Updated for Wan 2.2. I'm pretty happy with the results of this one. The movement seems decently high speed, even when using the lightning loras. It's likely even better without the lightning loras, but I haven't checked. All of the previews were created using the lightning loras. The workflow I used for the samples can be downloaded by using the training data download.
Here's the important part of the prompt:
A man is {standing|kneeling} in front of her positioned between her legs thrusting his penis back and forth in her vaginaDescription
FAQ
Comments (37)
what tool or website do u use for training wan loras?
I use https://github.com/tdrussell/diffusion-pipe. Check out my POV missionary one for more details on the training process.
could u post a WF,thx
Download the training images, it's really just a json workflow.
thanks! if possible 1.3B I2V needs some love too ^^
Does the i2v 1.3B model exist?
Hopefully a doggystyle one soon? T_T. Love your work!!!
I haven't been using the t2v for WAN very much after initial testing. Is there any particular reason you've created this for WAN t2v rather than Hunyuan? I.e. any benefits?
All of the Wan Loras I'm making I had already created for Hunyuan. I'm just doing it for both. Wan has the advantage of having good I2V capabilities.
Wan i2v is much better than Hunyuan, already far behind. Quality of trained open source models is moving fast. Don't stay attached to one for too long. Wan2.1 is much more "close" from KlingAI than Huynuan was.
@SD_AI_2025 Yeah, shortly after posting this I got Sage and Triton installed. I can actually generate quickly and have been using WAN a lot. It's great
Great job bro. All ur lora have been working very nice on my workflow. Im looking forward to a creampie lora to combine with this one. Keep up the great work. Love it !!!
where is training images?
I don't really provide the actual videos I trained on for various reasons, but if you're just curious to see my method and settings, you can check out the training data I uploaded to https://civitai.com/models/1331682/wan-pov-missionary.
Is this working with I2V?
I don't know, I have yet to try I2V. My other loras seem to work, based on people's comments, so it seems like this one should too. Let me know if it doesn't.
Yes, it is
Gives no errors when used with I2V, I'd say it works
Same training data as with your Hunyuan model? The Hunyuan Lora has more intense movement than the WAN one. I'm wondering if that is a general WAN thing or just unfortunate settings for WAN...
Yeah, it's hard to say. The one difference in training data with Wan is I include more still images to help it understand what it's supposed to look like in HD. Without it, things are very blurry and low res looking. But I wonder if it also affects movement somehow. I would hope it's smart enough not to let still images affect movement though.
Overall it feels like Wan and Hunyuan have different strengths and weaknesses. I'm not sure which one I prefer overall.
Is this lora for the 480p or for the 720p or works with both?
It was trained on the T2V model. As far as I know, it should work on every 14B wan model, including the I2V ones.
Thank you for your contribution to the best sex position and perspective known to humanity
has anyone seen how good this is prompting for different positions with a lower strengh?
YW!
Why my video is static? https://civitai.com/posts/15217877
I used OP prompt with a little modification
I'm shameless, but could you please consider making a mating press?
Am I the only one getting results that using this lora makes the resolution of the output a little bit more blurry? I've been trying every loras related to 'nsfw movements', and this has been the best in describing it so far, but lowering resolution thingy happens only in this lora...
Yeah, I think I understand the problem, but it's taking a long time to go through and fix the problem with some of my loras. For now, try to lower the weight of the lora as much as you can without ruining the movement.
@dtwr434 I think I've found a solution. When using Wan Video Wrapper nodes, connect 'WanVideo Lora Block Edit' node to the Lora Select node. Set block0-19 as T / 20-39 as F. The resolution is much improved.
Still, I'm not quite sure if this method harms the effect of lora. But I hope it's okay.
Its just the quality of the training inputs i imagine. The lora is great, you just have to run it at lower strength. Combining it with other nsfw loras really helps guide it since the strength is lower. Somewhere around .6-.75 works well for me with this lora.
This Lora is super versatile, can be used in a variety of positions to good effect .
Really good job !
My current favorite 👌
Side view missionary? More like side view missionary, side sex, and doggystyle, really any man on top position. Amazing work, really incredible.
@dtwr434 please update Base Model of your lora from Wan Video to Wan Video 14B i2v 480p so your lora can appear in new wan generator
This is a great lora for all side to side motions in my opinion, missionary or otherwise.
Your stuff is the best. Don't know if there's others that look better, but yours are the hardest to screw up!
Details
Files
wan_missionary_side.safetensors
Mirrors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
i2v_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
124_wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
mssn_npv.safetensors
side-view-missionary.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
wan_missionary_side.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.