About

About CameraControls

I'm Miles Donovan. Before I wrote a line of web software, I spent six years programming motion-control camera rigs — the computer-controlled cranes and dollies that indie films and product shoots use when a camera move has to be repeatable. Same arc, same speed, same framing, take after take. A shampoo bottle orbiting under studio light doesn't improvise, and neither did my rigs.

So when AI video tools arrived, I was excited — and then immediately frustrated. Every one of them asked me to describe the shot in words and hope. "Slow dolly in" might give you a dolly, a zoom, or a pan that drifts into a wall. Ask twice and you get two different moves. I'd spent years in a world where a camera move is a set of numbers you can write down, replay, and hand to the next shot. Prompts don't work that way.

The turning point was when image-to-video models started accepting real camera trajectories — actual keyframes with azimuth, elevation, and distance, the same grammar a motion-control rig speaks. But that capability lived only in developer APIs and ComfyUI nodes. No product put it in front of the people who actually need it: creators turning one product photo into a 360° orbit, or one portrait into a bullet-time shot.

CameraControls is that product. Upload one still, place keyframes on a timeline, and the scene holds still while the camera moves — exactly the way you plotted it. Presets like Orbit, Dolly Zoom, and Crane Up are just pre-filled keyframes you can retune. And because a trajectory is data, not a lucky prompt, you can run the same move across a whole product line.

A few things I hold this product to:

• Control beats novelty. If you can't repeat a shot, it's a demo, not a tool.
• Honest credits. Failed generations refund automatically, credits never expire, and paid renders include a commercial license.
• Camera grammar, documented. The blog and shot pages explain why a move works, not just which button to press.

— Miles Donovan, Founder & AI Cinematography Researcher