anything i show is always text to video, mostly synthetic images around 5000 - 500 real images - this is an beta version test but it still took around 32 hours of training, the next version is 27,000 real images, so it will take much longer. but - this works so i share
it you you want sexual actions, it might not work (the motion not the visuals) - use image to video loras like sexgod's stuff to it and it works, where as on its own sexgod loras dont work well text to video
this video was when the project was at 23k steps, i havent tested it since. - i moved onto the other lora. - should be slightly better then whats currently in the video.
