Ozen Lora

An Ozen from Made in Abyss LoRA. I'm a relative newbie at character LoRAs, so bear with me, but the results I got were satisfying, and considering the relatively low amount of art she has, I thought this could be nice. The token to invoke the lora 'ozensama'.

I copied YoungMasterFromSect's approach and am posting two versions,

the 'softer' one should be easier to convince to wear different clothes, while the 'harder' one should just mostly stick to her normal outfit.

This is trained on base novelai so any models off of that should do fine. The dataset contained no lewds and not much of how Ozen's bare arms should look like, so it won't be able to recreate that.

Model comparison

Previews

All previews are with the harder LoRA version, as well as my own model, you should achieve similar results with Anything V4.5 and other models.

Some of the images also use the 'dpep2 768' LoRA from the wonderful CloserToDeath, it improves backgrounds tremendously.

Sample Prompt
(best quality), watercolor \(medium\), (ultra-detailed), (best illustration), (best shadow), (an extremely delicate and beautiful), 1girl, solo, ozensama, ozen, black cape, collared cape, black eyes, multicolored hair, looking at viewer, black gloves, short hair, white hair, black hair, wristband, black jacket, belt, pouch, whistle, (empty eyes:0.8), pale skin, pants, very detailed background, mksks style, highly detailed background,
Negative prompt: multiple views, (low quality, worst quality:1.4), (bad anatomy), by bad-artist, bad-hands-5, bad-image-v2-39000, extra digit , fewer digits, (extra arms:1.2), bad hands,
Steps: 35, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 3877896406, Size: 768x512, Model hash: f30926bf30, Model: youcha, Denoising strength: 0.3, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: ozensama-000003(f2bba21f), AddNet Weight 1: 0.8, AddNet Module 2: LoRA, AddNet Model 2: dpep2 768(4f9473ab), AddNet Weight 2: 0.3, Hires upscale: 2.5, Hires steps: 15, Hires upscaler: 4x-AnimeSharp, Discard penultimate sigma: True

Sample Prompt
(best quality), (ultra-detailed), (best illustration), (best shadow), (an extremely delicate and beautiful), 1girl, ozensama, ozen, black cape, collared cape, black eyes, multicolored hair, looking at viewer, black gloves, short hair, white hair, black hair, wristband, black jacket, belt, pouch, whistle, (empty eyes:0.8), pale skin, detailed background, pants, very detailed background, paw pose, cat ears
Negative prompt: (low quality, worst quality:1.4), (bad anatomy), by bad-artist, bad-hands-5, bad-image-v2-39000, extra digit , fewer digits, (extra arms:1.2), bad hands,
Steps: 25, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 553564940, Size: 512x768, Model hash: f30926bf30, Model: youcha, Denoising strength: 0.3, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: ozensama-000003(f2bba21f), AddNet Weight 1: 0.8, AddNet Module 2: LoRA, AddNet Model 2: dpep2 768(4f9473ab), AddNet Weight 2: 0.5, Hires upscale: 2, Hires steps: 15, Hires upscaler: 4x-AnimeSharp

Sample Prompt
(best quality), (ultra-detailed), (best illustration), (best shadow), (an extremely delicate and beautiful), 1girl, ozensama, ozen, black cape, collared cape, black eyes, multicolored hair, looking at viewer, black gloves, short hair, white hair, black hair, wristband, black jacket, belt, pouch, whistle, (empty eyes:0.8), pale skin, detailed background, pants, forest, very detailed background, hand up, V, smile, cat ears
Negative prompt: (low quality, worst quality:1.4), (bad anatomy), by bad-artist, bad-hands-5, bad-image-v2-39000, extra digit , fewer digits, (extra arms:1.2), bad hands,
Steps: 25, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 2496261555, Size: 512x768, Model hash: f30926bf30, Model: youcha, Denoising strength: 0.3, Clip skip: 2, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: ozensama-000003(f2bba21f), AddNet Weight 1: 0.8, AddNet Module 2: LoRA, AddNet Model 2: dpep2 768(4f9473ab), AddNet Weight 2: 0.5, Hires upscale: 2, Hires steps: 15, Hires upscaler: 4x-AnimeSharp

Sample Prompt
(best quality), (ultra-detailed), (best illustration), (best shadow), (an extremely delicate and beautiful), 1girl, ozensama, ozen, (maid dress), black eyes, multicolored hair, looking at viewer, (black gloves), short hair, white hair, black hair, whistle, (empty eyes:0.8), pale skin,
Negative prompt: (low quality, worst quality:1.4), (bad anatomy), by bad-artist, bad-hands-5, bad-image-v2-39000, extra digit , fewer digits, (extra arms:1.2), bad hands,
Steps: 35, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 205714489, Size: 512x768, Model hash: f30926bf30, Model: youcha, ENSD: 31337, AddNet Enabled: True, AddNet Module 1: LoRA, AddNet Model 1: ozensama-000003(f2bba21f), AddNet Weight 1: 1.0

Contact

If you have any questsions you can bother me on Twitter

My pixiv if you're up for lewds: Pixiv

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference API
Unable to determine this model's library. Check the docs .