Reframe Guide

Why Reframe?

When you animate a character, the AI model treats the whole image as the stage. If your character fills the frame, there’s no room to move — attacks get clipped, jumps have no headroom, and run cycles feel cramped. Reframing places your character on a larger canvas before you animate, giving the model space to produce full motion.

Demon

Tight frame

→

10 frames

Reframed

→

10 frames

Cactus

Tight frame

→

8 frames

Reframed

→

8 frames

When to Reframe

You don’t need to reframe every animation. It matters most when the character needs to move beyond its resting bounds.

Reframe recommended

Attacks, slashes, spells — room to swing and follow through
Jumps, dashes — vertical or horizontal space for the motion

Usually fine without

Running or walking — the character stays in frame well on its own
Idle breathing, blinking — subtle in-place motion
Images that already have padding around the character
Small contained motions (head-turn, tail-wag)

How to Use It

Reframe controls are built into the Animate tab. When you select a starting image, the canvas preview on the right shows your image with reframe controls below it. Everything runs in your browser — no credits, no cloud jobs.

Select a starting image in the Animate tab — from your library or drag-and-drop one onto the panel.
Set Canvas Scale — controls how much larger the output canvas is relative to your image’s longest side. 1.0x = image fits exactly (padded to a square). 1.5x is a good default for most animations.
Choose Fill Method — Solid Color fills with a single color (auto-sampled from the image edge, or pick your own); Edge Extend stretches the edges outward; Transparent leaves the added space empty.
Position your character with the anchor grid, or drag the image directly in the preview to place it exactly where you want.
Submit — when you submit the animation, the reframed image is automatically sent to the AI model. Your original is never modified. You can also use Save to Library to save the reframed version as a standalone image.

Size Limit

The maximum output size depends on the animation model you’ve selected. For pixel-engine models, the limit is 256 × 256 px. For frame-engine models, it’s 2048 × 2048 px. The canvas toolbar shows a warning if your output exceeds the limit — lower the scale, or switch to a different model.

Tips

Match space to action — anchor to the bottom for jumps, to the opposite side for attacks.
Don’t over-scale — 1.3x–1.8x works well. If the character is too small relative to the canvas, the model may struggle to animate it.
Remix — when you remix an animation job that used reframing, all canvas settings are restored automatically so you can tweak and re-submit.

Learn More

For tips on prompting, frame counts, and starting poses, see the Animate Guide.