Cameras
A camera is your viewpoint into a 3D scene — frame the shot, then look through it.
In DNA, every 3D scene is seen from a camera. By default you fly around with the free viewport camera, but you can also drop a Camera node to lock down a specific shot, animate it, and render through it. The Camera node decides where you look from; the render.render node is what turns that view into pixels.
The viewport camera vs. a Camera node
When you open a scene you're navigating the free viewport camera — orbit, pan and dolly with the mouse to find your angle. Nothing is committed; it's just how you're looking around.
Add a Camera node (camera.viewport) when you want a named, savable, animatable viewpoint — a shot you can return to, keyframe, and render from. The node carries a position, a direction, and a lens.
You don't need a Camera node just to look around. Add one when you want to keep a framing, animate a move, or render a specific shot.
Looking through a camera
Tell the viewport to look through a Camera node and the viewport renders exactly what that camera sees — like the "active camera" view in compositing apps. There's one active camera at a time, and it drives the main view.
The best part: while you're looking through a camera, orbiting, panning, dollying and scroll-zooming all drive that node. You're not knocking the view off onto some disconnected free camera — you're posing the actual camera, and your moves are saved back onto it. Frame the shot by eye, and it sticks.
If the camera is parented to another camera (a rig), your navigation still moves the original camera at the start of the chain, and the rig re-derives the rest.
The camera you're looking through also drives gizmos, overlays and picking, so handles always line up with what you see — no drift between the rendered view and the tools.
Orthographic vs. perspective
The viewport has two projection modes:
Perspective — the natural, lifelike look where distant things shrink. This is what you want for most 3D shots.
Orthographic — parallel projection with no perspective falloff, ideal for technical/front-on framing and precise alignment.
Switching between them keeps your view stable — DNA matches the framing across the change rather than jumping. You can still orbit in orthographic mode (hold Alt and drag).
Some screen-space effects (like ambient occlusion) assume a perspective view and are skipped in orthographic mode.
Camera modes and lenses
The Camera node has three aiming modes:
Free — position the camera and rotate it directly.
Target — the camera always aims at a target point you set (great for orbiting a subject).
FPS — a first-person walk-around mode with mouse-look. Press Escape to release the mouse and stop looking through it.
Beyond a standard lens, the Camera node also offers lens models for creative looks — fisheye variants, focal length in millimetres, and curvature distortion — so you can go from a clean rectilinear shot to a wide fisheye without leaving the node.
For exact parameters:
| Parameter | Type | Default |
|---|---|---|
mode | String | "Free" |
position | Vec3 | (0, 0, 5) |
rotation | Vec3 | (0, 0, 0) |
target | Vec3 | (0, 0, 0) |
projection | String | "Perspective" |
focal_length | Number | 28 |
lens_model | String | "Rectilinear" |
curvature | Number | 0 |
perspective_inversion | Number | 0 |
ortho_height | Number | 10 |
near_clip | Number | 0.100 |
far_clip | Number | 1000 |
fps_speed | Number | 5 |
capture_width | Number | 512 |
capture_height | Number | 512 |