In short: pick photos with a clear subject in front of a distant background, use Zoom in for most shots, keep motion strength moderate (40–70%), and give each photo 4–6 seconds.
What parallax is
When you move your head, near objects shift more than distant ones. That difference — parallax — is one of the strongest cues your brain uses to perceive depth. A flat zoom (the "Ken Burns" effect) moves everything together, so it reads as a flat picture moving. Parallax moves layers at different speeds, and suddenly the picture reads as a place.
How AI estimates depth
VDOAPP uses Depth Anything V2 (small), a neural network trained on millions of images to predict how far each pixel is from the camera — from a single photo, with no special hardware. It learned the cues humans use: objects lower in the frame are usually nearer, distant things are hazier and smaller, edges overlap in a certain order.
With a depth map, the editor warps the photo on your graphics card for each frame of the animation, shifting near pixels further than far ones. The depth map is softened slightly so edges don't tear, and the picture is scaled up a touch so the frame's borders never show. The model is about 26 MB and is downloaded once.
Three motions
| Motion | Feels like | Use for |
|---|---|---|
| Zoom in (3D dolly) | Walking towards the subject | Portraits, products, dramatic reveals — the safe default |
| Pan across | A sideways tracking shot | Landscapes, streets, group photos |
| Orbit | Circling slightly around the subject | Hero shots, food, objects on a table |
Motion strength sets how far the virtual camera travels. Above about 70%, gaps behind the subject become more visible; 40–60% looks natural on most photos.
Choosing the right photos
- Great: a person or object clearly in front of a distant background — portraits outdoors, a building across a square, a mountain behind a hiker.
- Okay: indoor scenes with some depth, food shots with a table leading away.
- Poor: flat subjects (a document, a wall, a screenshot), aerial photos looking straight down, and scenes with fine overlapping detail such as tree branches against the sky.
Portrait blur
The same depth map powers Portrait blur: the nearest subject stays sharp and the blur grows with distance, like a wide-aperture lens. VDOAPP measures the depth range of each photo and keeps the front part — the whole person, not just the tip of the nose — in focus. Combine it with Zoom in for a cinematic still.
Step by step
- Add a photo and select it on the timeline.
- In the Clip tab, set Show photo for to 4–6 seconds. Motion needs time to be noticed.
- In the AI tab, under Photo depth, choose a 3D motion and adjust Motion strength.
- Optionally add Portrait blur, then preview. Try a Fade or Blur transition between 3D photos.
Tip: alternate motions in a slideshow — zoom, pan, zoom, orbit — so the rhythm doesn't become predictable, and cut on the beat of your music.
Frequently asked questions
Does this work on video clips?
No — 3D motion and portrait blur are for photos. For video, use the Blur background look in Behind the subject.
Why do I see a smeared edge behind the subject?
That's the area the photo never captured, filled in by stretching nearby pixels. Lower the motion strength or choose Zoom in, which reveals the least.
Is the photo uploaded?
No. The depth model runs in your browser.