[LoRA] Cartoon portrait卡通证件照

模型简介：
如果您喜欢这个 LoRa，请添加 ❤️ 以接收未来更新。
SD 使用和训练教程文档请访问：afdian，模型定制：vx: aimonstergo，国内镜像：卡通证件照｜LiblibAI

Prompt: best quality,cartoon_portrait, 1girl, solo, long hair, breasts, looking at viewer, simple background, brown hair, bare shoulders, brown eyes, closed mouth, upper body, earrings, sleeveless, lips, white sweater <lora:cartoon_portrait_v2:0.7>

采样器： DPM++ SDE Karras
模型： v1-5-pruned-emaonly (sd 1.5 原版模型)
Clip skip: 2
生成全身尺寸图像时启用 Adetailer
ADetailer模型： face_yolov8n

注意：
1. 不要使用恢复面，使用细节工具
2. 添加触发词来生成图像
3. 使用 controlnet tile 来放大图像

~~~~~~~~~~~~~~~~~~~~~~~o(*￣︶￣*)o~~~~~~~~~~~~~~~~~~~~~~~

主图提示词和参数：
Prompt: best quality, cartoon_portrait, 1girl, solo, long hair, looking at viewer, blonde hair, simple background, shirt, brown eyes,  white shirt, pink hair, multicolored hair, earrings, lips, <lora:cartoon_portrait_v2:0.7>

Negative prompt: bad-hands-5 EasyNegativeV2 ng_deepnegative_v1_75t bad-image-v2-39000

Steps: 20, Size: 1024x1280, Seed: 3737549008, Model: v1-5-pruned-emaonly, model: control_v11f1e_sd15_tile [a371b31b], weight: 1, Version: v1.4.1, Sampler: DPM++ SDE Karras, CFG scale: 4, Clip skip: 2, {"Method": "Mixture of Diffusers", "Upscaler": "R-ESRGAN 4x+", Model hash: 6ce0161689, resize mode: Crop and Resize, control mode: ControlNet is more important, "preprocessor: tile_resample, pixel perfect: True, "Tile Overlap": 48, starting/ending: (0, 1), "Upscale factor": 2, "Keep input size": true}, "Tile batch size": 4, "Tile tile width": 96, "Tile tile height": 96, Denoising strength: 0.3, preprocessor params: (-1, 1, -1)", "cartoon_portrait_v2: b7d71581b8b7", Tiled Diffusion upscaler: R-ESRGAN 4x+, Tiled Diffusion scale factor: 2