基于midjourney生成的的图片数据集合训练而来，特点是画风复古简约，生成的图比较偏近景，并且人物有点偏向圆圆的包子脸，试了几个模型好像都能不同程度的起作用，但总体而言checkpoint还是用noobai生成的图的画面更好，感觉可以用作书本插画，生成多人的图容易崩可能要开重绘修复， 关键字：simpleillust, simple colors, simple art, flat color, simple strokes, simple, simple lines, toon \(style\), 可选：chibi,wide view，low-angle view， high-angle view，bird's-eye view，dutch angle等等镜头或者距离相关的词<hr />Trained on a dataset generated by Midjourney, this model features a retro and simple art style. The generated images tend to focus on close-up compositions, with characters often having round, bun-like facial features. While testing different models, they all seem to work to varying degrees, but overall the checkpoint using Noobai-generated images delivers better visual quality, this lora may suitable for book illustrations. Multi-character images may require Adetailer to fix distortions.Prompt:simpleillust, simple colors, simple art, flat color, simple strokes, simple, simple lines, toon (style)Optional:chibi, wide view, low-angle view, high-angle view, bird's-eye view, dutch angle, and other camera/lens-related terms regarding perspective or distance.

Files with SHA256: 7c477625d9aaf3cae9fbe329ad6e9a71a49d4b820d0428f327be4444c5ba0ea0