Ayuli2424 あめ https://www.pixiv.net/en/users/7758885 Guidance Scale: 0 or whatever you want Recommended weight: 0.5-0.75 (0.7) After many versions, something roughly stable. Increasing the lora weight, will give more original style, but can lower the detail and create artifacts, above 0.8 is not recommended. Since this is Flux use natural language. Example prompt of the first image: a digital illustration in a vibrant, detailed style. an overgrown cityscape, waterfall. You can use a prompt of up to 512 tokens, you can experiment and write what you want FLUX is very flexible. You can specify different styles, anime, cartoon and painting are closest to the original style. It is better to specify, type of image, style, desired prompt and then the viewing angle. It is recommended to try top view and isometric angle.