Fields & sampling
A field is a value that exists everywhere in space — read it at any point to drive color, motion, shape, or any parameter you like.
Most values in DNA are a single number, color, or vector. A field is different: instead of one value, it gives you a different value at every point in space. Think of noise that fills a room, a gradient that fades across your canvas, or a distance that tells you how far you are from a surface. You don't set the value — you ask the field for it, wherever you are.
For the full menu of field flavors (scalar, vector, color, Volume, Distance Field, and more), see Fields. This page is about the three things you'll actually do with fields.
Drive any parameter spatially
The simplest use of a field is to let it vary something across space instead of keeping it flat.
Wire a field into a parameter and each point picks up its own value. A noise field plugged into color gives you mottled, organic variation. A gradient plugged into height pushes one side up and the other down. A radial field plugged into scale makes things grow toward the center.
This works because a field already knows how to read itself anywhere — at a point, along a surface, or across a whole canvas. You just point it at the thing you want to vary.
Fields adapt to whatever you ask of them. Feed a noise field into a color slot and it reads as grayscale shading; feed the same field into a position and it reads as a push in space. One field, many uses — no conversion step on your part.
Sample a field at your points
When you have geometry — points, particles, a mesh, a curve — you often want each point to grab the field's value at its own location. That's sampling.
Drop a sample node between your geometry and your field, and every point reads the field where it sits and stores the result as an attribute. Now you can:
Color particles by a noise field, so each one takes the shade of the space it occupies.
Push points along a vector field, turning a static cloud into a flowing one.
Read a Distance Field at each point to know how far it is from a surface, then fade or hide points that stray too far.
Because the value is now baked onto your points as an attribute, you can use it like any other attribute downstream — in an The Expression language, a filter, a color ramp, anything.
Sampling reads the field as it is right now. If the field animates over time, re-sampling on each frame keeps your points in step with it.
The field ↔ geometry round trip
Fields and geometry aren't separate worlds — you can cross between them in both directions.
Geometry → field. Turn a shape into a field and it becomes something you can read anywhere in space. A mesh or a set of points can become a Distance Field (how far am I from this surface?) or a Volume (what's the density here?). Once it's a field, you can sample it, blend it, displace it, or feed it into anything that takes a field — even though it started life as plain geometry.
Field → geometry. Go the other way and a field shapes geometry. Sample a vector field onto points to displace them. Use a Distance Field to carve, fill, or grow a surface. Render a Volume directly as soft, lit smoke or cloud.
This round trip is what makes fields so powerful for effects work: model something concrete, lift it into a field to manipulate it continuously, then bring it back to something you can see and render.
A common recipe: convert geometry to a Distance Field, blend or warp it in field-space (where shapes melt and merge smoothly), then render or re-mesh the result. The smoothness you get blending fields is hard to fake with geometry alone.
A quick mental model
A field = a value that exists everywhere.
Driving a parameter = letting the field set that value across space.
Sampling = each point grabbing the field's value at its own spot.
The round trip = geometry can become a field, and a field can shape geometry.
Keep those four ideas in hand and most field-based effects stop feeling mysterious.