Cameras
A camera is a real layer with a pinhole lens model. Without one, a 3D composition is rendered from a default view; with one, you control the shot.
Creating a camera
Insert → Camera, or the + Camera button in the viewport header. Either opens a real creation dialog where you choose the lens preset up front, rather than inserting a hardcoded default you then have to fix.
One-node and two-node
| Type | Behaves like |
|---|---|
| One-node | A free camera. You aim it by rotating it |
| Two-node | A camera with a point of interest. It always looks at that point; move either end |
Two-node is what you want for orbiting a subject or for a move that must keep something framed. One-node is better for a fly-through where the aim itself is the animation.
Focal length
Focal length is expressed in composition-space pixels, with familiar mm presets in the Camera section of the Style inspector tab. Pick the preset that matches the look you want — wide for drama and exaggerated parallax, long for a compressed, product-shot feel — and the underlying pixel value follows.
Focal length is keyframeable, so a zoom is available as a distinct move from a dolly. Animating both in opposite directions gives you a dolly zoom.
Depth of field
The camera drives depth-of-field blur. Layers away from the focus distance blur; layers at it stay sharp.
Because focus distance is a keyframeable property, a rack focus is two keyframes.
A sealed composition instance resolves its own camera and its own depth of field, not the host’s — see The composition boundary.
Grabbing the camera in the viewport
You can drag the camera itself, the way you would in After Effects: switch to an orthographic view, grab the camera’s position handle, and move it.
The position handle and the Point of Interest crosshair are independently draggable — for cameras and for spot lights. Dragging one never moves the other: a target that follows the body can never be aimed, and a body that follows the target can never be placed.
A camera and a light have no geometry, so neither is ever hit by ordinary layer picking. Their handles are hit-tested separately.
Parenting a camera
A camera or a light can be parented, to a null or to any other layer — the standard camera-on-a-null rig works. Parenting is bounded by the composition: you cannot parent across a composition boundary.
If you are working on the code: a drag knows where the pointer is in world
space, and a node stores parent-space values. Writing the world position
straight to a parented node applies the parent transform a second time on the
next frame, so the device leaps away from the cursor the instant you release.
Every write goes through Matrix4Math.toLocalPoint against the parent’s world
matrix.
Animating a camera move
The reliable pattern: frame the shot interactively, then keyframe. Use C to cycle orbit / pan / dolly and drag until the view is right, then set a keyframe on the camera’s position and orientation. Repeat at the end of the move.
Or start from a preset — Cinematic Pan 3D, Depth Push In and Orbit Tilt 3D in the Animate menu all write real camera keyframes you can then adjust.
Camera verbs (Layer → Camera)
Four commands, new in 0.7, for the moves people used to build by hand:
| Command | What it does |
|---|---|
| Create Orbit Null | Adds a null at the point of interest and parents the camera to it, so one Y-rotation keyframe pair orbits the subject |
| Set Focus Distance to Layer | Writes the selected layer’s distance into the camera’s focus distance once |
| Link Focus Distance to Layer | The same, as an expression — the focus follows the layer as either moves |
| Link Focus Distance to Point of Interest | Focus rides the point of interest, so a two-node camera is always sharp on what it looks at |
View → Viewport → 3D View → Look at Selected / Look at All aims the current view at the selection or the whole scene without touching the camera’s keyframes; it is a viewport move, not a scene edit.
Checking the shot
| Key | View |
|---|---|
| 1 | Active Camera — what the render sees |
| 2 | The last custom view you used |
Use the 2-up or 4-up multi-view layouts to keep an orthographic view open beside the camera view. Model in one, verify depth in the other. See Navigating 3D.
Related
- 3D layers
- Lights & materials
- The assistant’s
add_camera_movetool: Tool reference