嗨Stable Diffusion是一款利用深度学习的文生图模型支持通过使用提示词来产生新的图像描述要包含或省略的元素。 我在这里引入 Stable Diffusion 算法中的 Prompt 概念又被称为提示符。 这里的 Prompt 通常可以用来描述图像他由普通常见的单词构成最好是可以在数据集来源站点找到的著名标签比如 Danbooru。 下面我将说明 Prompt 的生成步骤这里的 Pro

"(best quality), (masterpiece), 1boy, martial arts, muscular, fierce, traditional Chinese clothing, holding a sword, mid-air jump, mountainous background"
"(best quality), (masterpiece), 1girl, martial arts, agile, graceful, hanfu dress, spinning kick, bamboo forest background"
"(best quality), (masterpiece), 2boys, martial arts, synchronized movements, serious expressions, samurai armor, sword fighting, cherry blossom garden background"
"(best quality), (masterpiece), 1boy and 1girl, martial arts, teamwork, determined expressions, flowing robes, counter-attack, ancient temple background"
"(best quality), (masterpiece), 1boy, martial arts, calm and collected, dragon robe, defensive stance, ancient palace background"
"(best quality), (masterpiece), 1girl, martial arts, elegant and deadly, silk dress, hidden weapons, moonlit lake background"
"(best quality), (masterpiece), 2girls, martial arts, fierce competition, colorful Hanfu, acrobatic moves, bustling marketplace background"
"(best quality), (masterpiece), 1boy, martial arts, blindfolded, monk's robe, unarmed combat, misty mountain background"
"(best quality), (masterpiece), 1girl, martial arts, mysterious and alluring, black leather outfit, grappling hook, abandoned castle background"
"(best quality), (masterpiece), 1boy, martial arts, training under a waterfall, straw hat, wooden staff, rocky gorge background