Resized to 68% of original (view original)

Prompt | source anime, masterpiece, best quality, <lora:mirin_chikuwa_style_v1_illustrious:1>, 2coleslaw, BREAK 1girl, monster girl, lamia, scales, slit pupils, fangs, serious, BREAK outdoors, cave, looking at viewer, glowing eyes, volumetric lighting |
---|---|
Negative prompt | lowres, worst quality, low quality, jpeg artifacts, bad anatomy, bad hands, bad quality, censored, watermark, signature, username, (mosaic censoring:1.5) |
Anonymous 06/12/25(Thu)09:53:13 No.11309084
BlueGlow.png (2.82 MB, 1248x1920)
>>11309032
The reason I used that model is that it's a fairly neutral model (similar to 102d vpred) that deals well with style/artist tags but has a greater bias towards e621 for more precise ball-size control (as seen on that goblin sequence) and better animal genitalia.
It does have a slight realism bias but that can be managed by other style tags or stamped out with negatives. You may also be interested in StableMondAI-SDG, which is an older, more 2d-focused version of that same model.