What a 2D tracker does and why you need it for 3D object placement

A 2D tracker in Nuke follows a flat, moving point in your video footage — like a person's hand or a car's bumper. Once Nuke knows where that point moves frame by frame, you can attach a 3D object to follow it. The tracker watches the 2D motion on your screen, and the 3D object rides along with it, staying locked to that spot even as the camera moves or the background shifts.

This is different from manually keyframing a 3D object's position. Instead of clicking to set the object's location on every frame, the tracker does the work for you. You point it at something in your footage that moves predictably, and it generates the motion data automatically. The 3D object then inherits that motion.

You need this when your footage has camera movement or when you want a 3D element to stick to a moving target. If your shot is locked on a tripod and you just want to place a 3D object in a static spot, you can skip tracking. But most real-world footage has some movement — a pan, a zoom, or an actor moving across the frame — and that's where tracking saves hours of manual work.

Key Takeaways

  • The 2D tracker node in Nuke watches a specific point in your footage and records its movement across every frame.
  • You attach a 3D object to the tracker's output, and the object inherits the tracked motion automatically.
  • Choose a tracking point that has clear contrast, stays visible throughout your shot, and moves in a predictable way.
  • After tracking, you link the 3D object's translate values to the tracker's output to lock the object to the tracked point.
  • If tracking fails partway through, you can manually adjust individual frames or restart the track from a different frame.

Setting up the tracker node and choosing your tracking point

Start by loading your footage into Nuke and creating a Tracker node. Plug your source footage into the Tracker node's input. In the Tracker node's properties panel, you'll see a viewer window showing your footage and a small square — that's your tracking region.

Move that square over the point you want to track. The best tracking points have high contrast: a corner of an object, a button on a shirt, a distinctive mark on a wall. Avoid flat, featureless areas like a plain white wall or a person's cheek. The tracker needs texture to lock onto. If your footage is soft or low-contrast, the tracker will drift or fail.

Adjust the size of the tracking region so it's just large enough to contain your chosen point plus a small margin of surrounding detail. Too small and the tracker loses context; too large and it picks up unwanted movement from nearby objects. A region about 40 to 80 pixels square works for most footage at 1080p.

Running the track and checking for drift

Once your tracking region is positioned, click the Track Forward button to start tracking from the current frame to the end of your shot. Nuke will step through each frame, measure where your point moved, and record that data. This usually takes a few seconds to a few minutes depending on your shot length.

After tracking completes, scrub through your footage and watch the tracking square follow your point. If it stays locked to your target throughout the shot, you're done with this step. If the square drifts away from your point partway through — especially toward the end — the track failed in that section.

Common reasons for tracking failure: the point leaves the frame, the point rotates or changes shape, the lighting changes dramatically, or the point becomes blurry. If you see drift, you can either restart the track from a frame where it was still working, or manually adjust the drifted frames by hand. For small drifts, manual adjustment is often faster than re-tracking.

Linking the 3D object to the tracker output

Create your 3D object — a cube, sphere, or imported model — in a separate part of your node graph. You'll also need a Camera node that matches your footage's perspective. If your footage has camera movement, the camera node needs to match that movement too, or your 3D object will appear to float unnaturally.

The tracker outputs two values: the X and Y position of your tracked point on the screen. You need to connect those values to your 3D object's translate X and translate Y. In Nuke, you do this by right-clicking on the 3D object's translate fields and selecting Link, then choosing the tracker's output channels.

Alternatively, you can use an Expression to pull the tracker data directly. Type an expression like Tracker1.track1.x into your 3D object's translate X field. This tells Nuke: "Make this object's X position equal to wherever the tracker says the point is." Do the same for translate Y.

Adjusting scale and depth when the object doesn't look right

Once your 3D object is linked to the tracker, play back your shot. The object should follow your tracked point across the screen. But it might look too big, too small, or like it's floating in front of or behind where it should be. These are separate from tracking — they're about how the 3D object sits in 3D space.

If the object is the wrong size, adjust its Scale value in the 3D object's properties. If it needs to move forward or backward in the scene, adjust its Translate Z value. If the object needs to rotate to match the angle of your footage, adjust its rotation values. None of these changes affect the tracking itself — they just position the object correctly once it's locked to the tracked point.

If your footage has perspective distortion (the camera is tilted or angled), you may need to adjust the 3D object's rotation to match. This is trial and error: rotate the object, play back the shot, and see if it looks natural. The goal is for the object to appear as if it's sitting in the same 3D space as your footage.

Handling camera movement and match-moving

If your footage has camera movement — a pan, a zoom, or a dolly — your 3D object will look wrong even if the tracker is perfect. The object will follow your tracked point, but it won't move with the camera. To fix this, you need to match-move your camera: create a camera in Nuke that mimics the camera movement in your footage.

Nuke has a Camera Tracker node that can solve camera movement automatically by tracking multiple points across your footage. This is more advanced than a straightforward 2D tracker, but it's the right tool when your shot has significant camera motion. The Camera Tracker outputs a camera that you plug into your 3D scene, and suddenly your 3D object stays locked to the world instead of drifting with the camera.

If your shot has only minor camera movement or if you want to keep things straightforward, you can manually keyframe the camera's position and rotation to match your footage. This is slower but gives you full control. The key is that your 3D object's position (linked to the 2D tracker) and your camera's position (linked to your footage's camera movement) need to work together.

Exporting your tracked data and troubleshooting common problems

Once your track is complete and your 3D object is positioned correctly, you can render out your composite. The tracker data stays embedded in your Nuke script, so if you need to adjust the object later, the tracking is still there.

If you need to use your tracking data in another program — like a 3D modeling tool or a different compositing software — you can export the tracker's output as a text file or as keyframes. Right-click on the tracker node and look for export options. Most 3D software can read Nuke's keyframe format.

Common problems: the tracker drifts because your tracking point is too small or featureless; the 3D object looks like it's floating because the camera movement wasn't matched; the object disappears because it's positioned behind the camera. For drift, try a larger tracking region or a different point. For floating objects, check your camera node. For disappearing objects, adjust the Z position or check that your camera's near clipping plane isn't cutting off your object.

Frequently Asked Questions

Can I track a point that moves off-screen partway through the shot?

No. The tracker needs to see your point on every frame. If it leaves the frame, stop the track before it exits, or start a new track from where it re-enters. You can blend two separate tracks together by manually keyframing the transition between them.

What if my tracking point rotates or changes shape?

The 2D tracker assumes your point stays the same shape. If it rotates significantly — like a person's hand turning — the tracker will drift. Use a different point that stays more stable, or switch to a more advanced tracker like the Camera Tracker that can handle rotation.

Do I need to match-move the camera for every shot?

Only if your camera moves. If your shot is locked on a tripod, you don't need camera tracking — just link your 3D object to the 2D tracker and you're done. Camera tracking adds complexity and is only necessary when the camera pans, zooms, or moves through space.

Can I manually adjust the tracker if it drifts?

Yes. Click on individual frames in the timeline, then drag the tracking square to the correct position. Nuke will update that frame's tracking data. This is useful for small drifts near the end of a shot, but if the tracker drifts throughout, re-tracking with a better point is usually faster.

Why does my 3D object look flat or wrong even though the tracking is correct?

Tracking only controls X and Y position on the screen. You also need to set the object's Z position (depth), scale, and rotation to match your 3D scene. Check that your camera is positioned correctly and that your 3D object's scale makes sense relative to your footage.