Usage of v1.0 and v1.1
Trigger words: She is lifting her tops, revealing her breasts and nipples Titty drop, flashing her breasts
Works with fp8, gguf variants, and FastVideo
LoRA strength: 0.5 - 0.8
Guidance scale: 9.0
Prompt: see the examples
Training details
See the "Training Data" file
Usage of v0 (not recommended for new users)
Trigger word: t1ttydr0p
Works with fp8, gguf variants, and FastVideo
This is still experimental and it seems not activated simply by the trigger word. You need to experiment with seeds and prompts to get the right result. Here is an example settings that I found it relatively working (also see the example videos and workflows):
num_frames: 41-61
seed: 665062624858284 (use this as a starting point)
Prompt:
t1ttydr0p, A professional video of a skinny Korean k-pop idol girl dancing on the concert stage. She lifts up her crop top shirt. After a second, she reveals her small breasts and nipples. She wears a crop top and a pleated mini skirt. In the background, there are other girls in the same group fully dressed. navel. Detailed smiling face with gorgeous makeup, high resolution.Sometimes, you need to reverse the video. I am not sure how to fix the training process to solve this issue...
About the example workflows of v0
HunyuanVideoWrapper (faster, recommended): Optimized for RTX 4090 or above (24GB+ VRAM) with sageattention and triton
Comfy: Uses a gguf and the official nodes (optionally you can use sageattention for faster generation)
Description
v1.0 comes with significant improvements:
Adhesion to the prompt got significantly better
No need to find good seed values anymore
Trigger words are changed to "Titty drop, flashing her breasts"
Notes
Recommended LoRA strength is 0.8. You will lose details if it is set higher.
The training dataset does not include many videos with full face. You may need to describe her face in the prompt.
This LoRA also changes the "default face" a little. I will try to fix this in the next version.
FAQ
Comments (31)
works decently. most of the times they start with their tits out and then put the top back on lmao
cool, if you don't mind my asking for longer form animations, how many frames are you training on?
Hello, you said it does not work with FastVideo but it does. First try I got a good video with only 6 steps.
Nice! I was in the middle of preparing a dataset for this exact subject. Do you have any advice for captioning videos? I've only done images up to now.
Awesome! Can you stack this with a character Lora?
can hunyuan video works with img2vid ?
Hi, I'm new to using LoRas. Is comfy the only place that I can use this at the moment?
NICE
Much more consistent, great job. It sometimes has some trouble with lifting up shirt showing odd straps underneath but it's way more likely to give a good result. I also noticed a lot of generation i do show a Nike logo even though i do not put that in the prompt, was some of training done with the Nike symbol present?
Thanks for including your training info and thoughts!
interesting work. thanks for including training details!
Great work, awesome update!
Great, one question: is this model faster than CogVideo img a video? I have a 12GB 3060 so I don't know if I can use it.
does it work without the keyword password? it's a lot of text to remember unless using the node that fetches the keywords online.
what tokens yields the largest boobs with this model/lora combo? (huge boobs:1.3),(gigantic boobs:1.3) or (huge breasts:1.3),(gigantic breasts:1.3)?
Has anyone faced the issue where only noise is generated for the video? I am unable to generate a video, only noise.
How do i prevent the splitting of the bra? She pulls it up but it turn in to 2 parts. Also i have some where its a complete tease. lol. She tries to flash but just plays with her boobs.
finally we are getting video loras
The only downside is that if the video is too long the drop reverses.
How is the LORA, with diffusion pipe or other? formed, what are the necessary PC capacities?
wish there was a video lora for boob flexing (like pec flexing but by a woman)
Does anyone have some good workflow to run this? I can't find anything working, constantly getting errors. What did you use? btw this is phenomenal work bro
Can you share where you got the material?
Seems to affect the face more than it should. Any chance of a new version that doesn't affect the face so much? (Not sure how that even works)...
I know I can try reactor or something, but trying to avoid that since I have character loras being used already.
try strength 0.33 you can lifted skirt :)
How do I achieve this effect? Are there any tricks or steps required?
would this work with Wan AI?
new in this version? faces were trained in too much in v1 so didn't use it, that fixed?
How do I do this? I have stable diffusion but have only made images, how to make videos like this?
wheres the training data file?
Sadly this won't be around for long, Civitai is taking all Tit drops down.