I created this lora to animate flux and sdxl images of women in showers and under waterfalls. While base hunyuan would animate the character and/or the shower/waterfall, it did a poor job of adding water content to the character. This lora attempts to add dripping, running, splashing water content/animation to the women characters in all the 'right' places.
Testing: i have gotten great results from hunyuan t2v. sadly, i am unable to get similar results from i2v which was the main focus of this lora.
Dataset/Training: 13x75 frame 24fps clips 1024x1024, trained at 256x256 for 2720 steps, 28 epochs
Description
initial version, trigger word is 'sh0w3rt1m3', additional training words, shower, showering, dripping, wet, dripping water,
FAQ
Comments (10)
looks quite good
thanks, unfortunately it doesn't work for i2v...
@tedbiv the real question is, is anything work on hunyuan I2V anyway ? I have yet to see any good resul of hunyuan I2V myself, compared to Wan at least...
i've had a couple of good i2v. mostly that don't involve a lot of movement...here's one with my roundabout lora. https://civitai.com/images/68413178
@NoArtifact https://civitai.com/images/68407415
@tedbiv Yeah it's decent, some of the best i've seen as far as hunyuan I2V is concerned, 95% of my attempts on those I2V hunyuan was straight to the trash, you did a better job than me for sure.
@NoArtifact don't get me wrong. those are the best of about 100... when it hits, it hits nice. mostly they are shit.
works very well !
glad you like it, post some of your creations...
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.