A-Ani Mix
์ธ๋ถ ์ ๋ณด
ํ์ผ ๋ค์ด๋ก๋ (1)
๋ชจ๋ธ ์ค๋ช
โจ A merged model, properly fine-tuned with large dataset-based LoRAs, then carefully re-merged again to refine and polish the overall aesthetics ๐จ
๐ด This is an anime-styled base model, designed to produce clean, stylized, and expressive visuals with a strong aesthetic focus โจ
๐งพ It also works very well with longer prompts, handling detailed descriptions smoothly while keeping strong coherence and style consistency โจ
๐ผ๏ธ Showcase Images: Generated completely raw | no ADetailer, no Highres.fix, and no img2img workflow used, to present the modelโs true base output quality โจ
โ๏ธ Samplers: Euler A, DPM++ 2M | both give stable results with good detail and consistency ๐ฏ
๐๏ธ CFG Scale: 5โ6 | recommended guidance for quality without overforcing/burning the imageโจ
๐ผ๏ธ Workflow: Highres.fix or img2img at higher resolution for cleaner final outputs ๐ (Optional)
๐๏ธ Face Detailing: ADetailer used for sharper and more accurate facial details. Always recommended for eyes๐ช
๐ Upscaler: Remacri, 4x-UltraSharp for crisp, high-quality upscaling and enhanced clarity ๐งพ
๐ผ Positive Quality Prompt (Boost)
amazing composition, ultra score, high score, depth of field, masterpiece, best quality, high quality, amazing quality, absurdres, highres, newest, best lighting, very aesthetic
(Optional: cinematic lighting, volumetric lighting, dramatic shadows, add more details)
๐ฝNegative Prompt
lowres, early, bad_quality, normal quality, lowres, bad anatomy, simple background, (worst quality, bad quality:1.2), jpeg artifacts, scan artifacts, ugly, deformed anatomy, bad proportions