How to make a photo move with real depth

Last updated 2026-09-01 · 8 min read

What is the short version?

A slow push into a photograph is the standard move and it always looks like a photograph. Separating the subject from the background and moving them at different rates makes the image read as a space with something standing in it.

1. Go to Vizard Agent and upload the photograph.
2. Say the mood — warm, cinematic, restrained.
3. Give the music, or describe the feeling it should carry.

What do you need before you start?

One good photograph and a piece of music. Vizard Agent separates the subject from the background on its own, so the real requirement is simply a picture with a clear subject in it, rather than a crowd scene where everything happens to sit at the same distance from the camera.

• The photograph. High resolution, clear subject.
• The music. Or its mood and tempo.
• The length. Thirty to sixty seconds.
• The words. Any text that should appear.
• The format. Vertical, usually.

What do you type into Vizard Agent?

Describe the feeling you are after rather than the camera move you imagine. Vizard Agent derives the moves from the music's own structure, so telling it the mood and the occasion produces a noticeably better result than specifying a pan direction and a speed.

Prompt

Variants worth knowing:

• Ask for depth. Otherwise you get a Ken Burns pan.
• Name the occasion. It sets the music and the pace.
• Ask for text behind the subject. It reinforces the depth.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real memory film built from a single photograph. Steps twenty and twenty-one are the technique: the subject comes out as its own layer, and everything after that depends on it.

1. Inspects the uploaded photograph and checks for any other media.
2. Checks the available effects, including the text-behind-subject recipe.
3. Probes the photograph's dimensions.
4. Searches the music library for tracks matching the mood and the tradition.
5. Downloads the candidates along with the photograph.
6. Detects the beats and tempo of both music candidates.
7. Analyses each track's mood and emotional arc before choosing.
8. Extracts a slice and detects its musical beats precisely.
9. Finds the audio onset peaks and measures the slice's loudness.
10. Extracts the subject as a cutout for the depth parallax layer.
11. Views the extracted cutout and checks its alpha channel.
12. Searches for an elegant serif typeface and inspects the source and cutout together.
13. Tests the first shot's camera move and renders it, then reviews the start and end frames.
14. Renders the second shot's move and reviews its frames.
15. Creates the typography overlay and renders a shot with the text behind the subject.

Step thirteen is the discipline. Each camera move is rendered as a test and looked at start and end before it goes into the film, because a parallax move that drifts too far exposes the empty edge behind the cutout.

What does the result look like?

The film this page is written from was built from a single portrait: several shots, each a different camera move over the same image, cut to a track chosen by analysing its mood and beat structure, with serif typography and a warm vignette.

The depth is the thing you notice without being able to name. The subject moves at one rate and the background at another, so the frame reads as a room rather than as a picture being zoomed into.

When does this not work well?

Parallax reveals whatever was sitting hidden behind the subject, and in a single photograph what is behind them is nothing at all. Vizard Agent keeps every move within what the image can actually support, rather than travelling far enough to expose the gap.

• Big moves. They reveal the empty area behind the cutout.
• Crowded photographs. No clear foreground to separate.
• Hair and fine edges. The cutout shows at the boundary.
• Low resolution. Moving into it exposes the pixels.
• Flat compositions. Nothing to separate in the first place.

How do you fix a result that came back wrong?

Say which shot is wrong and what is wrong with the move in it. Vizard Agent keeps the subject cutout, the full beat map and every rendered move test, so a gentler move or a different running order is a re-render rather than a rebuild from the photograph.

• "I can see the edge behind her." Move reduced in range.
• "It drifts too fast." Re-timed against the beats.
• "Put the text on another shot." Overlay moved.

How does Vizard Agent compare to doing it yourself?

By hand this is a cutout, a patched background plate and two layers animated in a compositing tool, per shot. Most people give up and use a zoom. Vizard Agent builds the layers and tests each move before using it.

Common questions

How many photos do I need? One is enough. Vizard Agent builds several shots from the same image.

Why not just zoom? Because a zoom looks like a photograph. Vizard Agent separates the layers so it reads as space.

What makes a good photograph for this? A clear subject against a background with depth. Vizard Agent will say when an image is too flat.

Will the edges look clean? Mostly. Vizard Agent checks the cutout's alpha channel and shows you the result.

Can it add text? Yes, and behind the subject. Vizard Agent layers the typography between the planes.

How does it choose the music? By analysis. Vizard Agent measures each candidate's mood, tempo and beats before picking.

Can I supply my own track? Yes. Vizard Agent detects its beats and builds the moves around them.

How long should it be? Thirty to sixty seconds. Vizard Agent uses several moves rather than one long one.

Can it use several photos? Yes. Vizard Agent gives each one its own move and cuts between them on the beat.

Does it work on old scans? If the resolution allows. Vizard Agent limits the move range on softer images.

What is actually being revealed behind the subject? Whatever the photograph does not contain, which is why the moves stay modest. A large parallax move needs a filled background plate, and that is a different and more visible kind of invention.

Can I get a vertical and a square version? Yes. Vizard Agent re-frames the moves for each format.

Does it check its own shots? Yes. Vizard Agent renders each camera move as a test first and reviews its start and end frames before that shot goes anywhere near the finished film.

Can it add a vignette? Yes. Vizard Agent builds a warm cinematic vignette texture and lays it over the whole film.

Does the music have to be licensed? Yours to check, as always. Vizard Agent searches libraries and tells you where each track came from.