Note: V2 can generate everything that V1 was capable of. Since the V1 model struggled with latex mask scenes, V2 has been fine-tuned specifically to handle them better. If latex mask scenes are not of interest to you, you may continue using the previous version instead. This LoRA includes a variety of scene styles: trample (thrt), footgag, latex mask (feet inside), and licking.Please note that in Wan2.2, the seed value has a very strong influence. If the desired scene doesn’t appear, first try changing the seed. If that still doesn’t help, slightly increase the LoRA weight.<ul><li>Works well with HIGH model at around 70–95% strength.</li><li>For the LOW model, 70% and above usually gives good results.</li></ul>Keep in mind this LoRA was trained on many different scenes. If a specific result doesn’t come out as expected, please don’t rage. Feel free to leave a comment instead, and I’ll try to help.Example prompts for each scene are available in the Google Drive link below. You can adjust and modify the scenes based on those samples.<ul><li>Only for footgag scenes, the trigger word “Footgagging” is required.</li><li>Other scenes don’t use trigger words.</li></ul>This LoRA works for both text-to-video (t2v) and image-to-video (i2v). I’ll continue to share more examples over time. Example Prompts: <a target="_blank" rel="ugc" href="https://drive.google.com/file/d/1c_cRa_vkZS24uLnnTZfShwiG0Ld-06Fv/view?usp=sharing">https://drive.google.com/file/d/1c_cRa_vkZS24uLnnTZfShwiG0Ld-06Fv/view?usp=sharing</a> ⚠️ Please avoid posting extreme scenes here.

Wan2.2 HIGH and LOW models. Check the models in the versions section below. t2v / i2v fmdm, fgag, thr trample, lick, latex mask,

Because the v1 version had difficulty with latex mask scenes, the v2 model was fine-tuned. If latex mask scenes aren't important to you, you can use the older version.

Files with SHA256: 709263ba352656447bf78c8db483fa74427c8b4f22127a1a9db58ca7fb3d9ba2