<a target="_blank" rel="ugc" href="https://github.com/KohakuBlueleaf/LyCORIS">LoHa</a> of Happy Meek (umamusume). Put LoHa file in your stable-diffusion-webui/models/lora folder and write LoRA notation in prompt to applying LoHa. &lt;lora:your_loha_name:weight&gt;All of my works is free. If you'd like to support me, feel free to <a target="_blank" rel="ugc" href="https://www.buymeacoffee.com/mhtLoRA">buy me a coffee</a>. 👍NOTICE<ul><li>After careful consideration, I decided not to produce LoRA of the characters below.<ul><li><a target="_blank" rel="ugc" href="https://umamusume.fandom.com/wiki/Etsuko_Otonashi">Otonashi Etsuko</a>, <a target="_blank" rel="ugc" href="https://umamusume.fandom.com/wiki/Misato_Akasaka">Akasaka Misato</a>, <a target="_blank" rel="ugc" href="https://umamusume.fandom.com/wiki/Junko_Hosoe">Hosoe Junko</a></li></ul></li><li>If anyone wants the characters above, I will consider making them.</li><li>Except for the above, I will make the rest of the characters in Umamusume: Pretty Derby, unless there is a special notice.</li><li>Something weird happened when training. If you interested in training, read comments on LoRA training info section.</li></ul>Recommended options<ul><li>LoRA weight 0.6 ~ 0.7</li><li>Trigger words<ul><li>Base<ul><li>happy meek \(umamusume\)</li></ul></li><li>Original Art :<ul><li>hair flower, hairclip, striped bowtie, blue bowtie, brooch, jewelry, collared shirt, dress shirt, white shirt, blue vest, white jacket, open clothes, open jacket, puffy short sleeves, frilled sleeves, white gloves, pleated skirt, blue skirt, boots, white footwear</li></ul></li><li>Tracen summer uniform<ul><li>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</li></ul></li><li>Tracen winter uniform<ul><li>tracen school uniform, ear ornament, hairclip, winter uniform, puffy long sleeves, sailor collar, purple shirt, white bowtie, horseshoe ornament, purple skirt, pleated skirt, purple thighhighs, brown footwear, loafers</li></ul></li></ul></li></ul>Use a model derived from or mixed with <a target="_blank" rel="ugc" href="https://huggingface.co/a1079602570/animefull-final-pruned/blob/main/model-001.ckpt">animefull model</a>.<ul><li>Settings<ul><li><a target="_blank" rel="ugc" href="https://github.com/AUTOMATIC1111/stable-diffusion-webui">Stable Diffusion WebUI by Automatic1111</a></li><li>DPM++ Series(SDE Karras, 2M Karras, etc.)</li><li>about 20 steps, CFG scale 3.5~6.5</li><li>I recommend <a target="_blank" rel="ugc" href="https://huggingface.co/hakurei/waifu-diffusion-v1-4/blob/main/vae/kl-f8-anime2.ckpt">kl-f8-anime2.vae</a></li><li>CLIP skip = 1</li><li>Use Hires.fix to get higher quality image<ul><li>Upscaler<ul><li>Latent, Latent (nearest-exact), Latent (bicubic antialiased) or other Latent series.</li></ul></li><li>Denoising strength 0.50~0.65</li></ul></li></ul></li></ul>LoRA training info<ul><li>trained on <a target="_blank" rel="ugc" href="https://github.com/kohya-ss/sd-scripts">sd-scripts by kohya_ss</a>.<ul><li><a target="_blank" rel="ugc" href="https://github.com/KohakuBlueleaf/LyCORIS">LyCORIS by KohakuBlueleaf</a></li><li>Thank you all a lot!</li></ul></li><li>based on <a target="_blank" rel="ugc" href="https://huggingface.co/a1079602570/animefull-final-pruned/blob/main/model-001.ckpt">Animefull-pruned model</a></li><li>129 images consist of<ul><li>4x low quality 47 images</li><li>8x medium quality 46 images</li><li>16x high quality 36 images</li><li>=&gt; 1 epoch = 1132 images</li></ul></li><li>replaced character feature tags with happy meek \(umamusume\)<ul><li>1girl, red eyes, pink hair, simple background, solo, tail, braid, white hair, official alternate costume, long hair, alternate hairstyle, brown eyes, horse ears, purple eyes, horse girl, brown hair, blonde hair, alternate hair length, bangs, horse tail, blunt bangs, animal ears, pink eyes, yellow eyes, short hair, grey hair, light brown hair, ponytail, medium hair, seiun sky \(umamusume\), bob cut</li><li>I think these tags are the essentials of Happy Meek or</li><li>unnecessary for training because of duplication.</li></ul></li><li>resolution : 768x768 with aspect ratio bucketing.</li><li>other training settings is included in training data or metadata of LoRA file.</li><li>Yukino Bijin, Neo Universe, Zenno Rob Roy, Light Hello, Tracen Staff, Ines Fujin, Inari One and Tanino Gimlet were trained with learning rate of text encoder(TE) which has five times higher than Unet learning rate.<ul><li>This is very interesting. Most of people uses TE lr as smaller than Unet lr. I did mistakes but got a good results. I don't know why these LoRA works. It may a clue to help to set proper learning rate.</li><li>I found this phenomenon in training of happy meek. During training, loss goes to nan. Happy meek has two tokens, so I think that higher TE learning rate with small tokens of trigger word make model boom.</li></ul></li></ul>All uploaded images are generated by <a target="_blank" rel="ugc" href="https://huggingface.co/WarriorMama777/OrangeMixs/tree/main/Models/AbyssOrangeMix3">AOM3A1</a>And, special thanks to my best friend. DH.

Happy Meek (umamusume)

<a target="_blank" rel="ugc" href="https://github.com/KohakuBlueleaf/LyCORIS">LoHa</a> of Happy Meek (umamusume). Please install <a target="_blank" rel="ugc" href="https://github.com/KohakuBlueleaf/a1111-sd-webui-locon">extension </a>on WebUI.After installing extension, put LoHa file in your sd/models/lora folder and write LoRA notation in prompt to applying LoHa. &lt;lora:your_loha_name:weight&gt;

Files with SHA256: f11b49900fd32d464ad08369e26c99c82e1afcef4f379ce0653054d6a21b35fa