Trained with offset noise


Trined on tachi-e (standing full body) with background from multiple mobile games. This LoRA is suited for both portrait & landscape orientation images.
Usage:
Trigger words:
white background, full body: tachi-e with background (separately described), try using (white background:1.5) if it's not working; the higher the strength of white background, the larger the white area
simple background, full body: tachi-e w/o background (no describe, portrait orientation)
white background, full body, border, outside border: tachi-e w/ background and border cut-out
white background, no human: scenery tachi-e w/o characters
Elements:
Clothes: (I'd emphasize for defeated tachi-e, use torn clothes, torn legwear, torn shirt, torn dress, also try adding and you'll see what happens)
Basic elements: fire/flame/water/ice/snow/lightning/electricity/clouds/wind/fog ... etc.
Color: {color} theme, use {color} background, white background and you'll get a two-tone background, also you can use colorful
Plants: tree/flower/floral background/meadow ... etc.
Animals: cat/dog/rabbit ... etc. and you can add monsters like ghost/zombie/dragon .. etc. See:
Architecture: gothic architecture/east asian architecture/cityscape/architecture/skyscraper/bridge/door ... etc.
Theme: fantasy/steampunk/cyberpunk ... etc. See:
Others:
Symbols: magic circle/hexagram ... etc. See:
Ornaments: book/chair/gem ... etc.
Expressions, postures & gestures: / /
Image composition: dutch angle/from above/fisheye ... etc. See:
All other things you can imagine
Use with other LoRAs: characters, poses, concepts etc.
Trained with offset noise