Update Version 2 Information
This release introduces an enhanced version of Lora. Unlike the previous version, which primarily focused on selfie generation, this new iteration is capable of producing both selfie and non-selfie images. It generally operates well alongside many other Loras, although there are a few exceptions where conflicts may arise. However, most Loras tested have shown compatibility.
To improve model performance, the term "Cum" has been omitted from captions. This decision was made based on the hypothesis that avoiding unfamiliar terms could enhance the model's learning of specific details.
Trigger Words:
Main Trigger: "WhiteLiquid" on her face (Note: no space between 'White' and 'Liquid').
Additional Triggers:
"WhiteLiquid" on her forehead
"White Liquid" on her clothes
"White Liquid" on her body
"WhiteLiquid" in her mouth
These triggers have been carefully selected to optimize response and accuracy.
Version 1
Cum Selfie XL By MakeThemComeAliveAiArt
"or in this case make them cum alive"
The model does what the title says, all you gotta do is add: "Selfie shot of ... (Woman, Name, Whatever) with cum on her face"
Disclaimer: This LORA is designed for creating NSFW (Not Safe For Work) content and is intended for use with fictional characters only. Please do not combine it with any real person or engage in any abusive activities. The creator of this LORA does not assume responsibility for any misuse and strongly encourages responsible and ethical use.
Description
As requested a smaller version of the V2.0 model. I would still recommend V2.0 but this one also seems to perform alright.
FAQ
Comments (10)
What refiner XL is recommended?
I didn't even use refiner with most the examples only the under water one I believe. I am not sure whick refiner as I use Mage.Space because my computer is to old for Stable Diffusion to create my images I am not sure which refiner they have build in.
@Makethemcomealive Ah, that explains it. The online processors are so much better than local, in my experience.
Man I want this but 870mb is rough.
I did train on lower but the quality just isn't that good, and with this version I was aiming for details
But next version I will aim for smaller
I can appreciate quality > size
@ethburner479 I still have the data set so I can do a rerun if I have the time in the next few weeks on lower dim, but do expect quality to be a bit less
Smaller version uploaded =)
@MakeThemComeAliveAIArt Very nice!

















