最近在网上看到了这种姿势，虽然能用控制网出但是不太稳定，于是使用我跑出的图进行了训练，能稍微稳定一点触发词：<pre><code>inverted cross bondage</code></pre><pre><code>1girl,solo,bondage, rope bondage, crucifixion, arms up, arms tied, legs spread,</code></pre>由于我不太会炼概念lora，模型可能存在一些不稳定的情况，需要调整权重和使用特定的提示词I've recently seen this pose online. Although it can be generated using ControlNet, it's not very stable. Therefore, I used the images I generated to train a model, which has made it slightly more stable.Trigger words:<pre><code>inverted cross bondage</code></pre><pre><code>1girl, solo, bondage, rope bondage, crucifixion, arms up, arms tied, legs spread</code></pre>Since I'm not very good at training concept LoRA, the model may have some instability issues. It might be necessary to adjust the weights and use specific prompts.

Files with SHA256: a8158066c5350af8c8392bb1ff47102045cadb1075db327cd9251432a4dda7c6