LTX-2I was the first to come out with an LTX-2 penis lora, and now you have several to choose from. Mine can be used to create penises from scratch (t2v) or to keep them from disappearing if you already have one (i2v).<ul><li>Version 1 - 650 images for 35,000 steps with the LTX trainer.</li><li>Version 2 - added videos to the version 1 images LoRA using the LTX trainer. It ended up being overbaked with dozens of checkpoints saved. I'm reluctant to release it, but it may be useful for image to video where you want a cumshot</li><li>Version 3 - this version was trained with the same image dataset as the rest of my current penis LoRAs as well as 55 videos, 512x512 x 121 assembled for training version 2. I used AI Toolkit and started all over again. I overbaked but saved snapshots along the way and rolled back and decreased the learning rate, and I am using the checkpoint from ~17k steps for this version 3 of the LoRA.</li><li>Version 4 - trained on improved captions, more videos. I have a preview on <a target="_blank" rel="ugc" href="https://huggingface.co/CoachBate/LTX-2.3/tree/main">HF</a> since I don't want to post here until I've given it more testing and I just haven't had time. Fixed the audio issue so you don't need to use the Advanced Lora Loader on it to mute out any noise (there's no noise).</li></ul>NOTE for version 3:I selected the option to train the audio from the video clips but the results just sound terrible and the white noise will dominate. You can turn off the audio from my LoRA with the LTX2 LoRA Loader Advanced node from comfyui-kjnodes. Set the video_to_audio, audio, and audio_to_video multipliers to 0 as shown in this screenshot. I hope to resolve this in a future version. Screenshot: <a target="_blank" rel="ugc" href="https://civarchive.com/images/121955512">https://civarchive.com/images/121955512</a>Check out these workflows that have renewed my faith in this model being actually useful: <a target="_blank" rel="ugc" href="https://huggingface.co/RuneXX/LTX-2.3-Workflows">https://huggingface.co/RuneXX/LTX-2.3-Workflows</a><span style="background-color:rgb(26, 27, 30);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">I welcome feedback but don't be a douche. If you're looking for chicks with dicks or uncut penises, I'm not the right creator for you. I create what I like, so sorry chicks with dicks and uncut penis lovers, not my thing. You're in luck because this site is full of that. Do let me know if you are able to prompt my stuff into doing sometime outside of the box pun intended with my creations. Version 1 required more detailed prompts to get good text to video results. <span style="background-color:rgb(26, 27, 30);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">You can also use my or other penis loras for image models to create your keyframes with penis in there, then animate them with this LoRA.<span style="background-color:rgb(26, 27, 30);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Grok will create filthy NSFW prompts for you for use with these models (at least at the time of this writing it does).

PENIS LoRA by CoachBate - LTX-2

LTX-2I was the first to come out with an LTX-2 penis lora, and now you have several to choose from. Mine can be used to create penises from scratch (t2v) or to keep them from disappearing if you already have one (i2v).<ul><li>Version 1 - 650 images for 35,000 steps with the LTX trainer.</li><li>Version 2 - added videos to the version 1 images LoRA using the LTX trainer. It ended up being overbaked with dozens of checkpoints saved. I'm reluctant to release it but you can message me for a copy via my <a target="_blank" rel="ugc" href="https://coachbate.ai">https://coachbate.ai</a> site</li><li>Version 3 - this version was trained with the same image dataset as the rest of my current penis LoRAs as well as 50 videos used for training version 2. I used AI Toolkit and started all over again. I overbaked but saved snapshots along the way and rolled back and decreated the learning rate, and I am using the checkpoint from ~17k steps for this version 3 of the LoRA.</li></ul>NOTE for version 3:I selected the option to train the audio from the video clips but the results just sound terrible and the white noise will dominate. You can turn off the audio from my LoRA with the LTX2 LoRA Loader Advanced node from comfyui-kjnodes. Set the video_to_audio, audio, and audio_to_video multipliers to 0 as shown in this screenshot. I hope to resolve this in a future version. Screenshot: <a target="_blank" rel="ugc" href="https://civitai.com/images/121955512">https://civitai.com/images/121955512</a>

Files with SHA256: ace0c62ef5cd08995cd02ad37a838978bc01d4d2b3e082207f5bc41dcadf98f2