?

Not now 😭 I might make one someday, I might not make it 😑

⚠warning

This is "LoHa"! ! ! ! Not LoRA! ! ! !

Doesn't work with sd-webui-additional-networks! ! ! !

Please read the notes displayed on Civitai! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

Oh LoRA version? it's over here

LoRA Version

👍Model

Some adjustments are required, but all fit well.

1. AOM

2. 7th layer

3. HeavenOrangeMix

4. kawaiDiffusionSD15_v4Nsight

5. linneamix

6. paleyMIX
   

7. Animelike Series (no longer exists in Civitai) [DELETED]

8. sita7taker

Trigger Word

hammann_\(azur_lane\),1girl

The words are the same as LoRA and LoHa, so I will use the same explanation!

Basically, the higher the percentage, the stronger the influence.

"solo" and "white_background" are also 100% triggers, but you should use them depending on the situation. Because "white_background" is strange even though the background is the ocean, isn't it?

QA

Q. What is version Lion_VerD?

he name of the scheduler used for training🦁

"D" was only created fourth. -> Ver.A,B,C,D

Users don't need to worry about it.

Q. Is it necessary to divide into DB, LoRA, and LoHa?

Since the number of learning is very small, it was necessary to make up for it by branching 😭

Also, LoRA and LoHa are not perfect. DB+LoRA is a good option for those who don't know "LoRA Block Weight". Of course, for advanced users, LoHa may be enough!

Q. Ret...

No Retrofit Hammann! ! It's not in the dataset!

Q. Where is the Retrofit version?

Not now 😭 I might make one someday, I might not make it 😑

Ai Hoshino

More epoch versions for this https://huggingface.co/KoRo8888/ai_hoshino

recommended epoch: Epoch 4 → 5(final)→ 1 → 3→ 2 Recommend model: animefull-latest, anything v4/3 Does Anime Screencaps better Recommended VAEs: clearvae(best), klf8anime

tokens: ai hoshino, animescreen cap(optional but gives better results) [use image filters and enable it then raise the "colors" on it to git rid of the desaturation 2~8]

V2 soon

Ai Hoshino

More epoch versions for this https://huggingface.co/KoRo8888/ai_hoshino

recommended epoch: Epoch 4 → 5(final)→ 1 → 3→ 2 Recommend model: animefull-latest, anything v4/3 Does Anime Screencaps better Recommended VAEs: clearvae(best), klf8anime

tokens: ai hoshino, animescreen cap(optional but gives better results) [use image filters and enable it then raise the "colors" on it to git rid of the desaturation 2~8]

V2 soon

Use these keywords to activate "Kyoto Animation screencap" "Anime screencap" recommended weight: 0.5~0.7

Use these keywords to activate "Kyoto Animation screencap" "Anime screencap" recommended weight: 0.5~0.7

This is for women with their tongues out. It has no eye rolling and is biased towards eye contact or having eyes closed. It is biased towards closeups but can be made to do other things easily. It can be made to push mouths open much wider than ahegao lora or the popular awaitingtongue embedding. It does well with side front or top shots. It has a tendency for split tongues especially when upscaling, reduce the cfg scale, and reduce the weight on 'tongue out'

This is a LOHA trained with 16/8/4/2, lr 2e-5 and 0.8e-5 for22 epochs on a set of 50 or so carefully curated super-close cropped images, all with identical tags "1 girl, mouth open, tongue out, extreme closeup"

You need both activation tags to make it work but you can play with the strength. It works well with any model.

Future versions may offer an ahegao option.

This is for women with their tongues out. It has no eye rolling and is biased towards eye contact or having eyes closed. It is biased towards closeups but can be made to do other things easily. It can be made to push mouths open much wider than ahegao lora or the popular awaitingtongue embedding. It does well with side front or top shots. It has a tendency for split tongues especially when upscaling, reduce the cfg scale, and reduce the weight on 'tongue out'

This is a LOHA trained with 16/8/4/2, lr 2e-5 and 0.8e-5 for22 epochs on a set of 50 or so carefully curated super-close cropped images, all with identical tags "1 girl, mouth open, tongue out, extreme closeup"

You need both activation tags to make it work but you can play with the strength. It works well with any model.

Future versions may offer an ahegao option.

This is a LoHa LoRA! You need the extension!
Why? LoHa have virtually no effect in output art style and they do a lot better with multi-concepts.

I've attempted this multiple times, this is the best I could come with.

Trigger words for V1.10 - You should probably avoid gens with the girl too far away!:

You can try "armsnet", "thighsnet", "topnet", "thighsnet".

Its hard to use, Hires denoise have a huge impact in final output. I think 0.7 is the sweet spot but try it yourself.

Follow the steps in the Model Version notes to reproduce my images.

Sample images generated with Hassaku.

All negative embeds should be easy to find. and Prompts should be included in the sample images! Click the little icon in the bottom right to see it.

It was trained on 120 512x512 images borrowed from Unstable Diffusion discord, 20 epochs using Animefull-final-pruned for the Training Model.

Danbooru tags generated by Waifu Diffusion 1.4 Tagger and fine tuned manually.

This is a LoHa LoRA! You need the extension!
Why? LoHa have virtually no effect in output art style and they do a lot better with multi-concepts.

I've attempted this multiple times, this is the best I could come with.

Trigger words for V1.10 - You should probably avoid gens with the girl too far away!:

You can try "armsnet", "thighsnet", "topnet", "thighsnet".

Its hard to use, Hires denoise have a huge impact in final output. I think 0.7 is the sweet spot but try it yourself.

Follow the steps in the Model Version notes to reproduce my images.

Sample images generated with Hassaku.

All negative embeds should be easy to find. and Prompts should be included in the sample images! Click the little icon in the bottom right to see it.

It was trained on 120 512x512 images borrowed from Unstable Diffusion discord, 20 epochs using Animefull-final-pruned for the Training Model.

Danbooru tags generated by Waifu Diffusion 1.4 Tagger and fine tuned manually.

提前声明：有大概率导致腿部丢失、下半身融合、扭曲。存在画风污染。

Advance statement: There is a high probability of loss of legs, fusion of lower body, and distortion.Artstyle overfitting.

建议生成横图。

It is recommended to generate a horizontal graph.

姿势基本只有这个单一的腿部交叉摩擦。触发词是“tribadism”，除此之外你最好加上“2girls”。还需要加上“bottomless”或者“nude”尽量保证你的角色穿少一点(至少裤子穿少一点)。你可以参考我的例图prompt：

The pose is basically just this single leg cross rub. The trigger word is "tribadism" and you had better to add "2girls". You also need to add "bottomless" or "nude" to try to ensure that your character wears less (at least less pants). You can refer to my example prompt:

tribadism, 2girls, yuri, tongue out, saliva,blush,bottomless, clitoris, pussy juice,spread legs,nipples,breasts out, on bed

提前声明：有大概率导致腿部丢失、下半身融合、扭曲。存在画风污染。

Advance statement: There is a high probability of loss of legs, fusion of lower body, and distortion.Artstyle overfitting.

建议生成横图。

It is recommended to generate a horizontal graph.

姿势基本只有这个单一的腿部交叉摩擦。触发词是“tribadism”，除此之外你最好加上“2girls”。还需要加上“bottomless”或者“nude”尽量保证你的角色穿少一点(至少裤子穿少一点)。你可以参考我的例图prompt：

The pose is basically just this single leg cross rub. The trigger word is "tribadism" and you had better to add "2girls". You also need to add "bottomless" or "nude" to try to ensure that your character wears less (at least less pants). You can refer to my example prompt:

tribadism, 2girls, yuri, tongue out, saliva,blush,bottomless, clitoris, pussy juice,spread legs,nipples,breasts out, on bed

This is mostly trained on her images from the past few years where she somewhat changed her appearance although some older pics snuck in.

It has affinity towards topless images so use negative prompts to avoid nsfw results.

This is mostly trained on her images from the past few years where she somewhat changed her appearance although some older pics snuck in.

It has affinity towards topless images so use negative prompts to avoid nsfw results.

*Sample images are auto-flagged and hidden... Login to see them...

Ying from Paladins

*Sample images are auto-flagged and hidden... Login to see them...

Ying from Paladins

LoCon made using Kohya LoRA Dreambooth https://github.com/linaqruf/kohya-trainer method.

His outfit is a bit complex for the AI, but I did my best. I may try to update it in the future using higher quality screencaps.

He's a pretty niche character, so I mostly made this for myself. He can look cool sometimes on different models. Batches are your friend. Just walk away from the computer for a while, eventually you'll get a good one. 😂

Descriptive prompts help (blue hair, hair over eye, yellow eye, etc)

LoCon made using Kohya LoRA Dreambooth https://github.com/linaqruf/kohya-trainer method.

His outfit is a bit complex for the AI, but I did my best. I may try to update it in the future using higher quality screencaps.

He's a pretty niche character, so I mostly made this for myself. He can look cool sometimes on different models. Batches are your friend. Just walk away from the computer for a while, eventually you'll get a good one. 😂

Descriptive prompts help (blue hair, hair over eye, yellow eye, etc)

Trained on 370 white old school swimsuit pics.

Trigger word: white school swimsuit.

Samples are generated on any4.5 with LoRA Block Weight.

Trained on 370 white old school swimsuit pics.

Trigger word: white school swimsuit.

Samples are generated on any4.5 with LoRA Block Weight.

Trained on clip skip 1 on top of ACertainty

Intermediate checkpoints can be found on huggingface https://huggingface.co/alea31415/kumabear-roukin-characters

The locon is extracted from the checkpoint

The first fours images are generated by the full checkpoint, and the remaining ones by respective model (but of course you can also use locon for roukin characters. I am just too lazy to prompt it)

Why do I train the two animes together?

I feel these two animes (light novels actually) have so much similarity that I really want to make some crossovers.

Moreover there is no reason to do single anime either. I plan to add shinmai renkinjutsushi no tenpo keiei next.

Trigger Words

KumaBear

• Atla

• Cliff

• Eleanora

• Fina

• Flora

• Gentz

• Misana

• Noire

• Shia

• Shuri

• Telmina

• Yuna

Roukin8

• Adelaide

• Beatrice

• Colette

• Sabine

• YamanoMitsuha

To get everything right you may need additional trigger words for outfits and ornaments. Here are some suggestions

• If you want to get the bear costume of Yuna you may add kigurumi, bear hood, animal hood, animal costume, hand puppet etc.

• Add Red bow for Fina/Shuri/Noire

• Add twin drill for Shia

• Add double bun for Flora

• Add scrunchie for telmina

Kumakyuu and Kumayuru are not tagged, but you may get something that look right by prompting with bears, stuffed animal etc.

Interestingly I can hardly take off the hood of Yuna during the early phase of training, but it becomes possible after longer training (actually now Yuna by default does not have hood though almost all the images of her have hood on!)

Many characters are missing from the two animes. I may update the KumaBear one at the end of the season with the following characters

Kumakyuu, Kumayuru, Lurina, Farrat (king), Kitia (queen), Karin, Sanya, Helen, Ans, Mylene, Cattleya

Dataset

KumaBear 5113

• anime screenshots 5042

• fanart 37

• official art 15

• light novel illustration 19

Roukin8 2948 (screenshots only)

Regularization ~30K

Training

• First trained for 9739 steps, resumed and trained for another 20494 steps

• clip skip 1, resolution 512, batch size 8, on top of [JosephusCheung/ACertainty](https://huggingface.co/JosephusCheung/ACertainty/tree/main)

• 2.5e-6 cosine scheduler, Adam8bit, conditional dropout 0.08

Trained on clip skip 1 on top of ACertainty

Intermediate checkpoints can be found on huggingface https://huggingface.co/alea31415/kumabear-roukin-characters

The locon is extracted from the checkpoint

The first fours images are generated by the full checkpoint, and the remaining ones by respective model (but of course you can also use locon for roukin characters. I am just too lazy to prompt it)

Why do I train the two animes together?

I feel these two animes (light novels actually) have so much similarity that I really want to make some crossovers.

Moreover there is no reason to do single anime either. I plan to add shinmai renkinjutsushi no tenpo keiei next.

Trigger Words

KumaBear

• Atla

• Cliff

• Eleanora

• Fina

• Flora

• Gentz

• Misana

• Noire

• Shia

• Shuri

• Telmina

• Yuna

Roukin8

• Adelaide

• Beatrice

• Colette

• Sabine

• YamanoMitsuha

To get everything right you may need additional trigger words for outfits and ornaments. Here are some suggestions

• If you want to get the bear costume of Yuna you may add kigurumi, bear hood, animal hood, animal costume, hand puppet etc.

• Add Red bow for Fina/Shuri/Noire

• Add twin drill for Shia

• Add double bun for Flora

• Add scrunchie for telmina

Kumakyuu and Kumayuru are not tagged, but you may get something that look right by prompting with bears, stuffed animal etc.

Interestingly I can hardly take off the hood of Yuna during the early phase of training, but it becomes possible after longer training (actually now Yuna by default does not have hood though almost all the images of her have hood on!)

Many characters are missing from the two animes. I may update the KumaBear one at the end of the season with the following characters

Kumakyuu, Kumayuru, Lurina, Farrat (king), Kitia (queen), Karin, Sanya, Helen, Ans, Mylene, Cattleya

Dataset

KumaBear 5113

• anime screenshots 5042

• fanart 37

• official art 15

• light novel illustration 19

Roukin8 2948 (screenshots only)

Regularization ~30K

Training

• First trained for 9739 steps, resumed and trained for another 20494 steps

• clip skip 1, resolution 512, batch size 8, on top of [JosephusCheung/ACertainty](https://huggingface.co/JosephusCheung/ACertainty/tree/main)

• 2.5e-6 cosine scheduler, Adam8bit, conditional dropout 0.08

LoHa of Winning Ticket (umamusume). Please install extension on WebUI.

After installing extension, put LoHa file in your sd/models/lora folder and write LoRA notation in prompt to applying LoHa. <lora:loha_name:weight>

All uploaded images are generated by AOM3A1

Recommended options

• LoRA weight 0.7 ~ 0.8

• Trigger words

  • Base

    • winning ticket \(umamusume\)

  • Racing competition costume :

    • ear ornament, hairclip, bandaid on face, necklace, bandeau, bracelet, open clothes, sleeveless jacket, hooded jacket, red jacket, wristband, partially unzipped, blue skirt, pleated skirt, shorts under skirt, bike shorts, sneakers

  • Steampunk :

    • official alternate costume, steampunk, single ear cover, feather ear ornament, goggles on headwear, low twintails, short twintails, bandaid on face, collared shirt, white shirt, crop top, red bandana, shoulder cutout, frilled sleeves, puffy short sleeves, fingerless gloves, brown gloves, midriff, belt, pouch, brown skirt, frilled skirt, miniskirt, black thighhighs, thighhighs under boots, brown footwear, knee boots

  • Casual :

    • casual, beanie, red headwear, ears through headwear, ear ornament, hairclip, bandaid on face, white shirt, long sleeves, sleeves past wrists, open jacket, red jacket, plaid jacket, fanny pack, denim shorts, short shorts, white footwear, sneakers

  • Tracen summer uniform :

    • bandaid on face, hairclip, ear ornament, tracen school uniform, summer uniform, serafuku, puffy short sleeves, purple bowtie, horseshoe ornament, sailor collar, sailor shirt, purple shirt, white skirt, pleated skirt, frilled skirt, miniskirt, zettai ryouiki, white thighhighs, brown footwear, loafers

Use a model derived from or mixed with animefull model.

• Settings

  • Stable Diffusion WebUI by Automatic1111

  • DPM++ Series(SDE Karras, 2M Karras, etc.)

  • about 20 steps, CFG scale 3.5~6.5

  • I recommend kl-f8-anime2.vae

  • CLIP skip = 1

  • Use Hires.fix to get higher quality image

    • Upscaler

      • Latent, Latent (nearest-exact), Latent (bicubic antialiased) or other Latent series.

    • Denoising strength 0.50~0.65

LoRA training info

• trained on sd-scripts by kohya_ss.

  • LyCORIS by KohakuBlueleaf

  • Thank you all a lot!

• based on Animefull-pruned model

• 71 images consist of

  • 4x low quality 27 images

  • 8x medium quality 17 images

  • 16x high quality 27 images

  • => 1 epoch = 676 images

• replaced character feature tags with winning ticket \(umamusume\)

  • 1girl, red eyes, horse ears, horse girl, tail, short hair, horse tail, simple background, solo, swept bangs, animal ears, brown hair, parted bangs, black hair, bangs

  • I think these tags are the essentials of Winning Ticket or

  • unnecessary for training because of duplication.

• resolution : 768x768 with aspect ratio bucketing.

• other training settings is included in training_data.zip or metadata of LoRA file.

And, special thanks to my best friend. DH.

LoHa of Winning Ticket (umamusume). Please install extension on WebUI.

After installing extension, put LoHa file in your sd/models/lora folder and write LoRA notation in prompt to applying LoHa. <lora:loha_name:weight>

All uploaded images are generated by AOM3A1

Recommended options

• LoRA weight 0.7 ~ 0.8

• Trigger words

  • Base

    • winning ticket \(umamusume\)

  • Racing competition costume :

    • ear ornament, hairclip, bandaid on face, necklace, bandeau, bracelet, open clothes, sleeveless jacket, hooded jacket, red jacket, wristband, partially unzipped, blue skirt, pleated skirt, shorts under skirt, bike shorts, sneakers

  • Steampunk :

    • official alternate costume, steampunk, single ear cover, feather ear ornament, goggles on headwear, low twintails, short twintails, bandaid on face, collared shirt, white shirt, crop top, red bandana, shoulder cutout, frilled sleeves, puffy short sleeves, fingerless gloves, brown gloves, midriff, belt, pouch, brown skirt, frilled skirt, miniskirt, black thighhighs, thighhighs under boots, brown footwear, knee boots

  • Casual :

    • casual, beanie, red headwear, ears through headwear, ear ornament, hairclip, bandaid on face, white shirt, long sleeves, sleeves past wrists, open jacket, red jacket, plaid jacket, fanny pack, denim shorts, short shorts, white footwear, sneakers

  • Tracen summer uniform :

    • bandaid on face, hairclip, ear ornament, tracen school uniform, summer uniform, serafuku, puffy short sleeves, purple bowtie, horseshoe ornament, sailor collar, sailor shirt, purple shirt, white skirt, pleated skirt, frilled skirt, miniskirt, zettai ryouiki, white thighhighs, brown footwear, loafers

Use a model derived from or mixed with animefull model.

• Settings

  • Stable Diffusion WebUI by Automatic1111

  • DPM++ Series(SDE Karras, 2M Karras, etc.)

  • about 20 steps, CFG scale 3.5~6.5

  • I recommend kl-f8-anime2.vae

  • CLIP skip = 1

  • Use Hires.fix to get higher quality image

    • Upscaler

      • Latent, Latent (nearest-exact), Latent (bicubic antialiased) or other Latent series.

    • Denoising strength 0.50~0.65

LoRA training info

• trained on sd-scripts by kohya_ss.

  • LyCORIS by KohakuBlueleaf

  • Thank you all a lot!

• based on Animefull-pruned model

• 71 images consist of

  • 4x low quality 27 images

  • 8x medium quality 17 images

  • 16x high quality 27 images

  • => 1 epoch = 676 images

• replaced character feature tags with winning ticket \(umamusume\)

  • 1girl, red eyes, horse ears, horse girl, tail, short hair, horse tail, simple background, solo, swept bangs, animal ears, brown hair, parted bangs, black hair, bangs

  • I think these tags are the essentials of Winning Ticket or

  • unnecessary for training because of duplication.

• resolution : 768x768 with aspect ratio bucketing.

• other training settings is included in training_data.zip or metadata of LoRA file.

And, special thanks to my best friend. DH.

Rei from Paladins

Her eyes are blurry :/

Rei from Paladins

Her eyes are blurry :/

My 2nd LoCon using Kohya LoRA Dreambooth https://github.com/linaqruf/kohya-trainer method.

Borders and being stuck inside things are pretty common, due to the poor quality images I trained it with (it particularly liked putting him INSIDE a microwave when using "microwave" prompt). He also has a bad habit of disappearing completely, but I think I know what I did wrong.

I plan on fixing all this sometime in the future using higher quality images and I'll hopefully gain more knowledge on how to make it better by then. For now it's good enough for me.

Inpaint and batches are your friend. 0.6~1 weight. Euler A. Hires fix.

My 2nd LoCon using Kohya LoRA Dreambooth https://github.com/linaqruf/kohya-trainer method.

Borders and being stuck inside things are pretty common, due to the poor quality images I trained it with (it particularly liked putting him INSIDE a microwave when using "microwave" prompt). He also has a bad habit of disappearing completely, but I think I know what I did wrong.

I plan on fixing all this sometime in the future using higher quality images and I'll hopefully gain more knowledge on how to make it better by then. For now it's good enough for me.

Inpaint and batches are your friend. 0.6~1 weight. Euler A. Hires fix.

This LORA is Sherlock Shellingford from Detective Opera Milky Holmes.

The main protagonist of the series. Sherlock is the airheaded leader of Milky Holmes, who, while known to be particularly clumsy yet is always also caring and loyal, excitable and friendly. Her Toys give her the power of telekinesis.

inital tags:

"Sherlock Shellingford, pink hair, hair rings", (or long hair)

main costume:

"hat, deerstalker, capelet, detective uniform"

"hair ribbon, school uniform, necktie"

"animal ears, cat ears, apron, maid"

You may want to use 7thLayerAbyss or AOM3A.

This LORA is Sherlock Shellingford from Detective Opera Milky Holmes.

The main protagonist of the series. Sherlock is the airheaded leader of Milky Holmes, who, while known to be particularly clumsy yet is always also caring and loyal, excitable and friendly. Her Toys give her the power of telekinesis.

inital tags:

"Sherlock Shellingford, pink hair, hair rings", (or long hair)

main costume:

"hat, deerstalker, capelet, detective uniform"

"hair ribbon, school uniform, necktie"

"animal ears, cat ears, apron, maid"

You may want to use 7thLayerAbyss or AOM3A.

Maeve from Paladins

Maeve from Paladins

Lian from Paladins

Lian from Paladins

weight: 0.6-0.8

weight: 0.6-0.8

*Sample images are auto-flagged and hidden... Login to see them...

Io from Paladins

huge_tail tag generates her tail back
bow_\(weapon\) tag attempts to generate her bow (or add it to Negative to remove it)

*Sample images are auto-flagged and hidden... Login to see them...

Io from Paladins

huge_tail tag generates her tail back
bow_\(weapon\) tag attempts to generate her bow (or add it to Negative to remove it)

First time making a LoCon using kohya LoRA dreambooth https://github.com/linaqruf/kohya-trainer

Trained on AnyLoRA model

Too hard to train his cross punisher and guns. But mostly shows his likeness ¯\_(ツ)_/¯
When I figure out how to make better models, I'll try to make it better. It's not perfect by any means.

0.6~1 weight depending on the style you want. I use euler a mostly. Descriptive prompts help (black hair, black eyes, cigarette, scruff, etc)

Sometimes it wants to make the suit all white, or parts of it, so try "white suit, black collar" in negative prompts.

First time making a LoCon using kohya LoRA dreambooth https://github.com/linaqruf/kohya-trainer

Trained on AnyLoRA model

Too hard to train his cross punisher and guns. But mostly shows his likeness ¯\_(ツ)_/¯
When I figure out how to make better models, I'll try to make it better. It's not perfect by any means.

0.6~1 weight depending on the style you want. I use euler a mostly. Descriptive prompts help (black hair, black eyes, cigarette, scruff, etc)

Sometimes it wants to make the suit all white, or parts of it, so try "white suit, black collar" in negative prompts.

A Character LoCon for Kama [Avenger] (カーマ/泡馍)

All ascensions in one LoCon！

三合一，小中大杯一次满足！

夏の堕落はチャーミング！ 咲き誇れ恋の花、出会いの渚に煩悩燦々！ 『愛もていたるは恋夏なり』！

• Recommeneded prompts: Please check my sample works!

• Trigger word: brown hair,red eyes,wolf ears,wolf tail

  • bare shoulders, blue fire,body markings, bracelet, breasts, earrings, fire, flower, hair flower, hair ornament, hair ribbon, jewelry, large breasts, long hair, lotus, nail polish, revealing clothes, ribbon, star earrings, thighs, two-tone hair, bangs, navel (大杯/ 3rd ascensions)

  • bare shoulders, bikini, bracelet,breasts, cleavage, collarbone, earrings, flower, frills, green bikini, green vest, hair flower, hair ornament, hair ribbon, hood, hooded vest, hoodie, jewelry, large breasts, long hair,lotus, necklace, pendant, ribbon, bangs, navel, braid, ponytail, (中杯/ 2nd ascensions)

  • bare shoulders, bikini, blue sky, blush, breasts, collarbone, flower wreath,green bikini, hair ribbon, head wreath,innertube,, miniskirt, nail polish, navel, rainbow skirt, ribbon, short hair,small breasts, white thighhighs, bangs,thighs, flower earrings (小杯/ 1st ascensions)

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, DanMix

• Some sample figures were generated with my Vivid Impactful Style - LoRA , Hitoimim Style - LoHa and Fuzichoco (藤ちょこ) Style - LoHa.

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

A Character LoCon for Kama [Avenger] (カーマ/泡馍)

All ascensions in one LoCon！

三合一，小中大杯一次满足！

夏の堕落はチャーミング！ 咲き誇れ恋の花、出会いの渚に煩悩燦々！ 『愛もていたるは恋夏なり』！

• Recommeneded prompts: Please check my sample works!

• Trigger word: brown hair,red eyes,wolf ears,wolf tail

  • bare shoulders, blue fire,body markings, bracelet, breasts, earrings, fire, flower, hair flower, hair ornament, hair ribbon, jewelry, large breasts, long hair, lotus, nail polish, revealing clothes, ribbon, star earrings, thighs, two-tone hair, bangs, navel (大杯/ 3rd ascensions)

  • bare shoulders, bikini, bracelet,breasts, cleavage, collarbone, earrings, flower, frills, green bikini, green vest, hair flower, hair ornament, hair ribbon, hood, hooded vest, hoodie, jewelry, large breasts, long hair,lotus, necklace, pendant, ribbon, bangs, navel, braid, ponytail, (中杯/ 2nd ascensions)

  • bare shoulders, bikini, blue sky, blush, breasts, collarbone, flower wreath,green bikini, hair ribbon, head wreath,innertube,, miniskirt, nail polish, navel, rainbow skirt, ribbon, short hair,small breasts, white thighhighs, bangs,thighs, flower earrings (小杯/ 1st ascensions)

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, DanMix

• Some sample figures were generated with my Vivid Impactful Style - LoRA , Hitoimim Style - LoHa and Fuzichoco (藤ちょこ) Style - LoHa.

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

strength: 1

models tested: walnutcreamBlend_herbmixV1, animefinal-full-pruned, chilloutmix_NiPrunedFp32Fix

use Hires. fix

prompts in pics

trained: Animefull-final-pruned.ckpt

I'm not very experienced, so any tips or constructive criticism is welcomed.

if you create some cool stuff, please share it, i want to see it!!

strength: 1

models tested: walnutcreamBlend_herbmixV1, animefinal-full-pruned, chilloutmix_NiPrunedFp32Fix

use Hires. fix

prompts in pics

trained: Animefull-final-pruned.ckpt

I'm not very experienced, so any tips or constructive criticism is welcomed.

if you create some cool stuff, please share it, i want to see it!!

Trained on AOM3. This is a LoCON, so please install the appropriate extension to be able to use it.

Consider tipping me through Ko-Fi!

https://ko-fi.com/muscleenjoyer

Trained on AOM3. This is a LoCON, so please install the appropriate extension to be able to use it.

Consider tipping me through Ko-Fi!

https://ko-fi.com/muscleenjoyer

✨✨Want your personalized LoRA or illustrations? Click here✨✨

あなただけのLoRAやイラストが欲しいですか？ ここをクリック

🎨 Pixiv https://www.pixiv.net/en/users/22936156

🐦 Twitter https://twitter.com/tsumasaky

📷 Instagram http://instagram.com/tsumasaky

✨✨Want your personalized LoRA or illustrations? Click here✨✨

あなただけのLoRAやイラストが欲しいですか？ ここをクリック

🎨 Pixiv https://www.pixiv.net/en/users/22936156

🐦 Twitter https://twitter.com/tsumasaky

📷 Instagram http://instagram.com/tsumasaky

You can support me on Ko-fi ☕ if you like my work.

Husbando LoCon is the successor of my previous work Husbando LoRA.

Husbando LoCon is able to generate explicit homoerotic hentai illustrations and illustrations of sexually suggestive males.

Features

• Easy to generate good-looking masculine characters using Danbooru tags.

• Good male genitalia.

• Good solo / solo focus homoerotic content.

• Homoerotic contents depicting 2 or more people are highly erroneous. You'll need luck and careful prompting to get things right.

Weight Guide

• For general purpose: 0.5 - 0.6. You can also use a lower weight such as 0.3 if you just want to enhance the sexiness of your character.

• For explicit art: Set the text encoder weight to 1.0 and U-Net weight to 0.5 - 0.6 by using Kohya's Additional Networks extension. (You need to toggle the "Separate UNet/Text Encoder weights" checkbox.)

How to use

• Select any checkpoint that works with Danbooru tags. This LoCon is trained from AnyLoRA and this should also work well on AOM series.

• Select a VAE if the checkpoint model you're using doesn't have a baked-in VAE, or else you'll find your generated images have dull colors. You can use vae-ft-mse-840000-ema-pruned, kl-f8-anime2, or NAI VAE.

• Make sure to follow the steps in How to use a LoCon. I suggest you start by setting the weight to 0.5.

• Generate at 512 base resolution and use hires fix to improve the details. Recommended upscaler: R-ESRGAN 4x+ Anime6B; denoising strength: 0.5.

You can support me on Ko-fi ☕ if you like my work.

Husbando LoCon is the successor of my previous work Husbando LoRA.

Husbando LoCon is able to generate explicit homoerotic hentai illustrations and illustrations of sexually suggestive males.

Features

• Easy to generate good-looking masculine characters using Danbooru tags.

• Good male genitalia.

• Good solo / solo focus homoerotic content.

• Homoerotic contents depicting 2 or more people are highly erroneous. You'll need luck and careful prompting to get things right.

Weight Guide

• For general purpose: 0.5 - 0.6. You can also use a lower weight such as 0.3 if you just want to enhance the sexiness of your character.

• For explicit art: Set the text encoder weight to 1.0 and U-Net weight to 0.5 - 0.6 by using Kohya's Additional Networks extension. (You need to toggle the "Separate UNet/Text Encoder weights" checkbox.)

How to use

• Select any checkpoint that works with Danbooru tags. This LoCon is trained from AnyLoRA and this should also work well on AOM series.

• Select a VAE if the checkpoint model you're using doesn't have a baked-in VAE, or else you'll find your generated images have dull colors. You can use vae-ft-mse-840000-ema-pruned, kl-f8-anime2, or NAI VAE.

• Make sure to follow the steps in How to use a LoCon. I suggest you start by setting the weight to 0.5.

• Generate at 512 base resolution and use hires fix to improve the details. Recommended upscaler: R-ESRGAN 4x+ Anime6B; denoising strength: 0.5.

LyCORIS of Felicia Day

LyCORIS of Felicia Day

LyCORIS of Emma Stone

LyCORIS of Emma Stone

LyCORIS of Cara Gee as Drummer from Expanse

LyCORIS of Cara Gee as Drummer from Expanse

LyCORIS of Anita Sarkeesian

LyCORIS of Anita Sarkeesian

LyCORIS of Angelina Jolie

LyCORIS of Angelina Jolie

LyCORIS of Nathan Fillion

LyCORIS of Nathan Fillion

LyCORIS of Michael Jordan

LyCORIS of Michael Jordan

LyCORIS of John Barrowman

LyCORIS of John Barrowman

Portland from Azur Lane

Portland from Azur Lane

LyCORIS of Dobromir Sośnierz

LyCORIS of Dobromir Sośnierz

LyCORIS of Andrzej Duda

LyCORIS of Andrzej Duda

Some of my LoRA's stopped working on the latest Auto1111 version for some reason. Seems to be an issue with LoCon LoRA's. To fix it all you need to do is install this extension then you can use them as normal again!

Elora from Spyro the Dragon!

Use tags "elora, furry, leaf-dress"
"leaf-dress" is for her default outfit. The majority of the dataset had her in the leaf dress so it'll sometimes affect other outfits you're going for, particularly if it's some other dress. adding leaf-dress to the negatives can help with that, as well as specifying things like "sleeves". I did my best to negate this but unfortunately like 99% of the Elora art out there has the exact same outfit so there's only so much I can do about it.

Some of my LoRA's stopped working on the latest Auto1111 version for some reason. Seems to be an issue with LoCon LoRA's. To fix it all you need to do is install this extension then you can use them as normal again!

Elora from Spyro the Dragon!

Use tags "elora, furry, leaf-dress"
"leaf-dress" is for her default outfit. The majority of the dataset had her in the leaf dress so it'll sometimes affect other outfits you're going for, particularly if it's some other dress. adding leaf-dress to the negatives can help with that, as well as specifying things like "sleeves". I did my best to negate this but unfortunately like 99% of the Elora art out there has the exact same outfit so there's only so much I can do about it.

Note #1: This is V2 of my previous Change-A-Character: Evilify Your Waifu Today! I had to make a different submission as this is a LOCON now and I couldn't have both on the same sheet.

Note #2: This is a Locon. Install the extension to get it to work. I've found them to do better when working on ideas that I want easily added to one of my LORA characters. Also, they're WAY less damaging to images, even at high weights.

I've uploaded V2 of this submission, which should allow for greater variability in terms of use, not overwhelming your normal pictures, and be usable in more places outside of anythingV3. It was trained on animefull_final with 72 images with the usual self-pruning and tagging and a wide range of different images. I like keeping my prompts simple for the first few pictures so that you can see it without a whole bunch of mumbo jumbo. Make your pictures better than mine though!

What does it do you ask? A Locon that will create a corrupted and evilify'd Waifu from whatever you give it.

How to use it: Pretty simple for this one. Just the Locon activation and the trigger word. Then add any other spice you want.

CorruptA - Main trigger word.

Otherwise, add whatever you want it to lean towards. Glowing eyes, skirts, you name it.

Extra, Extra!: Should be well trained and usable in most models. Images shown for example.

Picture Descriptions:

1-2: AnythingV3 using just this Lora with simpler prompts.

3: AnythingV3 combo'd with my released Seraphina (Disgaea 5) Lora.

4: AnythingV3 combo'd with my released Rebecca (Edens Zero) Lora.

5: AnythingV3 combo'd with my released Amiri (Iruma-kun) Lora.

6: AnythingV3 combo'd with my released Ghislaine (Mushoku Tensei) Lora.

7: AnythingV3 combo'd with my released Brainwashing Locon.

8: AnythingV3 combo'd with my released Tainted Nun Locon.

9: AbyssOrange with just this Lora.

10: PerfectWorld using just this Lora.

11: ChilloutMix using just this Lora.

What's next for this? Hopefully this one is better, if not I'll update it again.

What's next that isn't this?

More character Loras. (Ivy from Soul Calibur 5, Heine from Combatants Will Be Dispatched, Conqueror Lilias from E7, Rozalin from Disgaea 2, Tharja from Fire Emblem, Catherine from Catherine Full Body, Miku from Idaten Deities, Kuribayashi Shino from GATE, Rin Sohma from Fruits Basket, Kasumi from P5, Ann from P5, Ikumi Mito from Food Wars, Alice Nakiri from Food Wars, Chika Fujiwara from Kaguya-sama, Lulu from FF10, Momo from My Hero, Vivy from Vivy, Lucoa and Tohru from Dragon Maid, Dominique de Sade from Vanitas, and Dos from One Punch Man.) And maybe Wilhelm van Astrea, my favorite Re:Zero husbando or a Prinny from Disgaea.

V2's of some of the previous Loras. Specifically make them more diversely acceptable with other models and LORA's.

I also have some more quite NSFW Locon's. I decided to list them as well, so if people have preferences on which to fix up first let me know. Specifically, Symbiote Taking Over (imagine a venom-like symbiote taking over your waifu), being stepped on, Robotification, Petrified (I'm unsure if I should bother with this one anymore), Infection, Dominatrix (waifu becoming a dominatrix), Brainwashing Machine, Tentacles, Pillory, Masked Evil (A mask that possess' the waifu), Latex Corruption, Combat Drone (hard to describe but basically making them a mindless sci-fi combat warrior), JAV Supervillain, Jiangshi (Transforms into a Jiangshi), Mind Flayer Defeat, Harem Outfits, Hourglass Figure (just to make your waifu curvy), and Holding a Leash.

And, uh, I want to make a few "normal" Locon's so I have an "knight-ify" your character one coming up. And a forest fairy one as well. And a officer one. And a barbarian one.

Additionally, I love the style of Defaultz_17, so I'm making a Locon to mirror that.

Or just give me some random suggestion or priority opinions. That works too.

Note #1: This is V2 of my previous Change-A-Character: Evilify Your Waifu Today! I had to make a different submission as this is a LOCON now and I couldn't have both on the same sheet.

Note #2: This is a Locon. Install the extension to get it to work. I've found them to do better when working on ideas that I want easily added to one of my LORA characters. Also, they're WAY less damaging to images, even at high weights.

I've uploaded V2 of this submission, which should allow for greater variability in terms of use, not overwhelming your normal pictures, and be usable in more places outside of anythingV3. It was trained on animefull_final with 72 images with the usual self-pruning and tagging and a wide range of different images. I like keeping my prompts simple for the first few pictures so that you can see it without a whole bunch of mumbo jumbo. Make your pictures better than mine though!

What does it do you ask? A Locon that will create a corrupted and evilify'd Waifu from whatever you give it.

How to use it: Pretty simple for this one. Just the Locon activation and the trigger word. Then add any other spice you want.

CorruptA - Main trigger word.

Otherwise, add whatever you want it to lean towards. Glowing eyes, skirts, you name it.

Extra, Extra!: Should be well trained and usable in most models. Images shown for example.

Picture Descriptions:

1-2: AnythingV3 using just this Lora with simpler prompts.

3: AnythingV3 combo'd with my released Seraphina (Disgaea 5) Lora.

4: AnythingV3 combo'd with my released Rebecca (Edens Zero) Lora.

5: AnythingV3 combo'd with my released Amiri (Iruma-kun) Lora.

6: AnythingV3 combo'd with my released Ghislaine (Mushoku Tensei) Lora.

7: AnythingV3 combo'd with my released Brainwashing Locon.

8: AnythingV3 combo'd with my released Tainted Nun Locon.

9: AbyssOrange with just this Lora.

10: PerfectWorld using just this Lora.

11: ChilloutMix using just this Lora.

What's next for this? Hopefully this one is better, if not I'll update it again.

What's next that isn't this?

More character Loras. (Ivy from Soul Calibur 5, Heine from Combatants Will Be Dispatched, Conqueror Lilias from E7, Rozalin from Disgaea 2, Tharja from Fire Emblem, Catherine from Catherine Full Body, Miku from Idaten Deities, Kuribayashi Shino from GATE, Rin Sohma from Fruits Basket, Kasumi from P5, Ann from P5, Ikumi Mito from Food Wars, Alice Nakiri from Food Wars, Chika Fujiwara from Kaguya-sama, Lulu from FF10, Momo from My Hero, Vivy from Vivy, Lucoa and Tohru from Dragon Maid, Dominique de Sade from Vanitas, and Dos from One Punch Man.) And maybe Wilhelm van Astrea, my favorite Re:Zero husbando or a Prinny from Disgaea.

V2's of some of the previous Loras. Specifically make them more diversely acceptable with other models and LORA's.

I also have some more quite NSFW Locon's. I decided to list them as well, so if people have preferences on which to fix up first let me know. Specifically, Symbiote Taking Over (imagine a venom-like symbiote taking over your waifu), being stepped on, Robotification, Petrified (I'm unsure if I should bother with this one anymore), Infection, Dominatrix (waifu becoming a dominatrix), Brainwashing Machine, Tentacles, Pillory, Masked Evil (A mask that possess' the waifu), Latex Corruption, Combat Drone (hard to describe but basically making them a mindless sci-fi combat warrior), JAV Supervillain, Jiangshi (Transforms into a Jiangshi), Mind Flayer Defeat, Harem Outfits, Hourglass Figure (just to make your waifu curvy), and Holding a Leash.

And, uh, I want to make a few "normal" Locon's so I have an "knight-ify" your character one coming up. And a forest fairy one as well. And a officer one. And a barbarian one.

Additionally, I love the style of Defaultz_17, so I'm making a Locon to mirror that.

Or just give me some random suggestion or priority opinions. That works too.

由于训练集画风整体过于“浓厚”，会产生画风污染，有时候阴茎会产生畸形之类的。但是这个lora只是为了有趣，所以我觉得这没什么关系。

Because the overall style of the training set is too "thick", it will cause pollution of the style of painting, and sometimes the penis will be deformed.But this lora is just for fun, so I don't think it matters.

尽管不加应该也行，但是我还是建议加上“1boy,penis”，所以我把他们放在触发词里了。你也可以在负面里加入“multiple boys, multiple penises”避免左右两根阴茎融合在一起。

Although it should work without it, I still recommend adding "1boy, penis", so I put them in the trigger word. You can also add "multiple boys, multiple penises" to the negative to avoid merging the left and right penises.

weight 1 maybe ok

hiresfix

由于训练集画风整体过于“浓厚”，会产生画风污染，有时候阴茎会产生畸形之类的。但是这个lora只是为了有趣，所以我觉得这没什么关系。

Because the overall style of the training set is too "thick", it will cause pollution of the style of painting, and sometimes the penis will be deformed.But this lora is just for fun, so I don't think it matters.

尽管不加应该也行，但是我还是建议加上“1boy,penis”，所以我把他们放在触发词里了。你也可以在负面里加入“multiple boys, multiple penises”避免左右两根阴茎融合在一起。

Although it should work without it, I still recommend adding "1boy, penis", so I put them in the trigger word. You can also add "multiple boys, multiple penises" to the negative to avoid merging the left and right penises.

weight 1 maybe ok

hiresfix

Now you can support me on ko-fi: https://ko-fi.com/truefalse

Trained 10 epochs on 35 images of this car, i think that you can get really good results even with the 10th epoch, also you might want to use highres fix .5 it gives even better results!

image grid: https://files.catbox.moe/ovd2tm.png

Now you can support me on ko-fi: https://ko-fi.com/truefalse

Trained 10 epochs on 35 images of this car, i think that you can get really good results even with the 10th epoch, also you might want to use highres fix .5 it gives even better results!

image grid: https://files.catbox.moe/ovd2tm.png

2023/4/12 update

Upload Lycoris version (v5.3). Remove hyper-realistic images in dataset that may cause bad results. Try to make face more natural and beautiful.

2023/3/29 update

Due to the instability of Lycoris caused by sensitivity to learning rate settings, the updated model (5.2 version) is still conventional LoRA model. Use images published by some users as dataset to train the model. Fix the compatibility problem of non-NAI-based checkpoints. Now arbitrary anime model with NAI's VAE or kl-f8-anime2 VAE can also generate good results using this LoRA, theoretically. This version is a bit overfitted that will be fixed next time.

2023/3/24 Experimental Update

New model is trained using Lycoris method (all versions before 5.1 version are conventional LoRA model), pls install this extension to use this new model in webui. Adjust trigger tag to 'fashi-g'. If some problem occur during generating pls use old version.

2023/3/8 update

Update dataset and model. Decide to upload 5-epoch version whose color saturation is a bit better than higher-epoch version after careful consideration. Adjust rank to 128 to keep perfect face. The trigger tag is unchanged. The preview images are generated using latest version of YesMix and EasyNegative embedding.

2023/3/3 update

Reduce loss to ~0.11 using Lion optimizer. Btw, if you use Hires. fix, 0.55~0.6 weights of Denoising should be enough, otherwise the face may be slightly deformed. Btw, the preview images are generated using latest version of YesMix and EasyNegative embedding.

2023/2/28 update

Update dataset and model. Add trigger tag 'fashi-girl' to improve generating probability. Adjust rank to 32 to balance file size and quality.

2023/2/24 update

Prune tags to make the description more accurate.

2023/2/23 update

Merge previous LORA model that is easy to generate nice images to slightly improve color saturation and optimize face. Adjust rank to 32 to balance file size and quality. The preview images of new version are generated using latest version of YesMix.

2023/2/21 update

Adjust dataset to fix bad philtrum. Reduce training epochs to fix deformed limbs. The preview images of new version are generated using latest version of YesMix.

2023/2/20 update

Update and balance dataset to remain the images of mature female that fit personal aesthetics. New version of LORA model is trained using Lion optimizer discovered by Google Brain that is purportedly better than Adam(w). The preview images of new version are generated using latest version of YesMix.

2023/2/18 update

Update dataset and model. The preview images of new version are generated using latest version of YesMix.

2023/2/16 update

Clean the dataset to remove the images of unattractive characters. Use offset noise to improve ability to generate lighter/darker images. The preview images are generated using latest version of YesMix.

2023/2/14 update

Dataset is updated, so model is updated. Reduce the model file size to 1MB. The generating quality is unchanged.

Introduction

This is an experimental LORA model trained with about 100 images of fashionable girls that fit personal aesthetics. Therefore, this model is named as "Fashion Girl". Due to the small-scale dataset that are composed of realistic/photorealistic images, some output images will remain anime style. The 784mb VAEs (NAI, Orangemix, Anything, Counterfeit) are recommended. 0.6~0.8 weights should be enough. This model will be continuously updated as the dataset is updated.

2023/4/12 update

Upload Lycoris version (v5.3). Remove hyper-realistic images in dataset that may cause bad results. Try to make face more natural and beautiful.

2023/3/29 update

Due to the instability of Lycoris caused by sensitivity to learning rate settings, the updated model (5.2 version) is still conventional LoRA model. Use images published by some users as dataset to train the model. Fix the compatibility problem of non-NAI-based checkpoints. Now arbitrary anime model with NAI's VAE or kl-f8-anime2 VAE can also generate good results using this LoRA, theoretically. This version is a bit overfitted that will be fixed next time.

2023/3/24 Experimental Update

New model is trained using Lycoris method (all versions before 5.1 version are conventional LoRA model), pls install this extension to use this new model in webui. Adjust trigger tag to 'fashi-g'. If some problem occur during generating pls use old version.

2023/3/8 update

Update dataset and model. Decide to upload 5-epoch version whose color saturation is a bit better than higher-epoch version after careful consideration. Adjust rank to 128 to keep perfect face. The trigger tag is unchanged. The preview images are generated using latest version of YesMix and EasyNegative embedding.

2023/3/3 update

Reduce loss to ~0.11 using Lion optimizer. Btw, if you use Hires. fix, 0.55~0.6 weights of Denoising should be enough, otherwise the face may be slightly deformed. Btw, the preview images are generated using latest version of YesMix and EasyNegative embedding.

2023/2/28 update

Update dataset and model. Add trigger tag 'fashi-girl' to improve generating probability. Adjust rank to 32 to balance file size and quality.

2023/2/24 update

Prune tags to make the description more accurate.

2023/2/23 update

Merge previous LORA model that is easy to generate nice images to slightly improve color saturation and optimize face. Adjust rank to 32 to balance file size and quality. The preview images of new version are generated using latest version of YesMix.

2023/2/21 update

Adjust dataset to fix bad philtrum. Reduce training epochs to fix deformed limbs. The preview images of new version are generated using latest version of YesMix.

2023/2/20 update

Update and balance dataset to remain the images of mature female that fit personal aesthetics. New version of LORA model is trained using Lion optimizer discovered by Google Brain that is purportedly better than Adam(w). The preview images of new version are generated using latest version of YesMix.

2023/2/18 update

Update dataset and model. The preview images of new version are generated using latest version of YesMix.

2023/2/16 update

Clean the dataset to remove the images of unattractive characters. Use offset noise to improve ability to generate lighter/darker images. The preview images are generated using latest version of YesMix.

2023/2/14 update

Dataset is updated, so model is updated. Reduce the model file size to 1MB. The generating quality is unchanged.

Introduction

This is an experimental LORA model trained with about 100 images of fashionable girls that fit personal aesthetics. Therefore, this model is named as "Fashion Girl". Due to the small-scale dataset that are composed of realistic/photorealistic images, some output images will remain anime style. The 784mb VAEs (NAI, Orangemix, Anything, Counterfeit) are recommended. 0.6~0.8 weights should be enough. This model will be continuously updated as the dataset is updated.

I used cp decomposition to reduce the model size.

You may need to update your locon extension to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

I used cp decomposition to reduce the model size.

You may need to update your locon extension to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

Evie from Paladins

(Some good sample images are hidden... Login to see them...)

Evie from Paladins

(Some good sample images are hidden... Login to see them...)

I used cp decomposition to reduce the model size.

You may need to update your locon extension to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

I used cp decomposition to reduce the model size.

You may need to update your locon extension to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

I trained this LyCORIS by collecting about 200 images generated by NijiJourney V5 on the internet.

It's okay to set the weight to 1 (current version)

Set weights to 0.7~0.8 (previous version)

I generated the example images with AOM3.

I trained this LyCORIS by collecting about 200 images generated by NijiJourney V5 on the internet.

It's okay to set the weight to 1 (current version)

Set weights to 0.7~0.8 (previous version)

I generated the example images with AOM3.

I check Twitter every day, so I will respond when I receive a DM.
ツイッターは毎日チェックしているのでDMが来たら対応致します。

I check Twitter every day, so I will respond when I receive a DM.
ツイッターは毎日チェックしているのでDMが来たら対応致します。

First time using loha training characters

If there are missing features, you can try adding prompt words: black hair, blue fingerless gloves, black pantyhose, white boots,

Please check the prompt words in the example carefully.

第一次尝试loha训练角色

如果出现丢失特征，可以尝试增加提示词：black hair, blue fingerless gloves, black pantyhose, white boots,

请仔细查看例图的提示词

• 欢迎加入AI绘画模型博物馆QQ群

• ①群：791222249

• ②群：599962550

First time using loha training characters

If there are missing features, you can try adding prompt words: black hair, blue fingerless gloves, black pantyhose, white boots,

Please check the prompt words in the example carefully.

第一次尝试loha训练角色

如果出现丢失特征，可以尝试增加提示词：black hair, blue fingerless gloves, black pantyhose, white boots,

请仔细查看例图的提示词

• 欢迎加入AI绘画模型博物馆QQ群

• ①群：791222249

• ②群：599962550

This is a LoHa LoRA! You need the extension!
Why? LoHa have virtually no effect in output art style and they do a lot better with multi-concepts.

Tired of cute anime faces? Me too. Lets fix that!

Trigger words for V3.20 (old v3.20 gone! It was replaced by this hotfix):

positive: <lora:matureFemalev320:0.8>(mature female)

negative (you might have to do "shiny" and "shiny skin" in negatives): loli, cgi

You can also try "3d", "realistic", "lips apart", "small breasts", "thick thighs" and "wide hips" also helps out with more mature look!

I suggest not prompting for "1girl" at all, just go with "mature female" cus it was trained without "1girl" in dataset.

Follow the steps in the Model Version notes to reproduce my images.

Sample images generated with Hassaku.

All negative embeds should be easy to find. and Prompts should be included in the sample images! Click the little icon in the bottom right to see it.

It was trained on 207 512x512 images borrowed from Unstable Diffusion discord, 20 epochs using Animefull-final-pruned for the Training Model.

Danbooru tags generated by Waifu Diffusion 1.4 Tagger and fine tuned manually.

This is a LoHa LoRA! You need the extension!
Why? LoHa have virtually no effect in output art style and they do a lot better with multi-concepts.

Tired of cute anime faces? Me too. Lets fix that!

Trigger words for V3.20 (old v3.20 gone! It was replaced by this hotfix):

positive: <lora:matureFemalev320:0.8>(mature female)

negative (you might have to do "shiny" and "shiny skin" in negatives): loli, cgi

You can also try "3d", "realistic", "lips apart", "small breasts", "thick thighs" and "wide hips" also helps out with more mature look!

I suggest not prompting for "1girl" at all, just go with "mature female" cus it was trained without "1girl" in dataset.

Follow the steps in the Model Version notes to reproduce my images.

Sample images generated with Hassaku.

All negative embeds should be easy to find. and Prompts should be included in the sample images! Click the little icon in the bottom right to see it.

It was trained on 207 512x512 images borrowed from Unstable Diffusion discord, 20 epochs using Animefull-final-pruned for the Training Model.

Danbooru tags generated by Waifu Diffusion 1.4 Tagger and fine tuned manually.

If you create any good images using this LyCORIS, please consider sharing them by leaving a review. Thank you!

• This LyCORIS can help generate stereotypical fantasy elves, including classes such as warrior, ranger, druid, and priestess.

• Using Elf to trigger may give you a "high elf" or "hentai elf" vibe. "Pointy ear" is not a trigger word but can still work.

• You can generate classes without using "Elf". It will still give you pointy ears, but reduce the "high/hentai elf" vibe. Alternatively, you can use them to generate non-elf characters.

• Prompt "Fantasy" is not necessary but can add some flavor, and you can try other themes, such as cyberpunk.

If you create any good images using this LyCORIS, please consider sharing them by leaving a review. Thank you!

• This LyCORIS can help generate stereotypical fantasy elves, including classes such as warrior, ranger, druid, and priestess.

• Using Elf to trigger may give you a "high elf" or "hentai elf" vibe. "Pointy ear" is not a trigger word but can still work.

• You can generate classes without using "Elf". It will still give you pointy ears, but reduce the "high/hentai elf" vibe. Alternatively, you can use them to generate non-elf characters.

• Prompt "Fantasy" is not necessary but can add some flavor, and you can try other themes, such as cyberpunk.

Trained on screenshots.

Trained on screenshots.

Some of my LoRA's stopped working on the latest Auto1111 version for some reason. Seems to be an issue with LoCon LoRA's. To fix it all you need to do is install this extension then you can use them as normal again!

Falco from Star Fox!

Use tags "falco, falclothes, winged-arms"
Falclothes is for Falco's full default outfit.

I made this one as a request, hope you enjoy!

Some of my LoRA's stopped working on the latest Auto1111 version for some reason. Seems to be an issue with LoCon LoRA's. To fix it all you need to do is install this extension then you can use them as normal again!

Falco from Star Fox!

Use tags "falco, falclothes, winged-arms"
Falclothes is for Falco's full default outfit.

I made this one as a request, hope you enjoy!

We will comply with the ANYCOLOR secondary creation guidelines.
https://www.anycolor.co.jp/guidelines/

It seems that no particular intention is indicated. I pledge to create and publish works that respect the spirit of ANYCOLOR content.

特に個別の意向が示されていないと思われます。ANYCOLORコンテンツの精神を尊重した作品を制作、公表していく事を誓います。

We will comply with the ANYCOLOR secondary creation guidelines.
https://www.anycolor.co.jp/guidelines/

It seems that no particular intention is indicated. I pledge to create and publish works that respect the spirit of ANYCOLOR content.

特に個別の意向が示されていないと思われます。ANYCOLORコンテンツの精神を尊重した作品を制作、公表していく事を誓います。

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

3 - Prompts

I use wildcards to generate all of my images.

Positive:

RAW, (masterpiece, best quality, photorealistic, absurdres, 8k:1.2), nsfw, best lighting, complex pupils, detailed background, __composition-time__, __composition-location__,

1girl, SML, solo, upper body, (close up:0.4), __pose-head__, cowboy shot, standing, __pose-hand__, __face-expression__, shiny skin, visible skin pores, __hairlength__, curly hair, __hair-color__ hair, highly detailed skin,

__cloth-sexy__, chocker, see-through, <lora:LoHA-SML:1:LyCOFACE>

Negative:

badhandv4, missing fingers, extra digit, fewer digits, deformityv6, easynegative, verybadimagenegative_v1.2-6400, (low quality, worst quality, blurry:1.2), bad anatomy, bad proportions, bad-artist, character sheet, multiple views, jpeg artifacts, watermark, username, signature, text, pubic hair, cropped, monochrome, dof, backlighting, sketch, painting, censored, low key, high contrast, ribs, aged up, wrinkled skin, wire, skin spot, freckles, low key, high contrast,

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

3 - Prompts

I use wildcards to generate all of my images.

Positive:

RAW, (masterpiece, best quality, photorealistic, absurdres, 8k:1.2), nsfw, best lighting, complex pupils, detailed background, __composition-time__, __composition-location__,

1girl, SML, solo, upper body, (close up:0.4), __pose-head__, cowboy shot, standing, __pose-hand__, __face-expression__, shiny skin, visible skin pores, __hairlength__, curly hair, __hair-color__ hair, highly detailed skin,

__cloth-sexy__, chocker, see-through, <lora:LoHA-SML:1:LyCOFACE>

Negative:

badhandv4, missing fingers, extra digit, fewer digits, deformityv6, easynegative, verybadimagenegative_v1.2-6400, (low quality, worst quality, blurry:1.2), bad anatomy, bad proportions, bad-artist, character sheet, multiple views, jpeg artifacts, watermark, username, signature, text, pubic hair, cropped, monochrome, dof, backlighting, sketch, painting, censored, low key, high contrast, ribs, aged up, wrinkled skin, wire, skin spot, freckles, low key, high contrast,

We will comply with the ANYCOLOR secondary creation guidelines.
https://www.anycolor.co.jp/guidelines/

It seems that no particular intention is indicated. I pledge to create and publish works that respect the spirit of ANYCOLOR content.

特に個別の意向が示されていないと思われます。ANYCOLORコンテンツの精神を尊重した作品を制作、公表していく事を誓います。

We will comply with the ANYCOLOR secondary creation guidelines.
https://www.anycolor.co.jp/guidelines/

It seems that no particular intention is indicated. I pledge to create and publish works that respect the spirit of ANYCOLOR content.

特に個別の意向が示されていないと思われます。ANYCOLORコンテンツの精神を尊重した作品を制作、公表していく事を誓います。

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

3 - Prompts

I use wildcards to generate all of my images.

Positive:

RAW, (masterpiece, best quality, photorealistic, absurdres, 8k:1.2), nsfw, best lighting, complex pupils, detailed background, __composition-time__, __composition-location__,

1girl, ririz, solo, upper body, (close up:0.4), __pose-head__, cowboy shot, standing, __pose-hand__, __face-expression__, shiny skin, visible skin pores, __hairlength__, curly hair, __hair-color__ hair, highly detailed skin,

__cloth-sexy__, chocker, see-through,

Negative:

badhandv4, missing fingers, extra digit, fewer digits, deformityv6, easynegative, verybadimagenegative_v1.2-6400, (low quality, worst quality, blurry:1.2), bad anatomy, bad proportions, bad-artist, character sheet, multiple views, jpeg artifacts, watermark, username, signature, text, pubic hair, cropped, monochrome, dof, backlighting, sketch, painting, censored, low key, high contrast, ribs, aged up, wrinkled skin, wire, skin spot, freckles, low key, high contrast,

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

3 - Prompts

I use wildcards to generate all of my images.

Positive:

RAW, (masterpiece, best quality, photorealistic, absurdres, 8k:1.2), nsfw, best lighting, complex pupils, detailed background, __composition-time__, __composition-location__,

1girl, ririz, solo, upper body, (close up:0.4), __pose-head__, cowboy shot, standing, __pose-hand__, __face-expression__, shiny skin, visible skin pores, __hairlength__, curly hair, __hair-color__ hair, highly detailed skin,

__cloth-sexy__, chocker, see-through,

Negative:

badhandv4, missing fingers, extra digit, fewer digits, deformityv6, easynegative, verybadimagenegative_v1.2-6400, (low quality, worst quality, blurry:1.2), bad anatomy, bad proportions, bad-artist, character sheet, multiple views, jpeg artifacts, watermark, username, signature, text, pubic hair, cropped, monochrome, dof, backlighting, sketch, painting, censored, low key, high contrast, ribs, aged up, wrinkled skin, wire, skin spot, freckles, low key, high contrast,

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

3 - Prompts

I use wildcards to generate all of my images.

Positive:

RAW, (masterpiece, best quality, photorealistic, absurdres, 8k:1.2), nsfw, best lighting, complex pupils, detailed background, __composition-time__, __composition-location__,

1girl, ririz, solo, upper body, (close up:0.4), __pose-head__, cowboy shot, standing, __pose-hand__, __face-expression__, shiny skin, visible skin pores, __hairlength__, curly hair, __hair-color__ hair, highly detailed skin,

__cloth-sexy__, chocker, see-through,

Negative:

badhandv4, missing fingers, extra digit, fewer digits, deformityv6, easynegative, verybadimagenegative_v1.2-6400, (low quality, worst quality, blurry:1.2), bad anatomy, bad proportions, bad-artist, character sheet, multiple views, jpeg artifacts, watermark, username, signature, text, pubic hair, cropped, monochrome, dof, backlighting, sketch, painting, censored, low key, high contrast, ribs, aged up, wrinkled skin, wire, skin spot, freckles, low key, high contrast,

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

3 - Prompts

I use wildcards to generate all of my images.

Positive:

RAW, (masterpiece, best quality, photorealistic, absurdres, 8k:1.2), nsfw, best lighting, complex pupils, detailed background, __composition-time__, __composition-location__,

1girl, ririz, solo, upper body, (close up:0.4), __pose-head__, cowboy shot, standing, __pose-hand__, __face-expression__, shiny skin, visible skin pores, __hairlength__, curly hair, __hair-color__ hair, highly detailed skin,

__cloth-sexy__, chocker, see-through,

Negative:

badhandv4, missing fingers, extra digit, fewer digits, deformityv6, easynegative, verybadimagenegative_v1.2-6400, (low quality, worst quality, blurry:1.2), bad anatomy, bad proportions, bad-artist, character sheet, multiple views, jpeg artifacts, watermark, username, signature, text, pubic hair, cropped, monochrome, dof, backlighting, sketch, painting, censored, low key, high contrast, ribs, aged up, wrinkled skin, wire, skin spot, freckles, low key, high contrast,

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi - commissions open!
If you're looking for a >>LoRA Making Tutorial<<

Hard Swimwear

This one is for those who don't know if they want to have a swim or to have sex.

Why not both? Get this LoCon now!

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi - commissions open!
If you're looking for a >>LoRA Making Tutorial<<

Hard Swimwear

This one is for those who don't know if they want to have a swim or to have sex.

Why not both? Get this LoCon now!

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

LyCOKEEPBG:1,1,1,1,1,1,1,1,1,0.2,0.4,0.8,1,1,0.8,0.4,0.2,0.2,0,0,0.8,1,1,1,0,0

LyCOPALETTE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

LoHa of Gold Ship (umamusume). Please install extension on WebUI.

After installing extension, put LoHa file in your sd/models/lora folder and write LoRA notation in prompt to applying LoHa. <lora:loha_name:weight>

All uploaded images are generated by AOM3A1

Recommended options

• LoRA weight around 0.7

• Trigger words

  • Base

    • gold ship \(umamusume\)

  • Racing competition costume :

    • pillbox hat, brown headwear, red bowtie, bare shoulders, red dress, sleeveless dress, armband, white gloves, buttons, pouch, white pantyhose, thigh strap, white footwear, knee boots, high heel boots

  • Casual :

    • casual, pillbox hat, brown headwear, red jacket, open jacket, open clothes, ribbed shirt, white shirt, long sleeves, jeans, blue pants

  • Summer :

    • official alternate costume, sunglasses, necklace, jewelry, black one-piece swimsuit, casual one-piece swimsuit, cleavage, bare shoulders, arm tattoo, bracelet, thigh strap, black footwear, sandals

    • (optional) water gun

  • Tracen summer uniform :

    • tracen school uniform, summer uniform, pillbox hat, brown headwear, serafuku, puffy short sleeves, bowtie, horseshoe ornament, sailor collar, sailor shirt, purple shirt, white skirt, frills, miniskirt, zettai ryouiki, white thighhighs, brown footwear, shoes

Use a model derived from or mixed with animefull model.

• Settings

  • Stable Diffusion WebUI by Automatic1111

  • DPM++ Series(SDE Karras, 2M Karras, etc.)

  • about 20 steps, CFG scale 3.5~6.5

  • I recommend kl-f8-anime2.vae

  • CLIP skip = 1

  • Use Hires.fix to get higher quality image

    • Upscaler

      • Latent, Latent (nearest-exact), Latent (bicubic antialiased) or other Latent series.

    • Denoising strength 0.50~0.65

LoRA training info

• trained on sd-scripts by kohya_ss.

  • LyCORIS by KohakuBlueleaf

  • Thank you all a lot!

• based on Animefull-pruned model

• 255 images consist of

  • 4x low quality 80 images

  • 8x medium quality 100 images

  • 16x high quality 75 images

  • => 1 epoch = 2320 images

• replaced character feature tags with gold ship \(umamusume\)

• resolution : 768x768 with aspect ratio bucketing.

• other training settings is included in training_data.zip or metadata of LoRA file.

And, special thanks to my best friend. DH.

LoHa of Gold Ship (umamusume). Please install extension on WebUI.

After installing extension, put LoHa file in your sd/models/lora folder and write LoRA notation in prompt to applying LoHa. <lora:loha_name:weight>

All uploaded images are generated by AOM3A1

Recommended options

• LoRA weight around 0.7

• Trigger words

  • Base

    • gold ship \(umamusume\)

  • Racing competition costume :

    • pillbox hat, brown headwear, red bowtie, bare shoulders, red dress, sleeveless dress, armband, white gloves, buttons, pouch, white pantyhose, thigh strap, white footwear, knee boots, high heel boots

  • Casual :

    • casual, pillbox hat, brown headwear, red jacket, open jacket, open clothes, ribbed shirt, white shirt, long sleeves, jeans, blue pants

  • Summer :

    • official alternate costume, sunglasses, necklace, jewelry, black one-piece swimsuit, casual one-piece swimsuit, cleavage, bare shoulders, arm tattoo, bracelet, thigh strap, black footwear, sandals

    • (optional) water gun

  • Tracen summer uniform :

    • tracen school uniform, summer uniform, pillbox hat, brown headwear, serafuku, puffy short sleeves, bowtie, horseshoe ornament, sailor collar, sailor shirt, purple shirt, white skirt, frills, miniskirt, zettai ryouiki, white thighhighs, brown footwear, shoes

Use a model derived from or mixed with animefull model.

• Settings

  • Stable Diffusion WebUI by Automatic1111

  • DPM++ Series(SDE Karras, 2M Karras, etc.)

  • about 20 steps, CFG scale 3.5~6.5

  • I recommend kl-f8-anime2.vae

  • CLIP skip = 1

  • Use Hires.fix to get higher quality image

    • Upscaler

      • Latent, Latent (nearest-exact), Latent (bicubic antialiased) or other Latent series.

    • Denoising strength 0.50~0.65

LoRA training info

• trained on sd-scripts by kohya_ss.

  • LyCORIS by KohakuBlueleaf

  • Thank you all a lot!

• based on Animefull-pruned model

• 255 images consist of

  • 4x low quality 80 images

  • 8x medium quality 100 images

  • 16x high quality 75 images

  • => 1 epoch = 2320 images

• replaced character feature tags with gold ship \(umamusume\)

• resolution : 768x768 with aspect ratio bucketing.

• other training settings is included in training_data.zip or metadata of LoRA file.

And, special thanks to my best friend. DH.

Update：

V4 is MUCH better and easier to use, trained on anyv3, test on anyv5. The method below do still work but not necessary any more.

Compared to Bluemix (the 4g ckpt model), this LoHA is more capable with character LoRA, it will tend to be keeping character's facial characteristic and outfit, only change the line and coloring. (like a color filter rather than totally different art style).

If you like the character be fully switch to BA art style, generating OC or don't care about the LoRA character's outfit, I suggest you to use BlueMix instead.

Suggestion:

• use nature language to generate base environment (e.g. A girl is standing alongside the river.)

• weight 0.8-1, (higher value for better result but worse anatomy, vice versa.)

• base resolution: w or h lower than 768x768, over will cause the image broken more often.

• base model: suggest anyv3/v5, not recommend any oil painting/pastel model like AOM series or counterfeit. It CAN use, but is only a color filter.

• environment prompts (still work!):

  • Blue Archive-ish theme (Millennium?): tree, cloud, road, city, day, blue sky, outdoors

  • Japanese theme (Hyakkiyako?): cloud, day, outdoors, cherry blossoms,

  • Nightsky: night, night sky, starry sky,

  • You can try other tags generated by taggers of Blue Archive CGs.

===

I'm learning how to make a LoHA/artstyle model. This is my first step result, my PC isn't very good for testing so I release it for more feedbacks from your side, so this is basically an Alpha test, the model do work but not very user-friendly.

If you want to get a good result, you should know how to use lora block weight.

If you don't know how to use it but still have interest on learning some new tricks, then read the following guide.

My goal is make a artstyle LoHA that is capable to most of chara LoRAs.

So I trained it on Animefull-pruned, works perfectly on that one and Anything V3, will work on AnythingV4.5 without character LoRA, but almost won't work on Counterfeit or AOM3. It will more likely a color filter than an art style on those models.

Maybe I should train another version based on AnythingV4.5. If you have any suggestion about art style training, let me know. I will be very appreciate of that.

It can work without tags, but sometimes will make weird color.

Here are some tag groups for more stable results:

• Blue Archive-ish theme (Millennium?): tree, cloud, road, city, day, blue sky, outdoors

• Japanese theme (Hyakkiyako?): pink theme, cloud, day, outdoors, cherry blossoms,

• Nightsky: blue theme, night, night sky, starry sky, cityscape,

• You can try other tags generated by taggers of Blue Archive CGs.

You 'd better use it with LoRA block weight if you want to input a character lora. You can get it as webui plugin. Otherwise you should lower this model's weight to below 0.6, which will work bit poorly but still work, or it will make the image color very awful.

↓If you get something like this, lower the artstyle model weight, or use the method below.

When you use it with LoRA block weight, there are 2 methods:

• A: By default, you should remove your charactor lora's artstyle by this:

  <YourCharaModelName:1:1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0>

  don't forget to add :ALLto this model or it may won't be working.

  e.g. <BArstylev1c:1:ALL>

• B: If you noticed that your character's appearance has been obviously affected by this model, keep only the color of the artstyle model instead like this:

  <BArstylev1c:1:1,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1>

  In this situation, you can keep Charactor model in full.

  e.g. <YourCharaModelName:1:ALL>

• Basically, method A will works more like the original artstyle than method B.

• It seems that LoHA model have more block weights, I don't have much time to test since my PC will easily crash on it, if you have some information about it, please tell me. But so far using LoRA block weight is fine.

• ↓Still, if you get everything right but still got something like this, try to use the second way instead.

Method A when the charactor is affected.

Method B

Update：

V4 is MUCH better and easier to use, trained on anyv3, test on anyv5. The method below do still work but not necessary any more.

Compared to Bluemix (the 4g ckpt model), this LoHA is more capable with character LoRA, it will tend to be keeping character's facial characteristic and outfit, only change the line and coloring. (like a color filter rather than totally different art style).

If you like the character be fully switch to BA art style, generating OC or don't care about the LoRA character's outfit, I suggest you to use BlueMix instead.

Suggestion:

• use nature language to generate base environment (e.g. A girl is standing alongside the river.)

• weight 0.8-1, (higher value for better result but worse anatomy, vice versa.)

• base resolution: w or h lower than 768x768, over will cause the image broken more often.

• base model: suggest anyv3/v5, not recommend any oil painting/pastel model like AOM series or counterfeit. It CAN use, but is only a color filter.

• environment prompts (still work!):

  • Blue Archive-ish theme (Millennium?): tree, cloud, road, city, day, blue sky, outdoors

  • Japanese theme (Hyakkiyako?): cloud, day, outdoors, cherry blossoms,

  • Nightsky: night, night sky, starry sky,

  • You can try other tags generated by taggers of Blue Archive CGs.

===

I'm learning how to make a LoHA/artstyle model. This is my first step result, my PC isn't very good for testing so I release it for more feedbacks from your side, so this is basically an Alpha test, the model do work but not very user-friendly.

If you want to get a good result, you should know how to use lora block weight.

If you don't know how to use it but still have interest on learning some new tricks, then read the following guide.

My goal is make a artstyle LoHA that is capable to most of chara LoRAs.

So I trained it on Animefull-pruned, works perfectly on that one and Anything V3, will work on AnythingV4.5 without character LoRA, but almost won't work on Counterfeit or AOM3. It will more likely a color filter than an art style on those models.

Maybe I should train another version based on AnythingV4.5. If you have any suggestion about art style training, let me know. I will be very appreciate of that.

It can work without tags, but sometimes will make weird color.

Here are some tag groups for more stable results:

• Blue Archive-ish theme (Millennium?): tree, cloud, road, city, day, blue sky, outdoors

• Japanese theme (Hyakkiyako?): pink theme, cloud, day, outdoors, cherry blossoms,

• Nightsky: blue theme, night, night sky, starry sky, cityscape,

• You can try other tags generated by taggers of Blue Archive CGs.

You 'd better use it with LoRA block weight if you want to input a character lora. You can get it as webui plugin. Otherwise you should lower this model's weight to below 0.6, which will work bit poorly but still work, or it will make the image color very awful.

↓If you get something like this, lower the artstyle model weight, or use the method below.

When you use it with LoRA block weight, there are 2 methods:

• A: By default, you should remove your charactor lora's artstyle by this:

  <YourCharaModelName:1:1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0>

  don't forget to add :ALLto this model or it may won't be working.

  e.g. <BArstylev1c:1:ALL>

• B: If you noticed that your character's appearance has been obviously affected by this model, keep only the color of the artstyle model instead like this:

  <BArstylev1c:1:1,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1>

  In this situation, you can keep Charactor model in full.

  e.g. <YourCharaModelName:1:ALL>

• Basically, method A will works more like the original artstyle than method B.

• It seems that LoHA model have more block weights, I don't have much time to test since my PC will easily crash on it, if you have some information about it, please tell me. But so far using LoRA block weight is fine.

• ↓Still, if you get everything right but still got something like this, try to use the second way instead.

Method A when the charactor is affected.

Method B

Neo Universe from Uma Musume

勝負服:

victory outfit (v4) / shobofuku (v1 ~ v3)

トレセン学園:

tracen_school_uniform
tracen_track_uniform
tracen_swimsuit

Neo Universe from Uma Musume

勝負服:

victory outfit (v4) / shobofuku (v1 ~ v3)

トレセン学園:

tracen_school_uniform
tracen_track_uniform
tracen_swimsuit

A LoCon for Holo the Wisewolf (賢狼ホロ)

Holo (ホロ) ：ぬしよ、约束じゃそ。よいな！

Holo is one of my favorite anime characters.I believe I have done a great job in training this LoCon model. However, I have been considering whether to release this LoCon model, as it cannot generate images with the same art style as the Spice & Wolf manga or novel illustrations.

How should we define the character of Holo? By its art style? By its characteristics? By its outfit? Or by all of them? I am confused. If I only consider the characteristics of Holo, I would rate this model as 100/100. Vice versa, since I did not bake in any art style in order to keep the compatibility of this LoCon model, if I want the same art style as the Spice & Wolf manga or novel illustrations, I would rate it 10/100.

I find myself confused by this. While the model does an excellent job of capturing Holo's characteristics, I did not include any specific art style in order to keep the model generalizable.

Holo是我最喜欢的动漫角色之一。我认为我在训练这个 LoCon 模型方面做得很好。我一直在考虑是否发布这个 LoCon 模型，因为它无法生成与《狼与香辛料》漫画或小说插图相同艺术风格的图像。

我们应该如何定义 Holo 这个角色？是根据它的艺术风格？还是根据它的特点？或者是它的服装？还是以上全部？我很困惑。如果只考虑 Holo 的特点，我会将这个模型评为 100/100。反之亦然，因为为了保持这个 LoCon 模型的泛化性，我没有将任何艺术风格固化在其中，如果我们希望与《狼与香辛料》漫画或小说插图相同的艺术风格，我会将它评为 10/100。

我对此感到困惑。虽然这个模型能很好地捕捉 Holo 的特点，但为了使模型具有普适性，我没有包含任何特定的艺术风格。

• Recommeneded prompts: Please check my sample works!

• Trigger word: brown hair,red eyes,wolf ears,wolf tail

  • skirt,purple/blue shirt,vest,pouch

  • skirt,purple/blue shirt,vest,cape,pouch

  • purple shirt,vest,cape,pants,pouch

  • white dress,pouch

  • white shirt,skirt (All of these trigger words are for different outfits)

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, DanMix

• Some sample figures were generated with my Vivid Impactful Style - LoRA , Hitoimim Style - LoHa and Fuzichoco (藤ちょこ) Style - LoHa.

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

A LoCon for Holo the Wisewolf (賢狼ホロ)

Holo (ホロ) ：ぬしよ、约束じゃそ。よいな！

Holo is one of my favorite anime characters.I believe I have done a great job in training this LoCon model. However, I have been considering whether to release this LoCon model, as it cannot generate images with the same art style as the Spice & Wolf manga or novel illustrations.

How should we define the character of Holo? By its art style? By its characteristics? By its outfit? Or by all of them? I am confused. If I only consider the characteristics of Holo, I would rate this model as 100/100. Vice versa, since I did not bake in any art style in order to keep the compatibility of this LoCon model, if I want the same art style as the Spice & Wolf manga or novel illustrations, I would rate it 10/100.

I find myself confused by this. While the model does an excellent job of capturing Holo's characteristics, I did not include any specific art style in order to keep the model generalizable.

Holo是我最喜欢的动漫角色之一。我认为我在训练这个 LoCon 模型方面做得很好。我一直在考虑是否发布这个 LoCon 模型，因为它无法生成与《狼与香辛料》漫画或小说插图相同艺术风格的图像。

我们应该如何定义 Holo 这个角色？是根据它的艺术风格？还是根据它的特点？或者是它的服装？还是以上全部？我很困惑。如果只考虑 Holo 的特点，我会将这个模型评为 100/100。反之亦然，因为为了保持这个 LoCon 模型的泛化性，我没有将任何艺术风格固化在其中，如果我们希望与《狼与香辛料》漫画或小说插图相同的艺术风格，我会将它评为 10/100。

我对此感到困惑。虽然这个模型能很好地捕捉 Holo 的特点，但为了使模型具有普适性，我没有包含任何特定的艺术风格。

• Recommeneded prompts: Please check my sample works!

• Trigger word: brown hair,red eyes,wolf ears,wolf tail

  • skirt,purple/blue shirt,vest,pouch

  • skirt,purple/blue shirt,vest,cape,pouch

  • purple shirt,vest,cape,pants,pouch

  • white dress,pouch

  • white shirt,skirt (All of these trigger words are for different outfits)

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, DanMix

• Some sample figures were generated with my Vivid Impactful Style - LoRA , Hitoimim Style - LoHa and Fuzichoco (藤ちょこ) Style - LoHa.

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

We comply with Vshojo secondary creation guidelines.
https://www.vshojo.com/ja/legal
Please read the guidelines.
ガイドラインを一読するようお願いいたします。

We comply with Vshojo secondary creation guidelines.
https://www.vshojo.com/ja/legal
Please read the guidelines.
ガイドラインを一読するようお願いいたします。

This model is trained on a dataset of about ~50 or so images of anime/illustrated guitars, both electric and acoustic. The goal is to allow the user to make guitars reliably without having to do a lot of work. I've had a lot of issues with generating guitars on my own and hope this helps anyone interested as well!

The model was trained on Kenshi due to its versatility, but the images posted were generated on several different models, including AbyssOrangeMix and MeinaMix. It is a LoHa, so make sure you have the Lycoris extension before using. Trigger word is guitar, but specifying "electric guitar" or "acoustic guitar" along with it will help it differentiate the two.

Special thanks to IronCatMan for helping test this lora and providing some really great pics! Check out his awesome stuff!

This model is trained on a dataset of about ~50 or so images of anime/illustrated guitars, both electric and acoustic. The goal is to allow the user to make guitars reliably without having to do a lot of work. I've had a lot of issues with generating guitars on my own and hope this helps anyone interested as well!

The model was trained on Kenshi due to its versatility, but the images posted were generated on several different models, including AbyssOrangeMix and MeinaMix. It is a LoHa, so make sure you have the Lycoris extension before using. Trigger word is guitar, but specifying "electric guitar" or "acoustic guitar" along with it will help it differentiate the two.

Special thanks to IronCatMan for helping test this lora and providing some really great pics! Check out his awesome stuff!

Fuzichoco (藤ちょこ) style

極彩色の魔術師

An Artstyle LoHa

Fuzichoco is an artist I greatly admire. Fuzichoco's artwork is renowned for its exquisite lines, vibrant and colorful tones, and intricate and elaborate character and scene designs. Although my model cannot fully replicate Fuzichoco's style, attempting to imitate it still produces some surprising results.

I have tried over 10 different training versions, and finally, I have obtained a model that I am relatively satisfied with. Now, I am sharing it with everyone here.

Fuzichoco 是我非常喜欢的一位画师。fuzichoco老师的作画以精美的线条，明艳绚丽的色彩和精致繁复的人物场景设计闻名。我的模型并不能百分百还原老师的风格，但是只是这些模仿也能达到一些让人意想不到的效果。

我尝试了10多个版本的训练，终于我得到了一个我相对满意的模型。现在我就在这里把它分享给大家。

• Recommeneded prompts: japanese clothes,kimono,hair ornament,bow,white hair, animal ears (Please check my sample work for more details)

• Trigger word: You don't need any trigger word for this LoHa

• Recommeneded Weight: 0.7-1 (You can adjust it by your perference)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, Dan mix

• How to use LoHa/LyCORIS? Please read instruction of a1111-sd-webui-locon.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

Fuzichoco (藤ちょこ) style

極彩色の魔術師

An Artstyle LoHa

Fuzichoco is an artist I greatly admire. Fuzichoco's artwork is renowned for its exquisite lines, vibrant and colorful tones, and intricate and elaborate character and scene designs. Although my model cannot fully replicate Fuzichoco's style, attempting to imitate it still produces some surprising results.

I have tried over 10 different training versions, and finally, I have obtained a model that I am relatively satisfied with. Now, I am sharing it with everyone here.

Fuzichoco 是我非常喜欢的一位画师。fuzichoco老师的作画以精美的线条，明艳绚丽的色彩和精致繁复的人物场景设计闻名。我的模型并不能百分百还原老师的风格，但是只是这些模仿也能达到一些让人意想不到的效果。

我尝试了10多个版本的训练，终于我得到了一个我相对满意的模型。现在我就在这里把它分享给大家。

• Recommeneded prompts: japanese clothes,kimono,hair ornament,bow,white hair, animal ears (Please check my sample work for more details)

• Trigger word: You don't need any trigger word for this LoHa

• Recommeneded Weight: 0.7-1 (You can adjust it by your perference)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, Dan mix

• How to use LoHa/LyCORIS? Please read instruction of a1111-sd-webui-locon.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

WARNING: THIS IS AN EXTREMELY UNSTABLE FILE and the outputs have given us rabies and have sent us to hell. We'll be retraining this and giving you the NEXT version ASAP. For some reason training on open journey DOESNT work.

Trained on Open Journey.

Better outputs coming soon.

WARNING: THIS IS AN EXTREMELY UNSTABLE FILE and the outputs have given us rabies and have sent us to hell. We'll be retraining this and giving you the NEXT version ASAP. For some reason training on open journey DOESNT work.

Trained on Open Journey.

Better outputs coming soon.

Custom tags are above,

Trained on Anime Pastel Dream.

Models it looks great on: ANY of them - but Babes is really good :3

Custom tags are above,

Trained on Anime Pastel Dream.

Models it looks great on: ANY of them - but Babes is really good :3

Apex Legends says: TRANS RIGHTS

Ok so ALL OF THE NEW LYCORIS i'm uploading right now are BASELINE only using the kohya SS outputs

If you LIKE WHAT YOU SEE please download, review and adore it - cause we don't make money off our AI and we do it for the love of it.

IF YOU USE THIS ON A GENERATION SERVICE PLEASE CREDIT AND/OR REQUEST, I mean it's not in the commercial use guidelines, but it'd be nice to get credit LOL. I spent three hours or more sorting this one!

Here's your tags in case I forget!

Apex Legends says: TRANS RIGHTS

Ok so ALL OF THE NEW LYCORIS i'm uploading right now are BASELINE only using the kohya SS outputs

If you LIKE WHAT YOU SEE please download, review and adore it - cause we don't make money off our AI and we do it for the love of it.

IF YOU USE THIS ON A GENERATION SERVICE PLEASE CREDIT AND/OR REQUEST, I mean it's not in the commercial use guidelines, but it'd be nice to get credit LOL. I spent three hours or more sorting this one!

Here's your tags in case I forget!

Ball Joint Doll Model, will have BETTER outputs after I literally get a chance to test it on stream LOL, was only uploading this because poses aren't uploading but lycoris files are!

If you like more of this go see what else we got!

Ball Joint Doll Model, will have BETTER outputs after I literally get a chance to test it on stream LOL, was only uploading this because poses aren't uploading but lycoris files are!

If you like more of this go see what else we got!

It's an another experimental LoHa model, trained from 110 various sources images,mainly flowers. The tests found that flowers&bouquet performed better,which could add floral embellishments to clothes and add hair volume and lift.

You can use the LoHa to design clothes & face.

Weight 0.9-1.0 will perform better

It's an another experimental LoHa model, trained from 110 various sources images,mainly flowers. The tests found that flowers&bouquet performed better,which could add floral embellishments to clothes and add hair volume and lift.

You can use the LoHa to design clothes & face.

Weight 0.9-1.0 will perform better

I check Twitter every day, so I will respond when I receive a DM.
ツイッターは毎日チェックしているのでDMが来たら対応致します。

I check Twitter every day, so I will respond when I receive a DM.
ツイッターは毎日チェックしているのでDMが来たら対応致します。

Girl in the Flask (瓶中少女）

An Utility/Artstyle LoCon

You can use this Locon model to bottle up the characters you want and take them with you.

你可以使用这个locon模型可以让你把你想要的角色装进瓶子里带走。

What's new about v2m version?
I have added a new trigger word to this version, and it responds better to the prompt. I have also improved the quality of the tags, so now there will be fewer watermarks on the output images. Additionally, the generation of bottles is now more stable.

However, there still seem to be some areas that need improvement. If possible, I will release a new version soon.

V2m版本有什么新功能？

我已经在这个版本中添加了新的触发词，并且它对提示的响应更好了。我还改善了标签的质量，因此现在输出图像上的水印更少了。生成瓶子的稳定性也得到了提高。

然而，它似乎还有一些需要改进的地方。如果可能的话，我会很快发布一个新版本。

• Recommeneded prompts: simple background, full body, air bubble, water,submerged,minigirl (Please check my sample work for more details)

• Trigger word: bottle,jar,gib\(concept\)

• You can use this LoCon with other characters' LoRA.

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, Dan mix

• This LoCon may have some small issues, I will fix it soon.

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

Comparison of two versions (两版本比较):

1girl, solo,bottle, simple background, arms behind back,looking at viewer, kimono, full body,minigirl,gib\(concept\),submerged, underwater, floating hair, red hair, <lora:bottle_locon_v2m:0.7>
Negative prompt: verybadimagenegative_v1.3, EasyNegative,watermark,
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 5.5, Seed: 2265992698, Size: 512x1024, Model hash: c870e984cd, Clip skip: 2, ENSD: 31337

Saved: 00184-2265992698-1girl, solo,bottle, simple background, arms behind back,looking at viewer, kimono, full body,minigirl,gib_(concept_),submerged,.png

Human kind can not gain anything without first giving something in return. To obtain something of equal value must be lost. - Fullmetal Alchemist

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

Girl in the Flask (瓶中少女）

An Utility/Artstyle LoCon

You can use this Locon model to bottle up the characters you want and take them with you.

你可以使用这个locon模型可以让你把你想要的角色装进瓶子里带走。

What's new about v2m version?
I have added a new trigger word to this version, and it responds better to the prompt. I have also improved the quality of the tags, so now there will be fewer watermarks on the output images. Additionally, the generation of bottles is now more stable.

However, there still seem to be some areas that need improvement. If possible, I will release a new version soon.

V2m版本有什么新功能？

我已经在这个版本中添加了新的触发词，并且它对提示的响应更好了。我还改善了标签的质量，因此现在输出图像上的水印更少了。生成瓶子的稳定性也得到了提高。

然而，它似乎还有一些需要改进的地方。如果可能的话，我会很快发布一个新版本。

• Recommeneded prompts: simple background, full body, air bubble, water,submerged,minigirl (Please check my sample work for more details)

• Trigger word: bottle,jar,gib\(concept\)

• You can use this LoCon with other characters' LoRA.

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, Dan mix

• This LoCon may have some small issues, I will fix it soon.

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

Comparison of two versions (两版本比较):

1girl, solo,bottle, simple background, arms behind back,looking at viewer, kimono, full body,minigirl,gib\(concept\),submerged, underwater, floating hair, red hair, <lora:bottle_locon_v2m:0.7>
Negative prompt: verybadimagenegative_v1.3, EasyNegative,watermark,
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 5.5, Seed: 2265992698, Size: 512x1024, Model hash: c870e984cd, Clip skip: 2, ENSD: 31337

Saved: 00184-2265992698-1girl, solo,bottle, simple background, arms behind back,looking at viewer, kimono, full body,minigirl,gib_(concept_),submerged,.png

Human kind can not gain anything without first giving something in return. To obtain something of equal value must be lost. - Fullmetal Alchemist

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

I used cp decomposition to reduce the model size.

You may need to update your locon extension to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

I used cp decomposition to reduce the model size.

You may need to update your locon extension to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

首先我要承认这个lora不是那么容易使用，因为训练集确实比较难以搜集。生成一张令人满意的图片可能需要做不少尝试和筛选，目前最大的问题是经常会丢失脚部，但是我还是先把它发布出来再说。

First of all, I have to admit that this lora is not so easy to use, because the training set is really difficult to collect. It may take a lot of trial and error to produce a satisfactory image, the biggest problem so far is that the feet are often missing, but I will post it first.

由于一些原因，这个lora(locon)与我以往制作的不同。它并没有触发词，而是作为一个概念强化。你需要去使用prompt描述你想要的，如果你不知道怎么做，那么请看下列示例：

This lora(locon) is different from the ones I have made in the past. It does not have a trigger word, but acts as a conceptual reinforcement. You need to use the prompt to describe what you want, if you don't know how to do it, then please see the following example:

sandwiched, double penetration, group sex, 1girl, 2boys, mmf threesome, hetero, ass, veiny penis, blush, completely nude, vaginal, anal, leg up

在此基础上你可以做出改动，比如删去"leg up"改为“standing,standing sex”(尽管这比较困难，也许你需要提高权重)、更换人物构成为“3girls,futanari,futa with female”(futa会让画面更不稳定，你可以使用mmf的图像的controlnet来辅助)，不要忘了删除相反的prompt。

On this basis, you can make changes, such as deleting "leg up" and changing it to "standing, standing sex" (although this is more difficult, maybe you need to increase the weight), changing the character composition to "3girls, futanari, futa with female "(futa will make the picture more unstable, you can use the controlnet of the mmf image to assist),don't forget to delete the opposite prompt.

hiresfix是必要的，权重我一直使用1

hiresfix is necessary, weight I always use 1.

如果你发现使用人物lora后难以控制，我建议使用controlnet来进行辅助，首先生成一个不带人物lora的图片，然后将它放入controlnet中，最后加入人物lora后再次生成，以下为示例：

If you find it difficult to control after using the character lora, I suggest using controlnet for assistance. First generate a picture without character lora, then put it into the controlnet, and finally add the character lora and generate it again. The following is an example:

首先我要承认这个lora不是那么容易使用，因为训练集确实比较难以搜集。生成一张令人满意的图片可能需要做不少尝试和筛选，目前最大的问题是经常会丢失脚部，但是我还是先把它发布出来再说。

First of all, I have to admit that this lora is not so easy to use, because the training set is really difficult to collect. It may take a lot of trial and error to produce a satisfactory image, the biggest problem so far is that the feet are often missing, but I will post it first.

由于一些原因，这个lora(locon)与我以往制作的不同。它并没有触发词，而是作为一个概念强化。你需要去使用prompt描述你想要的，如果你不知道怎么做，那么请看下列示例：

This lora(locon) is different from the ones I have made in the past. It does not have a trigger word, but acts as a conceptual reinforcement. You need to use the prompt to describe what you want, if you don't know how to do it, then please see the following example:

sandwiched, double penetration, group sex, 1girl, 2boys, mmf threesome, hetero, ass, veiny penis, blush, completely nude, vaginal, anal, leg up

在此基础上你可以做出改动，比如删去"leg up"改为“standing,standing sex”(尽管这比较困难，也许你需要提高权重)、更换人物构成为“3girls,futanari,futa with female”(futa会让画面更不稳定，你可以使用mmf的图像的controlnet来辅助)，不要忘了删除相反的prompt。

On this basis, you can make changes, such as deleting "leg up" and changing it to "standing, standing sex" (although this is more difficult, maybe you need to increase the weight), changing the character composition to "3girls, futanari, futa with female "(futa will make the picture more unstable, you can use the controlnet of the mmf image to assist),don't forget to delete the opposite prompt.

hiresfix是必要的，权重我一直使用1

hiresfix is necessary, weight I always use 1.

如果你发现使用人物lora后难以控制，我建议使用controlnet来进行辅助，首先生成一个不带人物lora的图片，然后将它放入controlnet中，最后加入人物lora后再次生成，以下为示例：

If you find it difficult to control after using the character lora, I suggest using controlnet for assistance. First generate a picture without character lora, then put it into the controlnet, and finally add the character lora and generate it again. The following is an example:

A LoHa trained on a measly 10 images using my method. 1 batch size.

She will be more anime/human or she will be closer to the original depending on prompt combinations and version used.

A LoHa trained on a measly 10 images using my method. 1 batch size.

She will be more anime/human or she will be closer to the original depending on prompt combinations and version used.

LyCORIS of Tulsi Gabbard

LyCORIS of Tulsi Gabbard

It's hard to generate half-closed eyes on normal models. So I trained this lycoris to obtain better half-closed eyes faces.

Trained on 175 face pic.

trigger word: half-closed eyes.

Samples are generated on Any4.5 with LoRA Block Weight

It's hard to generate half-closed eyes on normal models. So I trained this lycoris to obtain better half-closed eyes faces.

Trained on 175 face pic.

trigger word: half-closed eyes.

Samples are generated on Any4.5 with LoRA Block Weight

LyCORIS of Taylor Swift

LyCORIS of Taylor Swift

I used cp decomposition to make the size smaller.

You may need to update your locon plugin to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

Apologies for my inactivity. I was experimenting with different hyperparameters.

I used cp decomposition to make the size smaller.

You may need to update your locon plugin to fully utilise this model!!!

This model may not be compatible with the AddNet plugin.

Apologies for my inactivity. I was experimenting with different hyperparameters.

LyCORIS of Summer Glau

LyCORIS of Summer Glau

LyCORIS of Sophie Turner

LyCORIS of Sophie Turner

LyCORIS of Natalia Dyer

LyCORIS of Natalia Dyer

LyCORIS of Morena Baccarin

LyCORIS of Morena Baccarin

LyCORIS of Melanie Lynskey

LyCORIS of Melanie Lynskey

LyCORIS of Maya Hawke

LyCORIS of Maya Hawke

LyCORIS of Maisie Williams

LyCORIS of Maisie Williams

LyCORIS of Lena Headey

LyCORIS of Lena Headey

LyCORIS of Kristen Schaal

LyCORIS of Kristen Schaal

LyCORIS of Katee Sackhoff

LyCORIS of Katee Sackhoff

LyCORIS of Katarzyna Figura

LyCORIS of Katarzyna Figura

LyCORIS of Jennifer Tilly

LyCORIS of Jennifer Tilly

LyCORIS of Iga Świątek

LyCORIS of Iga Świątek