Mikami Yua Lora for SD1.5Base model: majicrealistic v7 (SD1.5)三上悠亞 三上悠亚 LORA 用於 SD1.5基礎模型：majicrealistic v7 (SD1.5)________________________params suggested:weights : 0.8-1.0trigger word (recommended) : sanshangyouya推薦參數權重：0.8-1.0觸發關鍵詞（推薦使用）：sanshangyouya_________________________existed problems: watermark.現有問題：水印。_________________________enjoy yourself!

Hfda | 三上悠亞 Mikami Yua 三上悠亚 LORA

在這次的版本更新中，訓練圖片增加到了40張，並將Lora模型的rank提升至64，從而獲得了更細緻的面部特徵。 此外，透過在資料集標註中加入「Watermark」標籤，緩解了v1版本中的浮水印問題（但該問題仍然存在）。这次版本更新中，训练图片扩充到了40张，将Lora模型的rank提高到了64，从而获得了更细节的面部特征。 此外，通过在数据集标注中纳入“Watermark”标签，缓解了v1版本的水印问题（但仍然存在该问题）。In this version update, the number of training images has been increased to 40, and the Lora model's rank has been raised to 64, resulting in more detailed facial features. Additionally, by including the "Watermark" tag in the dataset annotations, the watermark issue present in version 1 has been alleviated (although it still persists to some extent).

Files with SHA256: 54fbbd1d3aa56e04ade1e675743e624c0b5b17b7805bb6ba3e9a31976fe86150