Current tools fix barrel distortion — the curved edges. We go deeper: we reconstruct how near and far objects actually relate to each other. The way your brain sees it, not your lens.
Lenses compress or stretch the distance between objects. Mountains shrink. Backgrounds flatten. People at the edges warp. Your eyes saw something magnificent — your camera recorded something mediocre. Current software fixes barrel distortion — the curved lines. Nobody fixes depth.
In cinema, there's a dolly zoom — the camera pulls back while the lens zooms in, creating that surreal depth shift. We start where dolly zoom ends: we reconstruct how human eyes actually perceive the scene.
We build a 3D depth map of the scene — measuring the distance to every object in the frame.
Our core is mathematics, not AI guesswork. Tensor analysis and affine transformations re-project the image to restore spatial relationships the way human vision constructs them.
When we re-project a scene, geometry shifts — buildings tilt, horizons stretch, trees lean. But key objects in the frame shouldn't warp. Semantic segmentation identifies what matters and locks it. Everything else adjusts to frame it naturally.
Sjona works on images already taken — years ago, any camera, any lens. The depth reconstruction applies retroactively.
Corporate strategist and entrepreneur. Builds animatronics, props, and 3D-printed systems for TV and film productions. Understands the image pipeline from set to screen.
20+ years driving digital transformation and IT operations. Lifelong photographer.