Update v3
Fixes the freezed frames issue of earlier versions. Overall better quality, movement and composition. Can generate shoe removal and also a sock removal, though sock removal is little bit tricky and can generate some weird transitions. I recommend trying with 0.7 strength on high noise and increasing from there. Samples made with DDIM/Beta, 6 steps and lightx2v. Share your videos in the gallery!
Update v2 High Noise: Lower step count with smoother transition to pose. This works better with some starting compositions. Might need to prompt "both feet" or character lifts only one feet up.
Feet Up LoRa for Wan2.2 I2V
The aim of this model is to help with the lifting feet up pose, with the character sitting or standing and lifting feet up in the air or on a table. It can make either clean or dirty soles, just prompt it!
Samples made with lightx2v loras, 6 steps, ddim/beta
Share your videos!

Prompts:
Basic
A woman putting her feet up on the table and flexing her toes wide, revealing wide bare soles with deep wrinkles.Zoom out
Camera zooms out slightly to show a woman sitting behind a glass table, A woman putting her feet up on the table and flexing her toes wide, revealing wide bare soles with deep wrinkles.Dirty feet
A woman putting her feet up on the table and flexing her toes wide, revealing extremely dirty bare soles with black streaks and deep wrinkles*This LoRA is provided as is, without any warranty or guarantee of any kind. Use it only in lawful, ethical, and responsible ways. The creator does not endorse or accept responsibility for any misuse, including illegal, harmful, or harassing applications. By downloading or using this resource, you agree that you are solely responsible for its use and any outcomes resulting from it.*
Description
WAN 2.2 i2v High Noise V3
Better movement. Fix for freezed frames in the earlier versions.
FAQ
Comments (9)
How did you fix the freeze frame issue of the previous versions? I'm having a similar issue when I try to make i2v loras.
V2 almost cured my foot fetish with the monstrosities it created but V3 looks much better now. Thanks for your efforts!
V3 Low Noise is marked as an SD 1.5 model.
using the high noise can sometimes unrealistically bend the sole towards the camera for some angles, just using the low noise is really good for foot details alone as wan makes them look weird natively
I see yellow soles... Everywhere.
This is probably the best most consistent lora I've tried for Wan. Amazing work.
My generations on v3 just go from image to a ghostly blur effect until nothing is visible. Is there a specific workflow for Comfy ya'll use or maybe I'm missing something with the Wan2.2 official workflow.
The result itself is good but the result is incredibly sped up for some reason and I can't figure out why.
Thank you very much for sharing. The effect is really great! I have a problem. After adding high and low noise, my processing speed has become very slow. I don't know how to solve it.
Details
Files
wan22_i2v_feetup_V3_high_noise.safetensors
Mirrors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
fj_h.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
high.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
wan22_i2v_feetup_V3_high_noise.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.