LA RAIN is a photo-driven film study of how generative models reinterpret weather as mood and texture rather than physical force.

 

As plausibility increases, the image can stand in for reality—believability becomes a perceptual effect, not a record of what happened.

​

Using a consistent set of photographs and footages as input, I ran repeated generative model tests and curated the moments where “weather” shifts from physics to atmosphere.

​

Across iterations, the system tends to preserve tone (haze, glow, grain) while breaking causality (gravity, depth continuity, and physical rain behavior).