I know there are a number of these cum loras already out already, but I wanted something with a bit more... volume. This lora was trained exclusively on photos of women just completely drenched in man-juice. Primarily facials, but also chest and body. It can help with both adding a few extra ropes to your already drenched subject when used with multiple loras, or it also does well with fixing realism on generations when used as a stand-alone lora.Lora weight use can range from 0.5 to 1.0. When stacking, I tend to use around 0.8 along with some strong prompting.Trigger word is: "plastered with cum". When I want a lot of cum, I will add one of the following combinations. Feel free to use as you see fit:<pre><code>((plastered with cum. facial. bukkake))</code></pre><pre><code>plastered, covered in cum. (facial). ((bukkake)). a cum tribute to the woman. She takes it all (cumshot:1.3) (cum on face:1.3)) (white thick cum:1.1)</code></pre>To be honest, I've had this lora trained for a while, but it produced a lot of watermark artifacts. I retrained it with the watermarking removed, but the end results are actually different than the prior version. I've been trying to determine which version of the lora is worth releasing, ultimately landing on Version B. If people feel like your subjects aren't basted and plastered enough, then let me know in the comments. I may end up releasing both, version A and version B. Just to give people options. Version A, in my opinion, gives slightly more mess than Version B. But with the down-side of occasional watermarking.Use with the Hunyuan Goonmaker workflow to achieve similar results: <a target="_blank" rel="ugc" href="https://civarchive.com/models/1315959?modelVersionId=1485540">https://civarchive.com/models/1315959?modelVersionId=1485540</a>

Plastered with cum

Trained on revised data set without watermarking. End result is slightly less mess, but slightly more stable generation patterning.

Files with SHA256: 67749fa97ff2e8c39a5811045db8554cef59feefe30a8e4831e9ba7825f48567