<h1 id="20241109v1.5-77f98py8b">20241109v1.5</h1>Utaha Kasumigaoka from Saenai Heroine no SodatekataTrigger Words: utahakasumigaokaTrained with Pony Diffusion V6 XL checkpoint.I have set a moderation on the gallery, so the NSFW content may not appear. Due to "flagged for review", many images will be delayed by many hours before they can be viewed.The original style has been enhanced, but the characters' body movements and hand details still need improvement. Gradually, I've been figuring out how to train a highly accurate character LoRA, even when the dataset's references are suboptimal (e.g., flashy visuals, complex poses).<h2 id="problems-8qp2qhri5">Problems</h2><ul><li>Bad hands, changing a checkpoint to generate images may be better.</li><li>Anatomy not good enough.</li><li>Delete many outfits from dataset for a continuous style.</li></ul><h2 id="usage-tips-o90gflu1x">Usage tips</h2><ul><li>The model's output may heavily rely on the checkpoint.</li><li>The character's face, eyes, and pupils can be repaired through ADetailer.</li><li>You could find the example prompts from my post.</li><li>My prompts are basically composed in the order of [character traits] + [style] + [expression] + [clothing] + [camera and action] + [background], and you can delete or modify them as needed.</li><li>Recommended weight: 1~0.6, adjust as needed until the character's appearance meets your requirements.</li><li>Upscale value recommendation is around 1.2~2.0, denoising strength is 0.2</li></ul><h1 id="20241103v1-ghca31c40">20241103v1</h1>This training has given me new experience; Megumi Kato should also be able to be improved through this method.<h2 id="usage-tips-qlmm9r4se">Usage tips</h2><ul><li>The model's output may heavily rely on the checkpoint.</li><li>"3d" tag must be added to the negative prompt to avoid uncontrollable outputs.</li><li>The character's face, eyes, and pupils can be repaired through ADetailer.</li><li>You could find the example prompts from my post.</li><li>My prompts are basically composed in the order of [character traits] + [style] + [expression] + [clothing] + [camera and action] + [background], and you can delete or modify them as needed.</li><li>Recommended weight: 1~0.6, adjust as needed until the character's appearance meets your requirements.</li><li>Upscale value recommendation is around 1.2~2.0, denoising strength is 0.2</li></ul>

Utaha Kasumigaoka PonyXL

utahakasumigaoka.safetensors

Kasumigaoka Utaha / Saenai Heroine no Sodatekata

v1.5

v1.0

20241109v1.5Utaha Kasumigaoka from Saenai Heroine no SodatekataTrigger Words: utahakasumigaokaTrained with Pony Diffusion V6 XL checkpoint.I have set a moderation on the gallery, so the NSFW content may not appear.Due to "flagged for review", many images will be delayed by many hours before they can be viewed.The original style has been enhanced, but the characters' body movements and hand details still need improvement. Gradually, I've been figuring out how to train a highly accurate character LoRA, even when the dataset's references are suboptimal (e.g., flashy visuals, complex poses).Problems<ul><li>Bad hands, changing a checkpoint to generate images may be better.</li><li>Anatomy not good enough.</li><li>Delete many outfits from dataset for a continuous style.</li></ul>Usage tips<ul><li>The model's output may heavily rely on the checkpoint.</li><li>The character's face, eyes, and pupils can be repaired through ADetailer.</li><li>You could find the example prompts from my post.</li><li>My prompts are basically composed in the order of [character traits] + [style] + [expression] + [clothing] + [camera and action] + [background], and you can delete or modify them as needed.</li><li>Recommended weight: 1~0.6, adjust as needed until the character's appearance meets your requirements.</li><li>Upscale value recommendation is around 1.2~2.0, denoising strength is 0.2</li></ul>

20241109v1.5
Utaha Kasumigaoka from Saenai Heroine no Sodatekata

Trigger Words: utahakasumigaoka

Trained with Pony Diffusion V6 XL checkpoint.

I have set a moderation on the gallery, so the NSFW content may not appear.

Due to "flagged for review", many images will be delayed by many hours before they can be viewed.

The original style has been enhanced, but the characters' body movements and hand details still need improvement. Gradually, I've been figuring out how to train a highly accurate character LoRA, even when the dataset's references are suboptimal (e.g., flashy visuals, complex poses).

Problems
Bad hands, changing a checkpoint to generate images may be better.

Anatomy not good enough.

Delete many outfits from dataset for a continuous style.

Usage tips
The model's output may heavily rely on the checkpoint.

The character's face, eyes, and pupils can be repaired through ADetailer.

You could find the example prompts from my post.

My prompts are basically composed in the order of [character traits] + [style] + [expression] + [clothing] + [camera and action] + [background], and you can delete or modify them as needed.

Recommended weight: 1~0.6, adjust as needed until the character's appearance meets your requirements.

Upscale value recommendation is around 1.2~2.0, denoising strength is 0.2

Creator: <a target="_blank" rel="ugc noopener noreferrer nofollow" href="https://civitai.com/user/sumai">Sumai</a>Recommended weight: 0.6~1.0, adjust as needed until the character's appearance meets your requirementsTrigger word: utahakasumigaokaUsage tips:<ul><li>The model's output may heavily rely on the checkpoint</li><li>The character's face, eyes, and pupils can be repaired through ADetailer</li><li>You could find the example prompts from showcase images</li><li>Prompts are basically composed in the order of [character traits] + [style] + [expression] + [clothing] + [camera and action] + [background], and you can delete or modify them as needed</li><li>Upscale value recommendation is around 1.2~2.0, denoising strength is 0.2</li></ul>

Files with SHA256: ec1f5790f3c3328bc8210bf2210cc2da6127d3ec4a676dcbc11163310d0a8ff5