Kindly share the first image directly in the Model Gallery. This is the best way to support and motivate creators!
For creating Lora & Concepts (SDXL | Flux | Z Image), feel free to Give me Direct DM via CivitAI. While I may not be able to create every concept, I will definitely bring the best ones to life with care and attention to detail.
Custom LoRA work:
1. 500 Yellow Buzz (Pony, Illustrious, SDXL)
2. 1000 Yellow Buzz (ZTurbo or Flux)
3. Delivered within a week
Models are delivered within a week. Partial or insufficient Buzz will be ignored and not refunded to keep the work sustainable.
Features:
Usage (Automatic1111 / Comfyui):
Sampling method: DPM++ SDE Karras / DPM++ 2M Karras / Eulera Automatic
Sampling steps: 24-35 Steps
Resolution: 823x1216 / 1024x1024 / 1216x823
CFG: 4-7
Clip Skip: 1
Upscale : 4x-AnimeSharp
Upscale By : Any
Denoising Strength: 0.35
Hires steps: 13-17Positive Prompt:
score_9_up, score_8_up, score_7_up, source_anime, detailedNegative Prompt:
score_6, score_5, score_4, simplified, abstract, unrealistic, impressionistic, low resolution, lowres, bad anatomy, bad hands, missing fingers, worst quality, low quality, normal quality, cartoon, anime, drawing, sketch, illustration, artificial, poor qualityRecombination's:
When prompting, I suggest not using the masterpiece, best quality tags, but rather preface your prompt with "Preferred Trigger word".
Don't use a lot negative embeddings at the same time. It definitely will ruin any possible output on all models! If you want to use negative embeddings - use one for quality, and one for hands (if you want a placebo for hands).
Resources:
Upscaler : 4x-AnimeSharp , 4x-Ultrasharp , 4x NMKD
Upscaler placement in directory
Comfyui,Automatic1111/ └── models/ └── ESRGAN/ └── [Your Upscaler Model File]This model is capable of generating both SFW and NSFW content. However, we recommend using it for SFW purposes, as there are many excellent results to be achieved with it.
Any misuse of the model is solely your responsibility, and we are not liable for any consequences. That said, the model is free to use and download.
Please consider supporting us by tipping, sharing the model, and following us on other platforms. Your support means a lot—thank you!
Description
Main Trigger Word: mikasa
FAQ
Comments (6)
It looks good. The first Mikasa I actually wanted to download.
But i DO have to ask, why are you using score tags? Those are only required in PONY models because of how the PONY maker made their checkpoint, they -as far as I know - do nothing for ILL, unless I've been misinformed.
Its Chatgbt genrated prompts i am too lazy to manually write 20 unique prompt, just ignore scores chatgbt adds that automatically, scored do nothing in ILL it just acts as plain text not prompt
score_9_up, score_8_up, score_7_up, source_anime, detailed
This thing in description that is given is for pony ie i was a pony creator before the release of ILL. Everthing Setting remain same apart from scores which is unnessary now, Tho Even if you add them it wont harm the image output
@thevoiid Fair enough. I was just curious.
I've seen a few others also putting score tags and I just happened to want to ask here finally.
But again, this is the first Mikasa model I've actually wanted to download. I LIKE Mikasa, but the other models from other creators just didn't look good enough.
@WorstAirtist The reason is simple Most creators lack large, high-quality datasets that properly match the anime art style and I don’t blame them—building good datasets is always difficult. Another issue is that many use CivitAI LoRA trainers, which often destroy style consistency and creativity. That’s why I consider CivitAI LoRA trainers poor choices. Either train locally or use Runpod
Wow. It's simply amazing. It's the best Mikasa model; she's identical to the anime version.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.