POV Squatting Cowgirl LoRA [1 MB] - CivArchive (CivitAI Archive)

POV Squatting Cowgirl LoRA [1 MB] - PSCowgirl

NSFW

This is another simple sex position LoRA for POV Squatting Cowgirl.

USE CLIP SKIP 2.

I suggest using this LoRA at 0.9-1.0 weight for best results.

Always add this prompt to activate it: <lora:PSCowgirl:0.9>, 1boy, penis, squatting cowgirl position, vaginal, pov

After that you can add other descriptive tags - If you remove penis from the prompt - It will give the effect of deeper penetration.

If you remove penis and vaginal - It will give the effect of her just sitting on you. If she is wearing a skirt or something similar - It will feel like it's implied sex - especially if you add cum.

If you add squatting to the prompt - It will show more of the position and it's a wider shot.

This was trained on NAI model - I have no idea if it works on realistic models. Images were generated on Corneo's 7th Heaven Mix using the Orange Mix VAE.

Embeds and LoRA used: Bad Prompt - Misty - Bea

Be sure to leave a comment, review and check out my other stuff here - you might like it.

Everything I'm making will always be free but if you want to show a little bit more support, you can buy me a coffee - https://ko-fi.com/ekune

Description

FAQ

Comments (41)

xpnvvxFeb 15, 2023

CivitAI

Thank you very much,and would you please make a lora of (extended downblouse)

btrockmanFeb 15, 2023

CivitAI

great one,thank you

JohnnyKnoxvilleFeb 15, 2023

CivitAI

I have no idea why, but tons of Loras I try and use throw errors in the Stable Diffusion console... They all say "Tuple index out of range"

ekune

Author

Feb 15, 2023

Not sure, brother. Never had the issue.

smsldooFeb 15, 2023· 1 reaction

Loras are generally either built on 1.x or 2.x. If you try to use a Lora from the wrong version (ie this on a 2.x model) it will throw that error.

trogly22859Feb 23, 2023

CivitAI

anyone have problems generating different color penises?

Larry_LePoopMar 2, 2023

CivitAI

I've liked your PoV LORAs so far; have you considered making a 'PoV Breast Grab'? I've had a surprisingly hard time getting ones that look good.

ekune

Author

Mar 2, 2023

I have actually.

alvinsu1029677Mar 6, 2023

CivitAI

Question: May I ask why every time I generate, there is no penis showing up? thank you！

ekune

Author

Mar 6, 2023

It sometimes happens - It heavily depends on the model. What model are you using and can you show me an example prompt? So I can try it, so see if there is anything I can add to prompt or remove to fix it.

185886008Mar 11, 2023

CivitAI

very wild model！

3247391426709Mar 22, 2023

CivitAI

It so cool

Freelancer604Mar 24, 2023· 1 reaction

CivitAI

not sure why my outputs keep generating 4 breasts lol, its like the upper torso is too long and it doesnt know what to put there

clevnumbJun 4, 2023

Shorten your vertical resolution...that also fixes double heads.

eleley123Apr 2, 2023

CivitAI

hi, your model is awesome. May i ask which checkpoint are you using

ekune

Author

Apr 2, 2023· 1 reaction

I was using AbyssOrangeMix2_Hard at the time.

DanseMacabreJesterApr 5, 2023

CivitAI

it is possible to remove the movement stripes under thighs?

maxwellcarey7777Apr 6, 2023

I have difficulties in installing these models and do inference on my computer, does anyone has any advices?

ZelinezApr 12, 2023

CivitAI

Hi, huge fan of all your pose-loras!

As a general question i was wondering if you could give some broad insight into the work you put into creating these?
Like how many pictures do you train on/steps/etc.?

Do you caption the pose itself for the training, or omit it? I've heard different approaches but yours seems the best afaik.

Do you create a "bad" lora from some pictures to generate more to then train a "good" one with those?

I'd be really happy to have some answers to these questions, especially by you, but also by anyone else that might have some more expertise than i do.

You're contributing to more amazing pose-loras coming from my end :P

ekune

Author

Apr 13, 2023· 2 reactions

https://rentry.org/59xed3#the-other-lora-training-rentry - I follow this guide.

I still check every once and a while to this day - they actively update it with new information and what settings/steps you need for everything.

ZelinezApr 13, 2023

@ekune Awesome, thanks a lot! :)

Could you maybe give one example of how you caption a specific picture to archive a pose? That would be super useful.

ekune

Author

Apr 13, 2023· 1 reaction

@Zelinez I use the WD14 Tagger Extension on Stable Diffusion to automatically caption all my pictures. I never use CLIP or Deepbooru.

It's usually accurate so I don't have to mess with the tags as much or at all most of the time.

ZelinezApr 14, 2023

@ekune Oh wow, I didn't think this autotagging would work for concepts (like poses) as many people say it's important NOT to describe the actual pose. This is all very helpful, thanks a ton!

ekune

Author

Apr 14, 2023

@Zelinez What alot of people do is use the auto tagger and remove any tags related to the pose - so in a way, they are right. I've done it before as well or sometimes I'll just leave all the tags - It works both ways.

Removing the tags basically makes it 'more' permanent in a way.

ZelinezApr 17, 2023

@ekune Interesting, what im trying right now is essentially to give it pictures that have or don't have the part that i want it to learn (so f.e. a couple pictures of the girl kneeling when im trying to train it to do squatting, just so it understands this is the part that is squatting in the picture - properly captioned of course)

Im definitely making progress, but my lora always seems to overfit the style a ton even though i don't want it to take the style into account at all.. I think my problem is that most my learning pictures have been generated in a similar style so it naturally will think that's what I' trying to do.

Is there any trick like a dampener just for style that could be applied here, or should i really just try to generate new learning pictures with different style loras so it can understand that's not what I want it to replicate?

ZelinezApr 23, 2023

@ekune I finally felt confident enough in my pose-LoRA to post it, haha. Man it took too much time, characters are so much easier..

In case you want to check it out:

https://civitai.com/models/48589?modelVersionId=53187

Hope it's ok to plug it here in the comments, if not just let me know - i'll delete this comment then.

Thanks again for the tips and links.

FxMx23Apr 29, 2023

CivitAI

Any way to change the height of the squat?

sgh617May 11, 2023· 9 reactions

CivitAI

Make a man have cowgirl sex while rubbing her tits!!

glowingariMay 16, 2023

CivitAI

Good LoRA :flushed:

claudshih2269782966Jun 25, 2023

CivitAI

Can I change from squatting to sitting ?

biros35293862Aug 13, 2023· 5 reactions

CivitAI

Is this only trained on huge anime tits? Does it only produce massive boobs? I'm looking for sex position LORAs that can accommodate amateur porn realism.

JaradisSep 2, 2023

CivitAI

This works really well on both Realistic Fantasy v2.0 and Realistic Vision v5.1, and my combined checkpoint of the two. Really well done. Would be nice if it also had a prompt where they would be on their knees riding and not always squatting with hands in front, but was able to do it with open pose of a reference image.

Chomp_CountryOct 10, 2023· 9 reactions

CivitAI

As I download newer LoRAs and am disappointed with the results time and time again, I frequently find myself going back to this LoRA over and over as a sort of tester (to make sure I didn't mess up a setting somewhere) because it always produces excellent and consistent results. And then I almost always end up forgetting about the finicky new one I downloaded to mess around with this one instead.

In short, this is the best NSFW LoRA I've ever used, and I've used a lot of them.

Aemeath1Oct 15, 2023

CivitAI

Will this model be revised? the generations are okay but its quite inconsistant in generating the desired posture

Aemeath1Oct 15, 2023

most of the time the hands are missing (behind back), model is yesmix

SakuragiKiriDec 26, 2023

CivitAI

这是我见过的给原图带来干扰最少的姿势LoRA，还有类似的优秀LoRA推荐吗

This is the pose LoRA that I have seen that brings the least interference to the original image, is there a similar excellent LoRA recommendation

dafebaJan 23, 2024· 9 reactions

CivitAI

hey man need an advice. I installed it correctly, use trigger words and lora itself in fooocus. with sdxl models, tryna make it realistic trying many checkpoints but it ain't workin anyway, what can be a problem?

challasvarias01106Mar 10, 2024· 8 reactions

CivitAI

Any chance to force the arms going on front on man's chest? 95% of the time the hands are behind.

Mumitroll95Aug 30, 2024

use pose and depth in controlnet

gamuhaji01233Feb 24, 2025· 4 reactions

CivitAI

guys, i need some help, i am using anything v3 and why everytime mine will not show penis :v

Vrtick25Aug 30, 2025· 5 reactions

CivitAI

Works fine, would recommend it for those who want to do NSFW related stuff

LORA

SD 1.5

by ekune

Download (Beta) View on CivitAI

Details

Downloads

203,616

Platform

CivitAI

Platform Status

Available

Created

2/14/2023

Updated

5/8/2026

Deleted

Trigger Words:

<lora:pscowgirl:1>, 1boy,squatting cowgirl position, vaginal, pov,

Files

PSCowgirl.safetensors

Size:

924.63 KB

SHA256:

39db07d5a50be7e0c8997c719509618016ecc8805a40e1578451f6da024e1c38