I2v and T2V Lora for spanking action
example of prompt:
1) T2V
integrated_multimodal_description: [Shot 1] spa_nk, The scene is set indoors, likely in a bedroom or dressing area, with a woman positioned in the foreground. The camera is framed from a low angle, focusing on the woman's lower back and buttocks. She is bent over a white desk with a black frame and a textured, possibly tiled, front panel. The woman is wearing a light blue, sleeveless t-shirt that is pulled up, exposing her bare backside. Her skin is covered in a pink, viscous substance, likely a lubricant or gel, which is smeared across her buttocks and is visibly dripping down her thighs. Her blonde hair is visible at the top of the frame. To the left on the desk, a red bottle and a black object, possibly a hairbrush or tool, are present. In the background, a doorway is visible, leading to another room. The lighting is bright and even, illuminating the entire scene clearly. A man's arm and hand enter the scene from the top left. He is wearing a light-colored, long-sleeved button-up shirt. His hand is positioned near the woman's back, and he appears to be holding a black object, possibly a hairbrush or a similar tool. The pink substance remains on her skin.
overall_soundscape: Sspanking sound
non_diegetic_music: na
2) I2v
For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.
subject_definitions:
<Subject 1> is the woman in <Picture 1>. She has long, wavy, dark brown hair, a radiant smile, and wears a light-colored, distressed open-front jacket over a dark garment, with a “FIA 2023” lanyard around her neck. Her vocal timbre is warm, natural, and expressive. <Subject 2> is the setting: an indoor or semi-enclosed venue with a dark gray door, white latch, and faint circular logo above — suggesting a public event space.
summary:
The sequence begins aligned exactly with <Picture 1>. <Subject 1> walks away from the dark gray door toward a flat-topped table and bends forward at the waist. She lifts her jacket and dark garment, exposing her underwear and bare buttocks. A man's left arm enters from the left side of the frame holding a black, flat stick. He raises the stick high and swings it downward rapidly, forcefully and repeatedly striking her bare buttocks. spa_nk. The camera remains steady as the impacts continue.
retention_analysis:
<Subject 1> (present throughout): fully_preserved - matches <Picture 1>.
<Subject 2> (present throughout): fully_preserved.
detailed_description:
The sequence begins aligned exactly with <Picture 1>. <Subject 1> walks away from the dark gray door toward a flat-topped table and bends forward at the waist. She lifts her jacket and dark garment, exposing her underwear and bare buttocks. A man's left arm enters from the left side of the frame holding a black, flat stick. He raises the stick high and swings it downward rapidly, forcefully and repeatedly striking her bare buttocks. spa_nk. The camera remains steady as the impacts continue.
overall_soundscape:
The audio consists of a series of sharp, violent, percussive slapping sounds, followed by a low, guttural groan. The man says : take this. The woman makes sound of pain.
non_diegetic_music:
None.
Description
FAQ
Comments (21)
Great work! Hope you keep refining it.
Small point of clarification: your I2V prompt example is technically for Ref2VA. I2V still utilizes the T2V model / prompting.
(Although, technically, you can use Ref2VA prompting with the FL2VA model; you just have to switch nodes to make references work.)
Oh. I am sorry. I have a lot to learn.
Would appreciate more of your feedback and direction , if possible
No need to apologize; you didn't do anything wrong! I was just making you aware that I2V is still technically the same as T2V in terms of prompting. You can read about it officially here: https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_base_en.md
And the official reference prompting, which you used for your I2V, can be found here: https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_ref_en.md
But like I said, it's not a big deal, because you can technically use them interchangeably with either model if you swap a single node. It's just best practice.
You're doing great. This is one of your first LoRAs and it's pretty good! Just keep learning.
@drowai443 are you saying you could encode the prompt and single image with the ref2v encoder while using the flv model? Like what "single node" are you referring to
@makiaeveli Yes, you just swap the MiniMax H3 Image to Video node for the MiniMax H3 Reference to Video node and then you can use references with FL2VA (for much better quality but at a slight hit to reference adherence). Or, perhaps even easier, just use a standard REF2VA workflow and swap the model to FL2VA. Same thing. It's pretty well known the FL2VA model is superior quality, even acknowledged by the MiniMax H3 team.
@drowai443 wow that's cool, I wouldn't even have thought to try it
Given that the previous similar model has disappeared from the website, I hope everyone will download it promptly upon seeing it.
Thanks for H3
havent tried but I bet it pairs with the ass jiggle lora, the collision one
very poor trained
Can you share your version, better trained?
@nnewtoai719 no mine is mine fuck you
@insanelime1012rk321 face it, you're a dick.
@insanelime1012rk321 waiting for you to share something better :-) Then we see who has it longer?
wonderful, going to make a great convo with whip one and fight one
is the fight lora available somewhere? could you please share the link?
@nnewtoai719 https://civitai.red/models/2853878/minimax-h3-combat-base-fight-motion-impact-drama-booster?modelVersionId=3246572 that one is cool, they always erase loras with weapons and combat, that one survived for now
which one is whip lora? )
@gonfinal12 Thank you for the link!
I might try to make a wrestling lora of some sort.
Have to see if I am capable to do it lol
Great, thank you! I get better results with face slaps with this than the other lora.
Fantastic! If you please share videos :-)