Depth Parallax

Parallax is just depth times movement

Move your head sideways and the things near you sweep across your view while the horizon barely shifts. The amount anything appears to move is its displacement divided by its distance — that is the entire phenomenon, and it is why parallax reads as depth so immediately that no one has to be taught it.

Give a flat photograph a depth value for every pixel and that arithmetic becomes available. Each pixel is displaced by its depth relative to a chosen focal plane; things in front of the plane move one way, things behind it the other, and the plane itself stays put. What you get is not a filter that suggests dimensionality — it is the correct shift for a viewpoint that moved.

Disocclusion is the whole difficulty

Move the camera and you start seeing surface that was hidden behind the subject a moment ago. A photograph contains no record of it. It was never captured, and nothing can recover it — this is not a limitation of the implementation, it is a limitation of photographs.

So the only question is which guess is least visible, and there are two honest ones. Stretch smears the edge pixels sideways across the hole, which is invisible against soft or textured backgrounds and obvious against a hard vertical edge. Background pulls a farther neighbour in to cover the gap instead, which holds up better where the background is busy and can drag the wrong thing across where it isn’t.

Neither is right in general, which is why both ship. And both have the same ceiling: the further you push the shift, the more hidden surface you demand, and eventually the guess becomes the picture. A restrained shift on a scene with real depth separation looks uncanny; a big shift on a flat one looks like a mistake. That ceiling applies to every 2.5D parallax method, here and anywhere else.

It carries no motion of its own

Depth Parallax displaces to a viewpoint; it doesn’t animate between them. The movement comes from the Motion stack — arm the shift there and you get the living-parallax loop, with the timing, easing and loop length under your control rather than baked into the effect.

That separation is deliberate across the app: an effect says WHAT, the Motion stack says WHEN. It also means the same parallax setup can be driven fast for a glitchy judder or slowly for the drifting depth loop that reads as a photograph breathing.

When you’d reach for it

For a still that has to move — a header, a loop, a post that needs to stop a thumb. And for stills too: a modest parallax shift is a genuinely three-dimensional way to recompose a photograph, which no crop can do.

Every option, from the app

Generated from the registry the app publishes, so this list is whatever the current build actually ships — not a copy of it.

Fill

StretchBackground

Related effects

• Depth of FieldA per-pixel circle of confusion computed from the depth inferred out of your flat photo, gathered through a configurable aperture — polygonal, catadioptric or anamorphic bokeh.
• Point CloudEvery pixel lifted into real 3D space by inferred depth and rendered as billboarded splats with true depth-buffer occlusion — six displacement fields, four materials, three environments, and a field channel another effect can drive.
• MetaballDepth points scattered into a volumetric density field and raymarched as an isosurface — blobs that genuinely fuse where they meet, with capsule connectors, cast shadows and ambient occlusion.

Open the app to try it; the free effects work with no signup. Open V3KTR · all the FX · how it works