Crop Video in After Effects Without Changing Composition Size

Why Cropping Without Changing Composition Settings Is a Core Skill

Every project eventually hits the same moment: a client asks for a vertical cut, an editor spots a stray boom mic in the top corner, or a logo needs to disappear from a shirt. The instinctive move is to open Composition Settings and shrink the canvas. That instinct is almost always wrong. The moment you change composition width and height, every layer, camera, null, expression, and guide inside that comp re-flows. Text wraps differently, particles that were carefully placed fall out of frame, and a stack of nested comps suddenly letterboxes in ways nobody asked for.

Cropping in After Effects is not a single button. It is a set of layer-level techniques that change what pixels are visible while leaving the canvas untouched. Once you internalize that separation between the canvas and the visible frame, editing becomes dramatically faster. You can produce a 9:16 version, a 1:1 version, and a square version from one master comp, all without duplicating or corrupting the original.

This guide walks through the practical methods in order of usefulness, the trade-offs between them, and the situations where each one wins.

Composition Size vs. Output Size vs. Visible Frame

Before touching tools, it helps to separate three things that beginners constantly blend together.

Composition size is the canvas: the fixed rectangle every layer renders into. Change it and you change the coordinate system for everything inside.

Output size is the render decision. In the Render Queue's output module, or in your encoding app, you can scale, letterbox, or crop the final export independently of the canvas. This is the safest place to make delivery-format decisions.

Visible frame is what the viewer actually sees after masks, mattes, track mattes, opacity, and parenting do their work. This is where cropping lives.

A quick decision rule saves a lot of rework:

• If only one layer needs hiding → use a mask or track matte on that layer.
• If a whole group of layers needs hiding → use a matte through an adjustment layer, or move those layers into a pre-comp.
• If the delivery format is smaller than the canvas → nest the master comp inside a second comp sized for delivery, or crop at export.
• If the canvas itself is genuinely wrong → that is the one case where Composition Settings is the right answer.

One clarification worth stating plainly: After Effects does not have a "Crop" property the way an image editor does. Cropping is achieved by controlling what renders, not by trimming a canvas rectangle. That distinction is what makes the non-destructive approach possible.

Layer Masks: The Most Flexible Way to Crop

Masks are the workhorse. They are non-destructive, they are per-layer, and they can be animated, feathered, and driven by expressions.

Drawing a rectangular mask quickly

Select the layer in the Timeline, then drag with the Rectangle tool on that layer in the Composition panel. Because a layer is selected, the drag creates a mask rather than a new shape layer. You can also use the Pen tool to click a custom polygon, or press the Ellipse tool for circular vignettes.

A fast workflow for hiding an edge intrusion:

1. Select the footage layer.
2. Draw a rectangle around the area you want to keep.
3. Set the mask mode to Add and check that the mask is inverted only if you drew around the problem instead of the subject.
4. Feather by one to three pixels if the surrounding content is soft; leave it at zero if the shot is tack sharp.

Hiding the mask edges matters more than people expect. If the surrounding area is a flat color, a mask will be invisible. If the surrounding area is moving detail, the mask boundary can read as a hard rectangle, so feathering or a slightly inset shape is required.

Mask modes and feathering

Each mask has a mode: None, Add, Subtract, Intersect, or Difference. Most crop work uses Add on a single mask plus, when needed, a Subtract mask for a second hole. Feathering is controlled through Mask Feather, and it accepts two values so you can feather horizontally and vertically differently. Mask Expansion grows the shape without redrawing it, which is handy for animating a reveal inward or outward.

Two limitations to remember. Masks have no motion blur of their own, so a fast-moving mask edge will look stepped unless you animate it with enough keyframes or add blur deliberately. And masks do not shrink the layer's bounding box, so effects placed after the mask may still sample content from outside the visible area. When that matters, a track matte is the better tool.

Animating masks without fighting them

For simple reveals, keyframe Mask Path and set the interpolation to something smooth. Sharp corners and slow, even movement rarely need more. For complex shape changes, the dedicated mask interpolation workflow smooths the transition between two very different paths and eliminates the twisting that comes from interpolating point-by-point. Keep your mask point count consistent between keyframes; adding or removing points mid-animation is the single most common cause of ugly mask morphs.

Track Mattes, Set Matte, and Shape-Layer Crops

A track matte crops a layer using another layer's alpha or luminance. The practical upside is crispness and control: you build the matte as a shape layer or solid, animate it with shape path keyframes, and use it across as many layers as you like.

Basic setup:

1. Create a shape layer above the footage.
2. Draw the region you want to keep (or the region you want to remove, then invert).
3. Set the footage layer's Track Matte to Alpha or Alpha Inverted.

The shape layer is automatically hidden when used as a matte, and its own opacity, feather, and position are fully animatable. Because shape layers support rounded corners, offset paths, and repeater-based patterns, you can build crop shapes that would be tedious to draw as masks.

For cropping many layers at once, put a single adjustment layer above the stack and apply the Set Matte effect, pointing it at your shape layer. Every layer below the adjustment layer is now cropped to that shape. This is the fastest way to reframe an entire scene — a full composition of footage, text, and graphics — into a new visible window without touching any individual layer.

A quieter option that surprises people: the layer blending controls include a preserve-underlying-transparency switch. When it is enabled, the layer becomes transparent wherever nothing exists beneath it in the stacking order. It is not a classic crop, but it is an elegant way to trim a top layer to the silhouette of the scene below.

The Transform Effect and Other Non-Mask Crops

The Transform effect (Effect > Distort > Transform) is frequently overlooked as a cropping tool. It moves, scales, rotates, and skews the content of a layer inside the layer's own bounds, without changing the layer's position property or the composition. That distinction matters in two situations.

First, when you want an effect further down the chain to keep operating on the full frame while the visible content shifts. Second, when you want motion blur on the movement itself: the Transform effect has its own shutter angle and sampling controls, so a fast reposition can blur convincingly even when the underlying property has no motion blur.

Other useful non-mask approaches:

• Grow Bounds on effects that clip unexpectedly, which lets a glow or blur extend beyond the original layer bounds rather than being cut off at the layer edge.
• Scale and position on a parented null. Parent the footage to a null, then offset the null. The footage moves as a unit while its own transform values stay clean for other uses.
• Corner Pin when the crop needs to be a perspective quad rather than a rectangle, for example fitting a screen into a shot at an angle.
• Opacity and blending modes for simple fades rather than geometric crops, which are often mistaken for crop problems when they are really edge-blending problems.

None of these change composition settings, and each one leaves the source footage intact, so revisions later in the project are painless.

Pre-Composing for Contextual Cropping

Pre-composing is where cropping becomes architecture rather than patchwork. When you pre-compose a set of layers, those layers move into their own composition with its own dimensions. You then scale and position that pre-comp inside the parent, and the parent's canvas is honored while the inner content is effectively cropped by the parent's frame.

When to pre-compose

Pre-compose when you need to treat a group as one object: a lower-third package, a hand-tracked element with its own internal timing, or an entire generated scene that has to be repositioned inside a delivery frame. Select the layers, pre-compose with Move All Attributes, and name the comp meaningfully. Then transform the pre-comp layer in the parent rather than each child layer.

One caution: a pre-comp creates a hard boundary at its edges. Effects that render outside the layer, such as glows and blurs, will be clipped at the pre-comp border. If that happens, either enlarge the pre-comp's dimensions at creation time, or enable continuously rasterize on the pre-comp layer so vector and shape data renders at full resolution.

Nesting a master comp for delivery frames

This is the cleanest crop of all. Keep your master comp at its full size. Create a second comp at the delivery size, then drag the master comp into it as a nested layer. The delivery comp's frame now acts as a cropping window: whatever falls outside is simply not rendered. You can duplicate the delivery comp for 1:1 and 4:5 versions, and because they all reference the same master, a single fix propagates everywhere.

Because nested comps are just layers, you can also scale and reposition the nested master to choose exactly which part of the frame survives the crop. Add guides at the safe-area lines so you can see what will be cut.

Managing nested comps and expression linking

Deep nesting gets slow. Once you are three or four levels deep, render time climbs and the timeline becomes hard to read. Flatten intermediate levels when the internal edit is final, and use collapse transformations on vector-heavy pre-comps to avoid unnecessary rasterization. If expressions inside the nested comp need to react to the outer comp, reach out with a layer reference expression instead of re-creating sliders in both places.

Expressions for Dynamic, Responsive Crops

Expressions turn a static crop into a system. The pattern is always the same: put a controller on a null or an adjustment layer using the expression control effects (Slider, Angle, Checkbox), then point the crop property at the controller.

Practical examples worth building:

• Feather from a slider. Put a Slider Control on the null named Crop Controls, then set the mask's Mask Feather to:

thisComp.layer("Crop Controls").effect("Feather")("Slider")

• Matte size from two sliders. On a rectangle shape layer used as a matte, express the rectangle path size from Width and Height sliders so an entire crop can be adjusted by dragging two controls instead of editing shape points.

[thisComp.layer("Crop Controls").effect("Width") ("Slider"),
 thisComp.layer("Crop Controls").effect("Height") ("Slider")]

• Offset a mask path with a slider. Add a small expression to the mask path that adds a single offset value to both axes, letting you nudge a crop without opening the mask shape.
• Auto-fit a crop to text. Use the bounding-box query for text layers so the crop hugs the text regardless of how the copy changes, useful for template work where copy is replaced frequently.

Two expression habits keep these systems maintainable. Name every controller layer so references read clearly, and avoid hard-coding layer indices, because a single reorder breaks the whole rig. Sliders and checkboxes are also far more legible to collaborators than a stack of mask keyframes.

Cropping AI-Generated and Variable-Aspect Footage

Generated footage introduces its own quirks. Models often return slightly different dimensions than requested, edges can carry warping or repetition artifacts, and subject framing can drift between shots. Cropping becomes part of normal cleanup rather than a final polish step.

A workflow that holds up:

1. Generate as close to the target aspect ratio as possible. Cropping a horizontal clip into vertical always throws away pixels and resolution.
2. Inspect all four edges at 200 percent zoom before compositing anything. Edge artifacts are cheaper to regenerate than to paint out.
3. Build the crop with a matte so you can adjust it later without redrawing masks.
4. If the crop must be aggressive, apply a detail-preserving upscale after scaling, and check for haloing around high-contrast edges.
5. Keep a safe margin of roughly five percent inside the intended delivery frame, because platform interfaces cover edges with buttons and captions.

For multi-format delivery, plan the reframe rather than the crop. Decide where the subject should sit in a vertical frame, then offset the footage so that composition works, rather than centering and then cropping symmetrically. Vertical versions of horizontal footage almost always look better when the subject sits slightly above center.

Export, Delivery, and Quality Control

Once the visible frame is correct, the export has its own crop options that never touch composition settings. The output module in the Render Queue includes a crop control that trims the rendered edges of everything in the comp. It is a blunt but fast tool for producing a slightly smaller final file from a finished render.

More controllable is the delivery-composition approach described earlier: nest the master, crop with the frame, and render from there. That gives you per-format control and a clean audit trail.

Quality checks worth running before every delivery:

• Scrub the full timeline at 100 percent zoom looking only at the four edges.
• Toggle the alpha channel view to confirm mask and matte edges have no unintended softness.
• Check text and captions against the safe area for each format.
• Confirm the render is not letterboxed unintentionally. Black bars usually mean a scaled nested comp, not a bad crop.
• Watch the final file on a phone. Edge problems that are invisible on a large monitor are obvious on a small screen.

Common Mistakes, Checklist, and FAQ

Mistakes to avoid

• Resizing the composition to crop. This is the root cause of most downstream breakage.
• Feathering aggressively on sharp footage, which produces a visible vignette halo.
• Animating masks with inconsistent point counts, causing twisting transitions.
• Forgetting that masks do not clip effects that sample outside the visible area.
• Cropping a low-resolution generated clip and then scaling up, which doubles the softness.
• Building ten separate delivery comps with duplicated edits instead of one master and nested children.
• Leaving a hard matte edge on a moving shot where the background is busy.

Quick checklist

• Canvas untouched.
• Crop method chosen for the right reason: mask for one layer, matte for many, pre-comp for groups, nested master for delivery formats.
• Edges inspected at high zoom.
• Safe areas respected per format.
• Controllers named and rigged with sliders where the crop will be revised.

FAQ

Does cropping reduce resolution? Cropping itself does not resample pixels; it simply hides them. Resolution only drops if you then scale the remaining image up.

Why does my mask edge look soft? Usually nonzero feather, or an alpha channel blended over a lower layer. Check Mask Feather and Mask Expansion first.

Can I crop only one layer and leave the rest alone? Yes, that is exactly what a layer mask or track matte does. It affects only the layer it belongs to.

What is the cleanest way to make a vertical version? Keep the master comp at full size, create a vertical delivery comp, and nest the master inside it. Reposition the nested master rather than rebuilding the edit.

Do masks carry through to the export? Yes, masks render as part of the layer. Alpha is preserved if the output module includes an alpha channel.

Should I crop or regenerate? If the subject is being cut badly, regenerate at the correct aspect ratio. Cropping should refine framing, not rescue composition.

Cropping is one of those skills that looks trivial and quietly determines how maintainable a project is. Treat the composition as a fixed canvas, treat the visible window as a design decision, and you gain the ability to reframe on demand without ever reopening Composition Settings.