This one is more stable LoRA of this character, images will give some glitchy lines or pixelized styles most often that is intentional, use the trigger keywords beside the loRA
also.... how I gonna put this as 'version' help
This is *experimental* LoRA training, to see and understand the process, please do not use or download via low quality image generation might happen.
use trigger keywords:
''1girl, solo, loonalone, brown eyes, brown hair, long hair, streaked sidelocks, ahoge, cherry earrings,''
choose clothing style:
''oversized hoodie, beige hoodie, wide sleeves, long skirt, high-waist skirt, plaid skirt, brown skirt,''
''cozy winter sweater, beige sweater, long sleeves, dark jeans,''
''sundress, sleeveless dress, floral pattern, white cardigan,''
lastly choose posing and composition style:
''masterpiece, best quality, good quality, newest, highres, absurdres,focused subject,8k,HDR''
''glitch art, dithering, chromatic aberration, pixel art, film grain''
''golden hour, crowded background, festival atmosphere, warm lighting, bokeh lights, depth of field,''