This is a Lora with enhanced walking posture for wan2.1/2.2, which can make the generated walking video more smooth and natural.
This LoRa has been updated to Wan 2.2, and three versions will be released this time,Greatly optimized the problem of small steps in Wan 2.1 version:
Lite: Training with fewer steps, avoids issues with holding a phone or causing wrinkles, but the walking posture isn't quite perfect.
Slow: Training with more steps, occasionally causing issues with holding a phone, but the walking motion is more sensual. Videos generated at a 16-fps frame rate will appear in slow motion.
Normal: Training with more steps, occasionally causing issues with holding a phone, but the walking motion is more sensual. Videos generated at a 16-fps frame rate are at normal speed.
i2v version updated to v3.0
Improved the issue of slow motion when characters walk, fixing it to a normal walking speed
i2v version updated to v2.0
This version has optimized the stability of walking, avoiding strange walking movements and adding some hand movements
The i2v_v2.0 version also adds some hand movements, such as:
adjusted her hair with handOr
hands on hipsI tried using Wan2.1 to generate videos of characters walking behind their backs but was unsuccessful. There were always instances of people walking backwards or making strange movements. So I trained this Lora, which will improve the success rate of generating walking back videos.
Trigger word:
walking from behindPerhaps increasing the number of sampling steps can lead to improvement
Description
FAQ
Comments (8)
Include some example prompts in your videos please.
I'll give you the prompt I wrote. This may be different from the prompt the uploader intended, but it worked well.
walking from behind. An anime-style illustration of a young East Asian girl walking away from the camera down a narrow street, her long, black hair swaying gently. She wears a white T-shirt and black athletic shorts with white trim, paired with black and white sneakers. The street is lined with plain, light-colored walls and utility poles, with potted plants adding a touch of greenery. The camera follows her from behind at a medium distance, capturing her full figure against the backdrop of a typical urban residential area in broad daylight, rendered with smooth Japanese cel-shading.
No need long prompts for I2v. Most of time, i only use the trigger 😂
You can download my example video and drag it into Comfyui. The video contains prompt and workflow
@ivever DUH! I forgot about that! Thanks.
@ivever How exactly would I use your example workflow for t2v instead of i2v? All of the examples for both models appear to be i2v.
@IcebergMM I have uploaded some T2V examples, but the effect is not good
@ivever that only works if you have the kaiji wrapper custome node, the text encoder is blank. I use normal comfyui workflow. You should add the prompt example
Details
Files
P002-The-Walking-Back-i2v-v10-000010_converted.safetensors
Mirrors
P002-The-Walking-Back-i2v-v10-000010_converted.safetensors
P002-The-Walking-Back-i2v-v10-000010_converted.safetensors
P002-The-Walking-Back-i2v-v10-000010_converted.safetensors
P002-The-Walking-Back-i2v-v10-000010_converted.safetensors
P002-The-Walking-Back-i2v-v10-000010_converted.safetensors
P002-The-Walking-Back-i2v-v10-000010_converted.safetensors