Knows two types of clothes: dress and bikiniFor PonyDiffusion V6 be sure to use the tag score_9The strength of LoRA depends on the prompt. Typically, the larger and more complex it is, the more force required, up to 1.0. But for simple and short prompts it is better to use a strength of 0.7.Model: <a target="_blank" rel="ugc" href="https://civarchive.com/models/260267?modelVersionId=293564">animagine-xl-3.0.safetensors</a> or <a target="_blank" rel="ugc" href="https://civarchive.com/models/257749?modelVersionId=290640">ponyDiffusionV6XL_v6.safetensors</a>Dataset size: 109Clip skip: 2 (no difference with clip1)dim32, alpha16lr=0.0003, constant, adafactorOptimizer extra arguments: scale_parameter=False relative_step=False warmup_init=Falseversion for SD1.5\NAI <a target="_blank" rel="ugc" href="https://civarchive.com/models/52832/yamada-elf-emily-or-eromanga-sensei">click</a>My tips for generating on SDXL for AUTOMATIC1111Resolution:<ul><li>1024 x 1024</li><li>1152 x 896 or 896 x 1152</li><li>1216 x 832 or 832 x 1216</li></ul>Don't use Hires. fix. Send the picture to img2img, set CFG 0.15-0.25, then select “SD upscale” in the script, you also need to select 4x-AnimeSharp or what you like best from upscale. All other settings are default. Click generate.For me 12gb VRAM is enough for this using --xformers

Files with SHA256: 7004fbaf6f543c86a23fcb39d0b659a0e906c668d6f07807a9a8a8e4de68af83