This is a lora test for split screen for groping. Trained with a very small dataset so it is very limited in freedom. This is V1. Might work on a V2 if people think it is worth the time. It should work for both t2v and i2v I2V Prompt example: integrated_multimodal_description: [Shot 1] The video presents a split-screen view of a single person. On the left, a young woman with long dark hair and bangs is visible from the chest up, wearing a white collared shirt and a blue and white striped tie, with her eyes closing at every touch for pleasure and her head tilting up and down in synchronization with the moaning expression of her moving mouth. On the right side of the screen, a close-up shows a man's hand actively lifting the woman's skirt and then inserting his hand directly inside the woman's underwear for gripping, rubbing, and sliding up and down against her pussy iside her panty with continuous, deliberate physical movement, tou_ch, creating a dynamic and engaging action sequence. The lighting on the left side is brighter, while the right side is and focused intensely on the movement. The camera is dynamic on the overall split-screen composition and the left and right screens are related in movements.overall_soundscape: The audio consists of the rhythmic clatter of a train moving on tracks, accompanied by a low, continuous hum, and the woman voice in pleasure.non_diegetic_music: na

Files with SHA256: 9e30eec8ad35b49cc757b499c9cf6e0a0ec0eb7ffe4435146318a0a3be8b072b