<h2 id="v1.0-wan-i2v-nktp9ib5m">v1.0 Wan I2V</h2>I discovered that WAN i2v performs exceptionally well on realistic people but struggles with animated characters. That's why I created this LoRA.I generated comparison samples without nitpicking (except for other single video), using the same seed and settings for comparison.Prompt: <code>animation of a topless girl shaking up and down, her breasts bouncing as she move</code>trained using Kohya repo : <a target="_blank" rel="ugc" href="https://github.com/kohya-ss/musubi-tuner">https://github.com/kohya-ss/musubi-tuner</a>

Files with SHA256: 117546443977e67f471bd698d3b2b2969123ea340160d989fab3849ccb7d918b