我第一次用 Krea 2 测试 😀

My first test with Krea 2 😀 These types of styles used to be only possible with Midjourney but no anymore, I am so excited to do a lot of LORA training and share them with everyone and I know the open source community been cooking with Krea 2 for some time but I believe my system prompt always can improve your results 😜😜😜 System prompt Create a [exact aspect ratio] horizontal [2D anime / illustrated] turnaround model sheet. The page contains exactly [three or four] complete full-body drawings of one fictional adult character, arranged [one evenly spaced horizontal row / another simple explicit layout] on [one shared baseline]. Every figure is shown from [crown to sole] at equal scale. From left to right: [front view], [three-quarter view], [profile], [rear view]. Every view has the same face, body proportions, haircut, clothing, and visible equipment. The character stands in [one neutral repeated pose], with [hand position] and [foot position]. The figures do not overlap. The character has [three to five identity anchors: face, skin tone, eye colour, hair silhouette, one distinctive feature]. They wear [one coherent outfit described through silhouette, materials, closures, and one or two fixed equipment items]. The page background is [plain drafting-paper colour]. Use [line treatment], [shading method], [lighting], and [material treatment]. The sheet contains only the declared full-body views, shared baseline, and empty background. Generate face studies, hands, props, expressions, colour swatches, and fabric callouts as a separate sheet.

我第一次用 Krea 2 测试 😀 这种风格以前只能用 Midjourney 才能做到，但现在不再是这样了。我非常兴奋要做大量的 LORA 训练，并把它们分享给大家。我也知道开源社区用 Krea 2 已经折腾了一段时间，但我相信我的系统提示词始终可以让你的效果更好 😜😜😜 系统提示词 创建一张 [exact aspect ratio] 的水平 [2D anime / illustrated] 转面模型表（turnaround model sheet）。页面恰好包含 [three or four] 张同一虚构成人角色的完整全身图，按 [one evenly spaced horizontal row / another simple explicit layout] 排列在 [one shared baseline] 上。每张图均以 [crown to sole] 等比例展示。 从左到右依次为：[front view]、[three-quarter view]、[profile]、[rear view]。每个视角的面部、身体比例、发型、服装和可见装备完全一致。角色以 [one neutral repeated pose] 站立，包括 [hand position] 和 [foot position]。图形之间不得重叠。 角色拥有 [three to five identity anchors: face, skin tone, eye colour, hair silhouette, one distinctive feature]。身穿 [one coherent outfit described through silhouette, materials, closures, and one or two fixed equipment items]。 页面背景为 [plain drafting-paper colour]。使用 [line treatment]、[shading method]、[lighting] 和 [material treatment]。 该图只包含已声明的全身视角、共享基线和空白背景。面部研究、手部、道具、表情、色板和面料标注另起一张图生成。