BEHIND THE WALL This is my first Contra challenge, and my entry for the CapCut Seedance 2.5 Video...BEHIND THE WALL This is my first Contra challenge, and my entry for the CapCut Seedance 2.5 Video...
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Peter's avatar
Peter Manda
• Aug 19
BEHIND THE WALL
This is my first Contra challenge, and my entry for the CapCut Seedance 2.5 Video Challenge.
I wanted to create a one-take that feels more like a directed fashion film than a sequence of AI-generated spectacle.
The film begins with a woman driving a black 1967 Pontiac GTO through a mountain landscape on her way to work. The camera moves around the car, pushes into the rear-view mirror, rises into space, then drops into Times Square as she arrives in the city.
She exits the car and walks toward a blank poster mounted on a building wall. As she begins to enter the poster, the camera continues independently into the hidden world behind the wall: a working fashion photography studio.
Inside, we see the next stage of the same journey — the shoot itself. The camera moves through the active studio, revealing the model being photographed, the crew, lighting, styling area and production environment.
The camera then returns to the same exterior wall, where the blank poster has become the finished campaign image created during the shoot.
The idea was to build a visual loop: we follow her through a stylized version of an entire working day — the journey to work, her arrival, the photoshoot itself, and finally the result of that work appearing back in the city. She enters an empty image, the camera reveals where that image is created, and the film ends on the completed campaign poster.
WORKFLOW
My usual workflow is different. I normally generate multiple separate shots with as much control as possible, refine them individually, and then manually edit them together.
Creating an entire 30-second sequence as one continuous generated take was therefore new territory for me.
For this project, I first built the film through controlled pre-production: character and vehicle sheets, exterior and interior environment sheets, a complete storyboard, and around nine anchor frames defining the key compositions and transitions.
These references were designed to control character identity, vehicle consistency, architecture, framing, spatial continuity and the camera path across multiple environments.
I then used a structured 30-second Seedance prompt to connect those pre-designed moments into one uninterrupted moving shot: mountain road → space → Times Square → poster → hidden photography studio → final campaign poster.
Working this way meant giving up some of the frame-by-frame control I normally have, but that was also what made the challenge interesting. It was a fun experiment and a useful experience in directing a generative model across a much longer continuous sequence.
Created with CapCut Video Studio and Seedance 2.5.
contraseedancechallenge
Guoshuai's avatar
Guoshuai Zhang
• Aug 19
Really polished. The last few seconds are especially strong 🔥
Peter's avatar
Peter Manda
• Aug 19
Thanks, that really means a lot - the ending was actually the core concept of the whole film, so I’m glad that part landed.
Back to feed
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started