How to Make a Depth Map From a Photo (Free, No Sign-Up)

By Ismail Farooq · 29 October 2026 · 7 min read

To make a depth map from a photo, run it through a monocular depth model such as Depth Anything V2 and save the result as a greyscale image, where white is near and black is far. The free 3D Photo & Depth Map tool does this in your browser from one ordinary photo, with no sign-up, no upload and no watermark.

What a depth map is

A depth map is a picture of distance. Each pixel stores how far that part of the scene is from the camera instead of its colour. Shown as an image, it looks like a soft black-and-white version of the photo: a face in the foreground is bright, the wall behind it is grey and the sky is nearly black.

Phones with two cameras or a LiDAR sensor record depth when you take the picture; that is how portrait mode knows what to blur. An ordinary photo has no depth stored in it, so it has to be estimated. A monocular depth model does this from a single image, using the same clues you do: things that overlap, things that get smaller with distance, perspective lines, haze and shading.

How to make a depth map, step by step

1. Open the 3D Photo & Depth Map tool and drop in a JPG, PNG, WebP or HEIC photo. It works on a phone or a computer.
2. Set Make to Depth map (greyscale PNG). Choose Depth map (colour) instead if you want a picture to look at: far is deep blue and near is yellow.
3. Press Make it 3D. The first time, the browser downloads the Depth Anything V2 Small model (about 50 MB with WebGPU, about 27 MB on the slower WASM fallback). After that it is cached.
4. Download the PNG. Nothing has left your device: the model runs in the browser tab, and the photo is never sent to a server.
5. Flip it if your software wants black near. Open the map in Image Adjustments and tick Invert.

What you get from one photo

The map is never enlarged past the photo, so a 1200 px photo gives a 1200 px map. The model judges depth at a lower resolution than that and the result is scaled up and lightly smoothed, which is why edges look soft. That softness is usually what you want for displacement, because sharp steps in a depth map tear the mesh.

The PNG is 8-bit: 256 levels of grey. That is plenty for a parallax effect, fog or a depth blur. For strong displacement on a big mesh, 256 steps can show as faint terracing; a little blur on the map, or a lower displacement strength, hides it.

How to use the depth map in Blender

This is the usual way to turn a flat photo into a shallow 3D relief you can light and move a camera around:

1. Add the photo as a plane: Add > Image > Mesh Plane (called Images as Planes in older versions), so the plane has the photo's shape and the photo as its colour.
2. Give the plane enough geometry. Add a Subdivision Surface modifier set to Simple with a high level, or subdivide it in Edit Mode. Displacement can only move vertices that exist.
3. Add a Displace modifier. Create a new texture, set its type to Image and open the depth map PNG.
4. Set the Displace modifier's Coordinates to UV, so the map lines up with the photo.
5. Lower the Strength until the relief looks right. Brighter is nearer, so near parts come towards the camera with no inverting.

Keep the camera close to the original viewpoint. The photo only shows the front of things, so swinging far round to the side reveals stretched edges where the background used to be hidden.

Other things a depth map is good for

• Depth blur in video: After Effects' Camera Lens Blur can use a depth map as its blur map, so the blur grows with distance rather than in a flat band.
• Fog and atmosphere: use the inverted map as a mask, and fog thickens with distance.
• Parallax for a post: the tool's own 3D motion photo option does this for you, with four motions (gentle circle, side to side, push in, up and down) and three strengths.
• Relief carving and lithophanes: CNC and 3D-printing software can turn a greyscale map into height. Expect a soft relief, not a precise model.
• Other ImgVX tools that use depth: Tilt-Shift blurs by measured distance for the miniature look, Photo Relighter lights the surfaces facing the side you choose, and Sky Replacer uses depth to find the sky.

For a phone-style portrait blur, the Portrait Blur tool is the quicker path; it separates the subject instead of grading the whole scene by distance.

Which photos give a good depth map

If the colour map shows a mistake, such as a reflection treated as a hole in the wall, paint over the wrong patch in the greyscale map with a matching grey in Paint Tools. A dark or blurry photo gives a muddy map; lift it first with Low Light & Dehaze or the AI Photo Enhancer.

Browser tool, Blender add-on or a hosted generator?

Other routes to the same map are a Blender add-on that runs a depth model inside Blender (paid add-ons exist), or a hosted generator that uploads the photo to a server and usually sets a daily allowance. Most of them, this tool included, run a version of Depth Anything, so the maps look similar. The differences are where the photo goes and what it costs: here it stays on your device, and there is no account, no daily cap and no charge. The free AI image editing guide explains how in-browser models work and what their first-run download is.

Frequently asked questions

Sources: tool behaviour checked on the 3D Photo & Depth Map page on 29 October 2026 (output sizes, formats, model download). Model and licence: Depth Anything V2 (Yang et al., 2024), github.com/DepthAnything/Depth-Anything-V2, where the Small model is Apache-2.0 and Base, Large and Giant are CC-BY-NC-4.0.