Sulphur 2 has been trained to know what a penis is, not a great looking penis, so coach is here to the rescue. Finetuned against Sulphur 2 base model checkpoint using my latest video dataset. You can use it with 10 eros for image to video workflows as well as sulphur 2 workflows. Some of the text to video samples are coming out great. especially when paired with my gay sex LoRA if that's your thing. If not, there's no gay sex, per se, trained in any of my penis LoRAs.

This LoRA can do text to video if creating from just a prompt. If you get backwards dicks or other oddities, prompting can usually fix it. For best results or more control over what you're going to get out of the video, use one of my penis image LoRAs to generate start and optionally end frames and use the 10 Eros checkpoint for i2v workflows. I put a modified workflow of 10 Eros' workflow in the optional files section here. I think my best penis images are created with either my Z Image Base or Qwen Image 2512 LoRAs. It uses the online Gemma text encoder, so you don't need to load the clip, and saves at least 20 seconds of time, you're welcome.

How to get a key:

1. Go to console.ltx.video

2. Sign up or log in

3. Generate a free API key

4. Copy it into the node’s api_key field