Hiya, this is my first poke at a LoRA with musubi-tuner. Krea2 seems a little light on pose LoRAs and this interaction is challenging to get when I try to prompt for it.It blends decently with other LoRAs in most of my tests, but you might wanna balance the overall strengths.27 image training set, auto captioned with Qwen Omni 30B. Selected epoch 16 after a few tests.<pre><code>Trigger: full nelson wrestling hold

Strength = 0.5 - 1.5, lower values the hold is more relaxed or loose, 1.5 seems more aggressive

Sampler = Euler / Simple, might work with others but that's what I normally test with</code></pre>Base prompt starting point, expand on it from there:<pre><code>full nelson wrestling hold, &lt;camera position&gt;, &lt;general setting&gt;

The attacker is &lt;character description&gt;.

The defender is &lt;character description&gt;.

&lt;additional details about character interaction&gt;

&lt;additional setting details&gt;</code></pre>Lmk if it works out or has issues and I might try another run with a bigger data set. ✌

Files with SHA256: c06e41a04b94ee116d0ab34222c0a030f040c01588f8d1209e90beb37c931a30