<h2 id="*****-this-lora-doesn't-work-alone-*****">***** THIS LORA DOESN'T WORK ALONE *****</h2>This is a slider LoRA. It doesn't contain image data so it won't be able to draw a penis with default Wan. It needs another LoRA with penises to work.What is does is put a scale from -1 to 1 on the absolute smallest and biggest possible penis the model can conceptualize -- not necessarily able to draw it.It won't be able to make a penis a size outside the range of what the other LoRA you use can do. If you push it outside of the range of the other LoRA, it will either do nothing or create flesh monsters depending on the LoRA.It's really just like using a bunch of adjectives in your prompt. What is it for then? Finer control.Positive values: Bigger. Absolute maximum 1.Negative values: Smaller, Absolute minimum -1.The real range will depend entirely on the other LoRA you use.I generated images with various LoRAs to show what it can do. (sorry no videos, it would have taken too long to test).The LoRA was trained with Ostris AI-Toolkit in concept slider mode: <a target="_blank" rel="ugc" href="https://github.com/ostris/ai-toolkit">https://github.com/ostris/ai-toolkit</a>Target class: penisPositive training prompt: big penisNegative training prompt: small penisAnchor class: personHonestly, I'm surprised this worked at all. It seems like Wan actually has a very good understanding of penises but no data to draw them. The sample images generated during training were absolutely not pleasant to look at.

Files with SHA256: 73d773bca54837bef34ab871df45149268c2d3e637dc46e1991402b4dfd86b85