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
FAQ
Comments (28)
I don't know if that's just me, but I prefer using the txt2vid even for img2vid... The txt2vid ones look much more natural, the img2vid always look like fake titties even if you have a saggy breasts image
The txt2vid version (3.0) was a continued training, and the I2V was a fresh training on a new dataset. I'll give the I2V a look and see if I can retrain it to give it more fluid motion. Thanks for the comment!
@ComfyTinker And thank you for your hard work!
add the bouncing boobs lora
Great work
Can you post this on tensorart?
Any advice on how to not have the penis head stuck? Seems no matter what i do if the penis head is visible in the intial frame it get stuck and causes unnatural morphing. Tested both t2v and i2v loras on various different strengths with zero difference.
Can you share your prompts @StableVibrations ?
Typically adding "Moving her breasts up and down dramatically" or "Engulfing the penis" would be my first thing to try.
@ComfyTinker
tbh tried basically every iteration of the example prompts i could think of.
last attempt:
She moves her breasts repeatedly up and down, stroking the mans penis. the penis is sliding between the soft flesh of her breasts.
any recommended cfg/shift values? currently running cfg:4.5 / shift 7.5 @25 steps
also to clarify a bit: the general motion of the titfuck is there, but the tip of the penis stays stuck in the top of the cleavage
@StableVibrations Are you not using the term "titfuck"?
@StableVibrations Also, Hunyuan or Wan? In ComfyUI locally?
@ComfyTinker Sorry for taking so long to reply, i've had to do some sooul-searching
it's wan i2v
It seems like it's the specific 4.5/7.5 combo that was the culprit, changing it to 4/7.5 works alot better
Latest models don't show up in the video upload page because of the new filters, ie: you have "Wan Video 14B t2v" as the base model but only models that say "Wan Video" show up when picking models/loras, if you try to upload something using this model you'll see what i mean.
Civitai released more granular model options, but those aren't usable with the on-site generator yet. I can't change it back, so we will have to wait for CivitAI to fix it. Civitai has been a mess lately, and I don't make any money from this. You can still download them to run locally.
Great work, but this is a titjob not titfuck. We need a version where the breasts are stationary and the penis moves.
Only getting small d!cks from this lora, however big, huge, massive, enormous, or giant I am prompting for it. I've seen posts with huge dicks below, what is the trick ? For some of them its not even in the prompt... ?
It's likely that a lot of them are using Img2Vid. I didn't train this LoRA on large or small dicks.
One of the best Loras in this website!
I2V or all purpose is still not available after Civitai's recent breakdown. The on-site generator can't find it for either Wan Video 14B i2v 480p or Wan Video 14B i2v 720p.
Can you re-upload the lora?
very nice
Just realised I have never left a review for this LoRA.
In my opinion one of the best. Much appreciated for your work.
The hit to miss ratio on this is on the high end of hit. Most gens come out good to great first time.
Sending some buzz.
<3
can't anymore use it on website generator ? :c
(In img to vid, wan 14b i2v 480p or 720p)
worked before in 480p
I have the same problem, cant add it as additional resource when using i2v. Only possible for t2v.
@Arag0rn CivitAI keeps screwing with its on-site generator settings. Not sure how to help on this one.
@ComfyTinker I think they're asking you to change your WAN 14B All Purpose "base model" type from Wan t2v to Wan i2v in order to use it in the on-site generator, I would also appreciate this.
@cranpeach Ahh, I'll upload my latest version as t2v and i2v separately.
Does this work with WAN I2V 720p or is it only 420p? Can't seem to get any movement on 720p.
You're commenting on the HunYuan version of the LoRA. The WAN version should work for both 480p and 720p. Can you share your prompt that isn't giving you movement?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.