A grayscale image in which brightness stands for distance: the nearer part of the scene is lighter and the further part darker. It records the shape of a scene and discards its colour and texture, which is what makes it usable as structural guidance.

Is this metric depth? Does it give real distances?

No. All three estimators return relative depth — how near or far parts of the picture are compared with each other, not how many metres away anything is. ZoeDepth is metric-oriented and comes closest, but treating any of these outputs as a measurement will give you wrong numbers.

Which estimator should I choose?

Depth Anything for most photos, and it is the default. MiDaS when you want the planes of a scene clearly separated rather than smoothly graded. ZoeDepth for rooms, architecture and 3D work where the relative scale matters most. The cost is the same for all three, so trying a second is cheap.

Can I use the output as a ControlNet depth input?

Yes, that is what most people generate one for. The map is a standard grayscale depth image and works anywhere a depth conditioning input is expected. Because the result opens in a generation session, you can wire it into a model here without downloading it first.

Can I do this for a video?

Not on this page — use the video depth map tool instead. It is a different model built for footage, and the difference matters: running a still-image estimator frame by frame produces depth that flickers, because nothing ties one frame's reading to the next.

What image formats can I upload?

JPG, PNG and WebP, one image per run. The depth map comes back as a standard image you can download or pass straight to another node.

Is my photo processed in my browser?

No. Some depth tools run entirely on your device; this one does not. Your photo is uploaded and processed on our servers, which is what allows the same three models to run regardless of the machine you are on. If keeping the file off a server matters more than the choice of estimator, a local tool is the better fit.

AI image depth map generator

AI Image Depth Map Generator: Turn a Photo into a Depth Map

Upload a picture and get a grayscale depth map back — brighter is nearer, darker is further. Pick the estimator that suits your scene, then use the map as structural guidance for generation, or as a displacement source for 3D and parallax.

Depth estimator

The general-purpose default. Start here.

2 credits

Street photograph receding to a distant junction

Depth map of the same street from Depth Anything: a smooth near-to-far gradient, parked cars picked out as pale masses, sky black

Examples

Depth Map Examples

Depth Anything — the default
MiDaS — flatter far separation
ZoeDepth — strongest contrast

How it works

Create a Depth Map in Three Steps

Step 1
Upload a photo
Drop in one JPG, PNG or WebP. Scenes with real distance in them — a street, a room, a view — read far better than a flat object on a plain backdrop.
Step 2
Pick an estimator
Three models, three different readings of the same scene. Leave it on Depth Anything if you are not sure; the section below says when the other two are the better answer.
Step 3
Download or keep working
The map opens in a generation session, so you can download it or wire it straight into a model as depth guidance without starting again.

Choose a model

Which Depth Estimator to Use

All three return relative depth — near versus far within one picture. None of them measures real-world distance, whatever the numbers look like.

Depth Anything

General-purpose relative depth that holds up across people, objects and outdoor scenes. The default, and the right answer for most photos.

MiDaS

Strong relative-depth layering, indoors or out. Reach for it when you want the planes of a scene clearly separated rather than smoothly graded.

ZoeDepth

Metric-oriented depth aimed at rooms, architecture and 3D spatial work. Closest to a real-distance reading of the three, though still not a measurement.

What it is for

What People Use a Depth Map For

ControlNet depth conditioning

The most common use. A depth map gives a generation model the shape of a real scene, so a new image keeps the spatial arrangement while its appearance changes completely.

Parallax and 2.5D motion

Displace a still along its depth channel and a flat photo gains a sense of volume when the camera moves across it.

Compositing and depth of field

A depth channel is what lets you fog, grade or blur by distance rather than by hand-drawn selection.

3D and spatial media

Depth maps drive displacement in 3D pipelines and feed the spatial photo formats that headsets read.

Credits and plans

Each depth map costs a flat 2 credits, whatever the size of the photo or the estimator you pick.

Basic

For hobbyists and explorers

$9.99/mo

Select Plan

1000 credits
Up to ~200 images/month
At least ~1000s video/month
Intelligent creative agent
Video editor
Latest image models, including GPT Image 2.5, GPT Image 2 and Nano Banana Pro
Latest video models, including Seedance 2
Realistic face uploads
Voice generation with ElevenLabs
No watermark
Unlimited template access

Pro

Max

For power users and teams

$149.99/mo

Select Plan

18000 credits
Up to ~6000 images/month
At least ~18000s video/month
Intelligent creative agent
Video editor
Latest image models, including GPT Image 2.5, GPT Image 2 and Nano Banana Pro
Latest video models, including Seedance 2
Realistic face uploads
Voice generation with ElevenLabs
No watermark
Unlimited template access

Team

For teams and studios

$329.99/mo

Select Plan

36300 credits
Up to ~12000 images/month
At least ~36300s video/month
Realistic face uploads
Share canvas, workflows, and assets with your team
Up to 10 members per team
Up to 5 teams
Role-based management
Team credit management and spending caps
Per-member usage tracking

See all plans

FAQ

Depth Map Questions

What is a depth map?: A grayscale image in which brightness stands for distance: the nearer part of the scene is lighter and the further part darker. It records the shape of a scene and discards its colour and texture, which is what makes it usable as structural guidance.
Is this metric depth? Does it give real distances?: No. All three estimators return relative depth — how near or far parts of the picture are compared with each other, not how many metres away anything is. ZoeDepth is metric-oriented and comes closest, but treating any of these outputs as a measurement will give you wrong numbers.
Which estimator should I choose?: Depth Anything for most photos, and it is the default. MiDaS when you want the planes of a scene clearly separated rather than smoothly graded. ZoeDepth for rooms, architecture and 3D work where the relative scale matters most. The cost is the same for all three, so trying a second is cheap.
Can I use the output as a ControlNet depth input?: Yes, that is what most people generate one for. The map is a standard grayscale depth image and works anywhere a depth conditioning input is expected. Because the result opens in a generation session, you can wire it into a model here without downloading it first.
Can I do this for a video?: Not on this page — use the video depth map tool instead. It is a different model built for footage, and the difference matters: running a still-image estimator frame by frame produces depth that flickers, because nothing ties one frame's reading to the next.
What image formats can I upload?: JPG, PNG and WebP, one image per run. The depth map comes back as a standard image you can download or pass straight to another node.
Is my photo processed in my browser?: No. Some depth tools run entirely on your device; this one does not. Your photo is uploaded and processed on our servers, which is what allows the same three models to run regardless of the machine you are on. If keeping the file off a server matters more than the choice of estimator, a local tool is the better fit.