I2V - Sex from behind (front facing) - CivArchive (CivitAI Archive)

I2V - Sex from behind (front facing) - v2.1_high_noise

NSFW

This lora is focused on intimate moments captured from front facing camera ideal for subjects facing the camera in doggystyle, plow, prone, or ¾ front positions, it can work from the side but no recommended. This LoRA's aim is to create accurate dynamic movement, as well more facial expressions:

Keywords, phrases:

→ Short thrusts, fast thrusts, slides in and out, penetrates
→ Fast, slow, rough
→ Pulls her hair, hands on neck

Workflow:

Here is a quickly organized and noted WF video. Save it and drag it into Comfy UI. https://civarchive.com/posts/25545461

Initial images:

For the V2.1: Starting images are a mixture of Z-image base checkpoint, and Pony CyberRealistic V15.0. Then I add do some enhancements with Zimage, Qwen, Qwen-edit to make them pop a little more.

The dark 'Raw' styled images are z-image and a style lora I'm working on.

Personal Settings:

Workflow posted above now, leaving this here for quick reference.

Settings for all V2.1 videos:

3 Advanced Ksamplers 8 step total, shift 5:

High noise, 2 steps, cfg:3 Euler, simple, add_noise: enable, return with noise: true, Without lightning lora.
High noise, 2 steps, cfg:1 Euler, simple, add_noise: disable, return with noise: true, High noise lightning lora.
Low noise, 4 steps, cfg:1 Euler, simple, add_noise: disable, return with noise: disable, Low noise lightning lora.

Description

Corrects over trained 2.0 version.
Adds penetrates as a keyword.

More training data with implied penetration to correct movement.

FAQ

Comments (35)

MMOFanDec 28, 2025· 3 reactions

CivitAI

This new version works quite well thank you for making it I’m leaving my old comment up only because people have taken the time to respond to it but as of right now I pronouncing this new version as of December 28, 2025 as AOK 👍🏻

The key is starting with a good image with the female in front and the man directly behind her on his knees

JellaiDec 29, 2025

Did you say that you're training an image lora to help with that? Is it Z-Image based?

MMOFanDec 29, 2025· 2 reactions

@Jellai No I use a different Text to Image workflow to get on image of female on all fours with naked man directly behind her, then use that image with the Lora.

YahulleyyyDec 29, 2025

how did yo uget it to woek, my characters in the right pose just move very slightly, barely any sex, strength 1 on lora

MMOFanDec 29, 2025· 1 reaction

@Yahulleyyy There is a penis LORA you can add to high and low that not only improves the look of the Wang but seems to make him thrust more aggressively.

YahulleyyyDec 30, 2025

@MMOFan yeh i had a Wang 8-) still no joy, ill keep trying

MMOFanDec 30, 2025· 1 reaction

@Yahulleyyy might just have a bad random seed, I swear to God I was trying to make one work today and it took me 20 tries I had to give up.

ShroomWombDec 29, 2025

CivitAI

is the version uploaded today different from the update yesterday?

daring_l

Author

Dec 29, 2025· 4 reactions

Sorry about the multiple notifications, I'm still new to releasing LoRas. I'll be more careful in the future but yesterday's release was overtrained. I removed it as the results weren't great. If it works well for you keep it and use it but todays will work better for more people.

ShroomWombDec 29, 2025· 8 reactions

@daring_l no need to apologize, you're the one doing something for free!

ron01468Dec 29, 2025· 2 reactions

CivitAI

Nice. Can it add the second person and do an insertion or is this more for animating a couple from an image where they are already in progress?

daring_l

Author

Dec 29, 2025· 2 reactions

If your starting image is setup in a certain way you might get something that looks close but all training data is "in progress".

Chp3286Dec 30, 2025

CivitAI

What are the initial images for these generating?

daring_l

Author

Dec 30, 2025

Models I used for the initial images in the lora gallery are from Pony CyberRealistic and Z-image.

ef4fJan 1, 2026· 3 reactions

@daring_l I THINK he means what's posture description

WhatTheGuyDec 31, 2025· 1 reaction

CivitAI

ahh to bad it's not T2V. The examples look great ;)

daring_l

Author

Dec 31, 2025· 2 reactions

I'll see what I can do.

thegriffinfamily3314710Apr 10, 2026

Just use a black image of your preferred dimensions as your starting image and it basically becomes T2V

WhatTheGuyApr 11, 2026

@thegriffinfamily3314710 from my experience loras that were trained on the I2V model aren't good at T2V. But haven't tried this one in particular

olympuscreativeApr 26, 2026

@thegriffinfamily3314710 never thought about that, ill give it some try, thanks for the tips

CyberAImaniaDec 31, 2025· 3 reactions

CivitAI

Can you pls share WF u used t generate this videos ?

daring_l

Author

Dec 31, 2025

Here is a quickly organized and noted WF video. Save it and drag it into comfyui. https://civitai.com/posts/25545461

asatru081Jan 3, 2026· 1 reaction

@daring_l not working. i get this message:
Unable to find workflow in 20251231_vid_00019.mp4

pratyushchandra1457Jan 25, 2026

There’s no workflow mate.

daring_l

Author

Jan 25, 2026

I just saved it again it works fine. But I uploaded it here as well. https://civitai.com/images/118675730

study520xMar 15, 2026

@daring_l i don't see where to use this lora

CyberAImaniaDec 31, 2025

CivitAI

Can you also please more explain this config:

3 Advanced Ksamplers 8 step total:

High noise, 2 steps, cfg:3 Euler, simple, add_noise: enable, return with noise: true, Without lightning lora.

High noise, 2 steps, cfg:1 Euler, simple, add_noise: disable, return with noise: true, High noise lightning lora.

Low noise, 4 steps, cfg:1 Euler, simple, add_noise: disable, return with noise: disable, Low noise lightning lora.

I just dont cant underfstand it.

Without fast lightning lora just 8 steps ??

Jusy 2 steps for high wan2.2 and 6 for low ?

daring_l

Author

Dec 31, 2025· 1 reaction

No sure which comment you made first but here is a quickly organized and noted WF video. Save it and drag it into comfyui. https://civitai.com/posts/25545461

MaxtorCivitaiJan 2, 2026· 3 reactions

CivitAI

Can we use a man coming from off the screen when is not in the image?

daring_l

Author

Jan 2, 2026· 6 reactions

I didn't plan on that use case. It may work in a tiny fraction of times but don't expect it to. I'll think about how to build that dataset for training.

R3G4LApr 20, 2026· 1 reaction

Can be done in SVI workflow....Use different LoRA or reduce strength for that scene, then switch back to the LoRA on the other parts etc etc.

omarpiosellidev541Jan 15, 2026

CivitAI

the best. but is there a method to increase fps?

bapybaraJan 29, 2026· 1 reaction

CivitAI

what is the best way to cut to this scene? are there loras that can help?

bapybaraFeb 2, 2026

@tombryder couldnt find it u have a link?

R3G4LApr 20, 2026

Use in SVI workflow.

LORA

Wan Video 2.2 I2V-A14B

by daring_l

Download (Beta) View on CivitAI

action

Details

Downloads

14,068

Platform

CivitAI

Platform Status

Available

Created

12/29/2025

Updated

5/13/2026

Deleted

Trigger Words:

sfb3hind, sfbehind

Files

sfbehind_v2.1_high_noise.safetensors

Size:

292.60 MB

SHA256:

e487de313c3c08383efd2dbb2b71175d380761746a4809737e156b66a1916d32