Hey bro, do a cumshot lora, plss

Is it possible to do the anal version, sir..?

Yup, another masterpiece. The subject’s face stayed the same in my test and worked as advertised.

Thank you.

Nice, Works perfecly!!!

Awesome, let's get an anal version

fucking legend

We have always trusted your LoRa models. I downloaded them without even reading the feedback. Your models are of the same high quality and consistency as always. I hope there will be more versions of LoRa with "sudden changes in scenes", such as riding sex, missionary sex, etc., because this is a new type of "reference image" that can be realized through LoRa.

The transition can be even faster, like the "Wan POV Missionary Insertion" lora, which can even take only 1 frame.

KING OF THE LORAS! WE MISSED YOU SO MUCH! DONT EVER LEAVE US AGAIN!

Cool, but it gives everyone large breasts... including the men doing the sexing. lol

edit: I am also getting a lot of body horror results every 1 out 4 or 5 gens or so, mutant bodies, extra genitals, deformed breasts, etc.

The GOAT is back

This model is excellent. The face remains unchanged, but I'd appreciate some additional body descriptions, such as (large/small/flat chested) and the amount of pubic hair. I tested it, and even if the original image showed flat chests and the prompt described them as flat chested, the generated video still showed larger breasts.

I'd also appreciate models for other poses.

Really unique and awesome idea for a lora! Great work! 😜

Looking forward to more poses. Great!

i am not getting any results, as the face just talks, using both low and high with 1 strength, i use the rapid aio I2V model, which is a single unified model for 2.2 , anyone outthere who can guide me some?

🐐

Incredible stuff, can't wait to see what you do next

niu bi!

Great idea, but like Sex Fov Slider, there's no control over the prompt, and it gives every generation the exact same body and pussy which makes all generations look exactly the same

would love a penis pop out of vagina lora

Another legendary Lora from you. You make some of the bests and more easy to use anywhere, really.

Recommended settings for steps with or without lightning?

The good: Does a decent job with likeness preservation (a very big deal). Good motion (also a big deal).

The bad: Forces pubic hair on 90+% of the generations regardless of what you do. Kind of decides what it wants to do with breast size/body shape regardless of what you input.

Still overall a win, but room for improvement as well.

Bro, you must be overdosing on fish oil — your brain’s too smooth with these LoRAs

Great idea, but it seems like the dataset lacks diversity. It always creates the same body type and same large breasts, ignoring how the girl in the original picture looks.

How to use this I'm getting tis message pLease help me anyone? [Outputs based on this image must be PG, PG-13, or they will be blocked and you will not be refunded. ]

I couldn't get it to work, (comfyUI, local)

If the subject is changing in your video then hook clip vision in node in your ComfyUI.

this is amazing. I hope there is an updated version of this with more trained version. This works great but is limited by one type of body.

Love it, works very well. Would like to see a cut scene to other positions, doggy style, and blowjob

I wonder why people train these. Isn't it easier to just prompt these 2 things or do i2v, and then splice them together in some editing program? Seems like extra steps to me.

Can you do an update with more body types and ethnicities please? Awesome lora otherwise.

This lora is fun and thanks for creating!

Has anyone found a way to prompt for different body type or titty type?

greatest of all time

you cooked again! Thank you for the great lora

they are all fat, why? is a bias from the dataset?
is there a way to prevent each leg to be as wide as my couch?

The transition dont work, the caméra just go down someone have the same problem?

Really fun concept! Would love to see more angles in future versions.

This is great. For anyone with character inconsistency I find the low noise lora on lower strength (0-0.5) + high noise lora on a little bit lower strength (0.4-0.7) helps a lot. Also allows for a little bit of room for interpretation in the transitions/POV.

This works really well!

For anyone wondering what the point is, this can take any image and make a video of that character getting fucked. This means you can use images that otherwise would be hard to "transition" into the action you want. The cut solves this!

Room for improvement: The guy doing the fucking is always white, despite prompting otherwise.

you are the best lora creator

Lmao, this is one of the best lora's for generating total bullshit! Congrats, made my day!

Try lowering the strength for anime, ~0.7 works good, idk the best value.
Also try ~0.85 for real images too for more consistency.

do these work with the lightning loras?

please do more angles of this

Amazing Lora as usual :).
Would love to see the same Lora but for anal :)

awesome lora. please expand on it so we can tweak it with the prompt! Such as different color dicks, different tit sizes, and also to keep it more accurate to their weight and size?

Amazing works as always.

Best Wan LoRA of the Year.

Man, all of your models 2.2 are fking fantastic and produce the best quality of all the ones i've tried, i'm just hoping for an insertion but no pov, Thanks for sharing this models

This should win a nobel peace prize.

Can someone tell me if the pubic hair is random or not?

Any tips to generate it?

Would love to see a cut to side fucking

this cut scenes are pure gold, please create more.

超级惊人的一致性，绝绝子！！！

i tried to add this to my resources when posting but it couldnt find it. wish i could add it so i can share my work with it.

Amazing lora, the best this year, hope you can do more

any workflow to execute this on 8gb vram laptop

What should be the model strength set at for high and low models, what step count and what's the cfg set to for high and low? cause i am getting it to generated but it's not coming out very clear

Versatility 10/10

Beginner-Friendly 10/10

Reality 10/10

Review: MVP Contender

A lot of my generations are leaving the face/head of the woman out entirely. Any ideas? What CFG/steps should I be using?

wonderful lora

make pov view pls i love it

This is just amazing work once again. May I please request a Nelson hold type?

This Lora is great. I learned that if you have two people in frame that you can prompt them go at each other if you specify the transition will be a wide shot.

Thanks! Does this work with the Lightning 4-step LoRA? I'm having trouble making it work on my Comfy setup. The resulting vids are just blobs that melt into each other...

I love your work, your lora's are always the best and works so well with so many different layouts. The details and quality is top notch, the shape of the penis and motion are perfect. The only thing is I wish you had the motion of pulling out. I try to make longer videos and use yours to start but at the end I want to have them pull out and match what was going in. Nothing comes close to the quality and always distorted or cutoff. Would love if you made a "pull out" lora. Keep up the great work.

how is this supposed to work? it cannot be as simple as a i2v and just the lora doing that?

BRO!!! where are you? i need you in my life!!! LMFAOOOO!

为什么我生成的视频，一转换画面，就看不到女人的脸，只有在做爱的下半身。

Look pretty good! Would consider making an anal version?

BEST LORA IN CIVITAI, SCORE: 100/10

dude this is what I've been looking for for months

i need to mess with it a bit but it seems a bit overtrained? at 1.0 faces that are not like the training data get changed after the smash cut. Overall wonderful lora though! Training for wan 2.2 is a pain so I appreciate that there are dedicated people making loras for it!

This lora is reallty top notch !
but how to custom body proportions ?
Seems like adding loras to change breasts, fatness etc do not work for me.

Nice lora, works really well a lot of the time, but every once in a while it produces cthuluian flesh horrors.

The view is always from bottom side on generated videos. Any suggestions? I'm using the default prompt and the official ComfyUI workflow for WAN 2.2. This happens all the time and for all generated videos, regardless of whether I'm using lightning loras or not.

Can you ejaculate on the breasts, mouth, face, and inside the vagina?

please make these loras with other poses, definitely with a doggy style, great lora

No control over body weight. All bodies tend toward chubbiness. Even very skinny builds become fuller. No realism. I ran hundreds of generations and got the same result across different strength parameters. This LoRA overrides the base model when it comes to weight. Unfortunately, it's unusable. There's a lot of work needed here.

you save me bro! no other lora can work such great performance like this in 8 varm!

Works great on rtx2060 vram 6Gb, ram 64, cpu ryzen5 5600x, workflow - wan2-2-i2v-fp8-Lightning (4 steps).

Generation time: 8 sec of video (512x712, long:101fr, 24 fr/sec, + multiplier 2 (RIFE VFI) ) - first_iter: ~850 sec. , next iterations: 779s, 735s, 688s

negativ: The character is trying to fiddle with something with his fingers, there might be a slight facial shift, extra hands.

But overall, it works fine!

please make this lora for different breast sizes , so it can be used for I2V also , cause this version makes everyones bigger than actual . thanX

this is brilliance. whatever you decide to lay upon us from this point is a blessing.

best lora i used, cause you can change or add other prompts. thx, bro.

Agreed with other comments here, this lora has too strong an influence on physique: thick to chubby women with large fake-looking breasts. Great if you like plastic pornstars, bad if you want natural women.

Great model. Ignore the idiots complaining about not being able to get their preferred body type. They are too stupid to use a proper input image or additional prompting or additional loras. The gallery shows that the limitation is them not the model.

Can anyone explain how to set the settings if you use WAN through Pinokio?

incredible stuff, I hope more is made

Great lora series! Although, I have seen the comment referring to tuning regarding women's traits. I normally don't have any issues with i2v for the women.

However, even if I prompt for example "a Black man enters the frame", I get a White guy 100% of the time, which tells me tuning was done that way. Any workarounds or solutions?

Otherwise, a great series of loras.

more please it;s awesome

how to make a lora like this？

@ Everyone Please provide tips for using this lora on civitai generator. I already tried on site but no good results. Thanks. PS Use the lora with these onsite images please. https://civitai.com/posts/6356416

This Lora is the most perfect Wan2.2 Lora I have ever used on Civitai, bar none. I am not overestimating it; on the contrary, I have a thorough understanding of Wan2.2. Based on my experience, I believe this is also thanks to the highest-quality training dataset.

10/10 LoRA concept, but...
Can you make another version of the model that works much more natively with Civitai? Or at the very least, make the trigger word not explicit because a majority of the time I get false flagged and won't go through. Like why can't you just make the tag simply <smashcut> There are LoRAs that only use singular trigger words instead of full phrases.

It does minimal mistakes and it's very consistent.

if it only produces body horror, try this prompt, this works regularly for me:

a woman is posing for a photo in (the location). <SM4SHCUT> The scene changes and a man is thrusting his penis into and out of the pussy of the same woman in the same location.

One of the best if not the best Loras for WAN 2.2. Really looking forward to more type of smash cut!

how do you even use this???
I put the Lora just before the two K samplers and used your prompt, only garbage pixelated resutls nothign really happens?

Using the original Wan2.2 14B i2v workflow, I replaces the loraloadermodelonly node with the lora node from here. I then moved the clips input to Load Clip and the output clips to the CLIP text Encode. Adjusting nothing else but KSampler steps, the video starts off fine, then becomes grainy.