Try if you want, but I am not sure this lora helps the generation of h3 for the same scene.
This is a testing lora.

It works for both I2V and T2V

Example of prompt:

grated_multimodal_description: [Shot 1] A white woman with shoulder-length straight black hair sits upright on the floor directly facing forward in front of a white leather sofa with two large plush black fur throw pillows against a pale turquoise wall. She wears a tight-fitting orange top and a thick black spiked choker. Immediately to her right at face level, a man is kneeling or standing close beside her, holding a tool directly aligned with her face. From the nozzle of the tool held tightly in his hand right next to her, a heavy jet of liquid pssMB shoots horizontally forward, colliding directly with her chin and splashing outward aimed precisely toward her open mouth. Her eyelids remain partially lowered, brows drawn together in concentration as droplets accumulate along her chin and upper lip line. Facial muscles tense intermittently with slight puckering motions. The camera hovers just below eye level, capturing the close spatial alignment between the man's tool and her face.

overall_soundscape: silence except for gentle bubbling noises emanating softly from the device held in-frame by the man, dynamic background dialogue, shifting liquid stream sounds, gagging, and ambient indoor noise.

non_diegetic_music: na