模型简介: ※A LoRa model trained by a friend with multires noise. This is a LoRa model that trained by 70+ img generated with NijiV5,and it's the multires noise version. This model can generate exquisite Qinghua-style clothing, decorations, and new Chinese-style backgrounds. weight 0.8-1.0 maybe generates better imgs. teapot, blue dress, hair ornament, chinese clothes, vase and so on are suggersted tag. ~~~~~~~~~~~~~~~~~~~~~~~o(* ̄︶ ̄*)o~~~~~~~~~~~~~~~~~~~~~~~ 示例提示词和参数: Prompt: 1girl, head,red eyes,hair ornament, flower, lattice, solo, table, scroll, sitting, window, round window, jewelry,, paintbrush, long hair, chinese clothes, indoors, vase,, earrings, looking down, long sleeves, hair flower, chair, hanfu, blue dress, bracelet, , beads, tassel, dress, holding, calligraphy brush, windowsill, teapot, closed mouth, shawl, blue flower, <lora:porcelain-000004:1> Negative prompt: extra fingers,fewer fingers,(low quality, worst quality:1.65), (bad anatomy), (inaccurate limb:1.2),bad composition, inaccurate eyes, extra digit,fewer digits,(extra arms:1.2), Steps: 26, CFG scale: 8, Sampler: DPM++ 2M Karras, Seed: 2017260175, ENSD: 31337, Size: 512x832, Model: mixProV4_v4, Model hash: 61e23e57ea, Hires steps: 9, Hires upscale: 1.75, Hires upscaler: 4x-UltraSharp, Denoising strength: 0.4, Clip skip: 2 |