MF Hip Hop
self project · music driven ai motion study
music: Onyx, "MF Hip Hop" · concept, ai & compositing: ZEDZARA · 2026
this étude started from sound rather than image. the only rule was simple and strict: let the music decide. Onyx's "MF Hip Hop", heavy, gritty, shouted from a New York stoop, dictated everything that followed: what the world looks like, how it moves and who lives in it. the track's weight became the camera, pressed low to the asphalt and bent through a fisheye lens. its grit became the image itself: monochrome, grain, damp concrete and overcast streets. and its swagger became the people: figures with exaggerated proportions, oversized hands and boots, baggy denim collapsing over heavy soles, bodies that tower over the lens the way the beat towers over the room. nothing in the frame is decorative; every element is there because the music asked for it.
the world was built in Midjourney, with GPT Image as an unexpected collaborator: together they pushed the characters' proportions further than either could alone, until every figure carried the same attitude as the track. subway cars, deli corners, skateboards, a pack of BMX riders, a hand trailing out of a car window. fragments of a city seen from the curb up.
motion was the real test. video models are trained on how bodies normally move, and they resist anything else: a few frames in, they quietly correct the anatomy, shrinking hands and boots and pulling the bodies back to balance. shot after shot drifted toward average, until Seedance 1.0 held the exaggeration from the first frame to the last. After Effects completed the picture with compositing, grain and texture, so the footage reads as if shot through a dusty lens rather than generated.
no brief, no client. just finding out how far the tools can be pushed before they push back.
concept, design, ai & compositing: ZEDZARA
music: Onyx, "MF Hip Hop"
tools: Midjourney, GPT Image, Seedance 1.0, After Effects
non commercial personal study. music © its respective owners.
Instagram: @zedzara_
this étude started from sound rather than image. the only rule was simple and strict: let the music decide. Onyx's "MF Hip Hop", heavy, gritty, shouted from a New York stoop, dictated everything that followed: what the world looks like, how it moves and who lives in it. the track's weight became the camera, pressed low to the asphalt and bent through a fisheye lens. its grit became the image itself: monochrome, grain, damp concrete and overcast streets. and its swagger became the people: figures with exaggerated proportions, oversized hands and boots, baggy denim collapsing over heavy soles, bodies that tower over the lens the way the beat towers over the room. nothing in the frame is decorative; every element is there because the music asked for it.
the world was built in Midjourney, with GPT Image as an unexpected collaborator: together they pushed the characters' proportions further than either could alone, until every figure carried the same attitude as the track. subway cars, deli corners, skateboards, a pack of BMX riders, a hand trailing out of a car window. fragments of a city seen from the curb up.
motion was the real test. video models are trained on how bodies normally move, and they resist anything else: a few frames in, they quietly correct the anatomy, shrinking hands and boots and pulling the bodies back to balance. shot after shot drifted toward average, until Seedance 1.0 held the exaggeration from the first frame to the last. After Effects completed the picture with compositing, grain and texture, so the footage reads as if shot through a dusty lens rather than generated.
no brief, no client. just finding out how far the tools can be pushed before they push back.
concept, design, ai & compositing: ZEDZARA
music: Onyx, "MF Hip Hop"
tools: Midjourney, GPT Image, Seedance 1.0, After Effects
non commercial personal study. music © its respective owners.
Instagram: @zedzara_
scenes
left on the cutting room floor
stills that never made it into motion.
stills that never made it into motion.
