Bite that lip!
Wan 2.2 I2V v1
Definitely happier with this run than the last one. Still recommend highest resolution you can manage. Also, I had really terrible results using SmoothMix checkpoint. Not sure why. Best results I have seen are using the base wan. Test videos use Q6_K.gguf models. Also, I recommend messing with the lightning lora weight. I got better stability using 1-1.5 weight on lightning lora, or 2-2.5 if using other loras for sex acts for more speed/movement.
Wan 2.2 I2V v0.1
I'm seeing better results at higher resolution. So 720x1280 seems to be the best. Lower resolutions have sometimes worked, but only if the face is large and the primary focus of the image. My guess is that in the compressed latent space, the lips are just too small at lower resolution for the model to focus on, much less for the Lora to affect. So if you need small resolution, make sure the lips are big in the image reference.
I have a version of the core workflow I use here:
https://civarchive.com/models/2076123/wils-style-workflow
Description
v1.0 - Illustrious
Challenges:
-Checkpoint dependent. Only model I've found so far is WAINSFW does well. Feel free to experiment.
-Style shift. For it only working on wai, it adapts the style pretty hard. So it doesn't generalize well.
Tags:
--biting own lip
--seductive
--confident
--mischievous
--hesitant
--worried
FAQ
Comments (3)
PornworksAnimeDesireNSFW checkpoint also seems to handle this lora just fine!
that said, it REALLY wants to increase the size of the subjects head. At lower values 0.7-ish its fine, but the higher the value, the bigger her head becomes.
At 1.2 the head is already 1.5 times the standard size.
To deal with style shift and in general I usually put face detailing loras inside the adetailer prompt box, this lora also works very nice like that and one can up the strength quite a bit too.
Simply awesome. This should be in every model.