
依然是用一个全身图片训练的,更换姿势还是略有一定的难度,使劲加权重,
人物提示词看情况和权重配合使用能尽量改变动作,Marie Therese+0.2都行,
发饰原因,AI概率画成双马尾,
Google Translate:
It is still training with only one full-body picture, and it is still a little difficult to change the posture, so increase the weight.
Character prompts can be used in conjunction with weights to change actions depending on the situation, Marie Therese+0.2 is also possible.
描述:
我发现这角色脑袋下面都是对称的,就镜像了几张图尝试抵消那一成不变的动作。
Google Translate:
I found that the character below the head is symmetrical, so I mirrored a few pictures to try to offset the unchanging movements.
训练词语: marie therese
名称: Marie Therese v2.0.safetensors
大小 (KB): 147567
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success