For more LoRAs and updates, Join my Discord:
https://discord.gg/ZVWVhT43GW
You can deploy MiniMax on RunPod with my template: https://get.runpod.io/minimax-template
Follow me here so you see the next one.
Recommended settings:
T2V/I2V:
Turbo LoRA - 8-steps version at strength 0.20
Cumshot LoRA - strength 0.9
Sampler - Euler
Scheduler - ddim_uniform (Optionally include the ExtendIntermediateSteps node and add 2 steps starting at sigma 0.85)
R2V:
Turbo LoRA - 4-steps version at strength 0.85
Cumshot LoRA - strength 0.9
Sampler - Euler
Scheduler - beta (Optionally include the ExtendIntermediateSteps node and add 2 steps starting at sigma 0.85)
V0.5:
Expanded dataset (56 videos) with more angles
A thorough description of the cum texture helps a lot!
V0.1:
A video LoRA for the cumshot itself: cum leaving the cock, travelling, and landing on her face/tits/pussy/sister/father in law
This is mostly a demo as I am testing different LoRA training settings, dataset is very slim and limited to 1 POV position, but it does other position as well.
I will get to training another version with a more robust dataset in the future.
Prompting
It was captioned in plain, blunt language, so write prompts the same way:
hmcumshot3, he shoots a big load of thick white cum from his cock in five thick
spurts. She is a petite blonde with her eyes shut and her mouth open. This pov
shot shows her kneeling on a dark wood floor.Related
HMPussy and HMPenis are anatomy adapters. This one is an action and it stacks on top of them.
Description
Larger dataset with more angles, slightly better cum texture, but this is mostly prompting.
FAQ
Comments (8)
Good work on these dude we all appreciate it. Thank you for not doing that bs with the early access for a bagillion buzz too <3
Just to add an edit after using it. I've noticed half the cumshots disappear mid shot. I've trained some loras for wan and stuff and the key is that in your dataset make sure none of the videos have incomplete cumshots in them. That's the only way to get rid of it
Thank you!
The physics work great but it glitches the face on the whole video when face is included in the prompt.
it's odd, using fl2va but with the reference workflow, character references lose likeness, but the lora works and looks good with the extensive description. however, switching backto the ref model in the reference workflow, the likeness is fixed again, but the lora produces weird substance, not even white and more like a slushy ice cream, until you actually reduce the extensive description to something simple like : thick semi-translucent white cum , and not overcomplicate. so depending on which workflow/model, the lora handles differently. So you can make it work in both, just need simple vs complex descriptions depending on the model. Even if it takes a little experimenting to tweak, by miles one of the best loras so far.
v0.5 looked bad at first, but then I copied the whole paragraph of the cum description as you said, now it's absolutely awesome!
Thanks for the feedback, prompting is incredibly important!
@HearmemanAI you should display the prompt requirements more prominent in your model description! I see in the comments that a lot of people don't realize how to prompt when trying out the Lora for the first time
I really wanna get this working but it just looks like a bucketload of paint gets launched at the girls face, I tried making my prompt more specific like translucent w/ weak trajectory, etc but no dice
For reference I2V
Turbo 8-step w/ 1.0 strength (tried others)
Epic Cumshots @ 1.0 strength
Used good verbiage to make sure its really weak cum, still insane toothpase look
Anyone have a workflow I can take a peak at? Loving H3 but trying to get it to play nice, maybe its my WF