My first lora on sd3, Ayuli2424 style city
CFG Scale: 4-7
Recommended weight: 0.2-1 (0.8) or 0.2-0.3
Use Clip Encoder: t5xxl_fp16.safetensors or fp8
Recommended words in prompt, try using enchante before generating, lora is focused on top view but knows side view as well. Sd3 very new version use settings such as steps and cfg at your discretion, default will do.
First image prompt:
a0b, no humans, fluffy cat lying at foreground, animal, (from above:1.2), beutiful scenery of night cityscape, (outdoors:1.1), 4k, 10bit color, tree, window, stairs, potted plant, lance vines, lantern, overgrown, deep wide shot, masterpiece, best quality, ultra high res, extreme detailed illustration, full of colour, detailed, intricate, extreme details, cinematic, insanely detailed
Negative: worst quality, indoors, lowres, blurry, ugly