For on-site generation, I recommend only using models designed for WAN2.1 or earlier. It does not appear that Civitai can get 2.2 models to work. You can use the Universal model with both high and low noise steps, but if you want the best quality, I included both versions as well.
This LoRA is trained on conversational language. You can describe your scene in plain English or Chinese, and it should adhere to your prompt well. Below are some inspirations for you if you need it.
Positions:
On her back
Kneeling
Perspectives:
POV
Enhancement phrases:
She bounces her body up and down
She bounces her breasts up and down with her hands
All showcase videos have the workflow metadata embedded. You can download the video and drag it into ComfyUI to replicate them with their prompts, seeds, and settings.
Description
More training, new datasets. No trigger word needed, just makes boobs bouncier.
FAQ
Comments (23)
So glad to see an updated version for on-site 720p. Great lora, thank you!
Your all purpose LoRA is my favorite LoRA. Imagine my delight to see new versions of it! :D
Been doing some limited testing as I haven't had a lot of time.
Results are great, still a lot of issues with penis shape changing, hit and miss. (I guess it cant gen what it cant see.)
Tried some penis LoRAs but I feel it doesn't improve much.
Bounced some tests with a few new workflows I'm testing. Some of them have weird penis things happening but thought I would upload anyway.
Going to test with old workflow again and I'll update.
Much appreciated for your work. Thumbs up.
Thanks for the feedback! I don't have any decent penis datasets to train off of, and WAN doesn't have very much in its base model, so it's definitely something that's lacking. I'll see if I can get a large dataset and train something into it to improve. Do you have any examples you can share so I can track progress?
As a separate statement: This LoRA is all purpose still. I uploaded it 3 times so that people can use it with the CivitAI generator, but it's the same exact LoRA for I2v and t2v.
@ComfyTinker Oh it's one of the best LoRAs in my opinion. ;)
You can see by my posts.
I imagine it's difficult anyway as the breasts hide the penis most of the time and after that it seems to just draw lines from the top to the bottom.
Still, sometimes it creates magic. ;)
From my testing it works really well in two situations, small breasts (where most of the penis is visible), or insanely big breasts (where you can pretty much just see the tip) and it doesnt move up too much.
I'm still testing.
I appreciate the work mate. Solid.
Chucked you a tip too my man. Legend.
Hey, you've got a good amount of different versions. Which one would you recommend for Image to video for the 480p wan video model?
If you're generating locally, just download the all-purpose model. They're all the same file, just uploaded under multiple versions so people can use the on-site generator.
the titfuck itself works great but I can't seem to get cum to look good at all. for some reason no matter how much I specify white and translucent, it comes out as this weird olive oil-colored substance and it's kinda hilarious
try using a c-msh0t lora and describing the c-m as milky white
Has anyone managed to get this lora to have the woman on her back with the man straddling over? No matter what prompt I use, it just won't do it. (Using with character lora if it makes a difference).
this is the best one ive used so far. the only other one thats close is POV Titty Fuck -T2V & I2V by playtime_ai. that one's TF movements are faster and more of what one expects BUT you have to use the high lora and it gives alot of blur(even at 0.5). this one doesn't, i even set it to 2.0 strength at 5sec, 8 steps and it works great. the interesting thing is this lora has alot of mouth movement and the girl is always talking lol
Playtime_AI "enhances" other people's LoRAs. I remember comparing tensors between my LoRAs and theirs awhile back and noticed their is a lot more overlap than there should be if they trained from scratch. LoRAs are not copyrightable, so nothing wrong with that, but I don't know that they are curating their datasets perfectly.
Which, um, @playtime_ai_ lora can do titfucking? Did I miss one? Playtime, you know about this?
@ComfyTinker You could not be more wrong. ALL of my models are trained off of my own curated datasets. I have never and would never use anyone else's model as a starting point or to "enhance".
@chesthole Do I know about my own Titty Fucking lora? I sure do! It is not on the site right now... My original account got taken down. If you go to my huggingface, you can find most if not all of my old models.
Don't know what am I doing wrong but it was REALLY difficult to spawn a penis while using this... I only saw the girl grabbing her tits and moving, but no penis between her tits. After 5 tryouts it appeared and was a bad quality dildo-looking penis.
Not using any other Lora only this one.
I'm currently working on a much larger and more broadly trained LoRA for this on a personal datacenter.
This LoRA was originally trained on the movement and included a lot of simulated titjobs/titfucking on dildos. Also, since the penis is often obscured, it impeded training. I'm hoping future models have better male genitals, but for now please be patient :)
please make one for ltx2!
This model is putting out weird stuff recently. What changed?
On-site generation has not been very reliable. Are you doing local generation or on-site?
@ComfyTinker I was using on-site, but it has become completely unable as of late. I moved to local and I am seeing much better results. Thanks for the response.
Details
Files
BetterTitfuck_v4_July2025.safetensors
Mirrors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
T2V_BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors
BetterTitfuck_v4_July2025.safetensors