ImageBuddy documentation

Drop an image and ImageBuddy trims it to the artwork, keys the background out to transparency, wraps it in an outline and a drop shadow, pads it, recolors the backdrop, resizes it, sets a title and caption around it, stamps a watermark over the result, and can trace it to a clean SVG — all over a live preview so every cut-out is obvious at a glance. This page walks through each stage.

Installation

ImageBuddy comes two ways. The download from bendansby.com/apps/imagebuddy is a standard .dmg — open it and drag ImageBuddy.app to your Applications folder. The store version installs and updates the usual way.

Requires macOS 13 Ventura or later, on Apple Silicon or Intel. The direct-download build auto-updates via Sparkle; you can also check anytime via the app menu's Check for Updates…. The direct build runs a free 7-day trial with every feature unlocked, then a one-time license unlocks unlimited use — buy it on its own or as part of the Surprisingly Useful bundle. The store version is paid through the usual pipeline with no separate license key.

ImageBuddy only ever reads and writes the files you drop on it or pick from a dialog. Nothing is uploaded — every stage of the pipeline runs on your Mac.

The preview canvas

Drop an image onto the window (or open one with ⌘O) and it appears on a checkerboard, with a striped canvas around it. The checkerboard is the tell: anywhere it shows through the image is a transparent pixel, so every cut-out the pipeline produces is obvious at a glance. Everything renders live — nudge a slider or flip a switch and the preview updates immediately.

A dashed output-bounds frame traces exactly what a Save will write, and a size badge on it reports the pixel dimensions of that output. Because stages like Stylize and Add text can make the result larger than the source, the badge is the honest measure of what you'll get — watch it as you dial the settings in the right-hand sidebar.

On a large image the preview keeps up with a slider by rendering at preview resolution while you drag, then settling to the full-size result the moment you let go — so a drag may look fractionally soft mid-flight, and the output-size badge holds its last exact value until the drag ends rather than showing an approximation.

Trim & crop

ImageBuddy's oldest trick: shave the empty border off an image down to the artwork.

Auto trim

Turn on Trim to content and ImageBuddy finds the bounding box of everything that isn't transparent or background-colored, and crops to it. A tolerance slider widens or narrows what counts as "empty border" — raise it if a faint edge halo is being kept, lower it if real artwork is being clipped.

ImageBuddy samples the border first to work out the artwork's actual background color — white, black, or anything else — and trims around that detected color rather than assuming white, falling back to white if the border isn't confidently one color. If a differently-colored region at the edge is being kept, that's the same detection doing its job.

Manual crop, rotate & flip

You can also draw a crop by hand, with a loupe for pinpoint placement and arrow-key nudging of the crop edges. Rotate and flip the image in 90° steps to reorient or mirror it before the rest of the pipeline runs — for anything that isn't a quarter turn, see Straighten.

Straighten

Turn on Straighten to level a crooked shot by any angle — the free-rotation counterpart to the 90° rotate buttons. Drag the Angle slider up to 45° either way, or use the − / + buttons beside it to nudge by a tenth of a degree at a time, which is usually what a horizon actually needs. A thirds grid appears over the preview while the stage is on, to judge the result against something straight, and a Reset link returns the angle to zero.

By default the output crops in to the largest straight rectangle that fits inside the rotated image, so there are no empty corners — which means the result comes out a little smaller than the source, with nothing scaled up to fill the frame. Turn on Keep the whole image to grow the canvas to fit the rotation instead, leaving the corners transparent for Add background to fill, or to save as-is in a format that holds alpha.

Straighten runs early — right after rotate/flip, before the crop — so everything downstream measures the levelled image, and the manual crop editor draws on already-straightened pixels. Because rotating moves the pixel grid, changing the angle clears any exclusion marks you've painted.

Remove the background

Turn Remove background on to lift the subject off its background and out to transparency. There are two modes.

Auto

The default is a border-connected flood fill: ImageBuddy first works out the border's actual color — white, black, or any other solid color — then floods inward from the edges of the image, clearing the connected region that matches it. Because it's a flood from the outside in — not a global "clear every matching pixel" — interior details survive: eye whites, the counters inside letters, the inside of a speech bubble, anything the flood can't reach and anything that genuinely doesn't match the detected color. A tolerance slider widens the match so an off-tone or faintly tinted background still keys cleanly.

When you do want the enclosed matching-color pockets gone too, turn on Include internal areas. That switches keying to a global mode that also punches out matching regions the border flood can't reach. Leave it off for the safe, subject-preserving default; turn it on as the deliberate, more aggressive choice. For the rare patch that even the tolerance slider — or Include internal areas — would swallow but you want kept, see Mark exclusions below.

Subject

On macOS 14 or later, ImageBuddy can instead lift the salient subject with an on-device Vision cut-out — the system finds the foreground object and masks everything else away, regardless of the background's color. This is the one to reach for on photos and busy backgrounds where a single background color doesn't describe the thing you want to remove.

Keying runs first in the pipeline, so everything downstream — the trim box, the outline and shadow, padding — measures the real cut-out rather than the original rectangle.

Mark exclusions

Sometimes a threshold slider alone can't tell a stubborn white patch — a logo on a shirt, a highlight, a bright reflection — from the background you're trying to key out. Mark exclusions lets you hand-protect exactly those pixels before the key runs.

With Remove background turned on and set to Auto mode, click Mark exclusions and the preview swaps into a paint canvas. Drag over an area to protect it; hold ⌥ while dragging to erase marks instead. A brush-size dial controls how wide the stroke is, and a trash button appears once marks exist so you can clear them and start over. Click the same button — now labeled Done Marking — to return to the normal preview.

A painted mark isn't just a wider tolerance — it's absolute. Marked pixels are never cleared by the key, and they block the flood fill from crossing them, so a mark protects a patch even Include internal areas would otherwise swallow.

Exclusion marks belong to a specific image's pixels, not to its settings. They don't travel with Sync all across the batch queue, and a newly added image never inherits another image's marks. They're also cleared automatically by anything that moves the pixels underneath them — rotating, flipping, or a Save Over Original that reloads the file — since the mask's coordinate space would no longer line up with the image.

Exclusion marking only applies to Auto mode; it has no effect on Subject mode's Vision-based cut-out.

Stylize

Turn on Stylize for the classic die-cut sticker look — an outline hugging the cut-out, a soft drop shadow underneath, or both. The two are independent: tick either on its own, or both together.

Outline

A solid band traced around the silhouette, with its own width and color. White is the sticker default, but any color works, including a translucent one — the outline keeps its clean anti-aliased edge either way.

Shadow

A blurred, offset copy of the silhouette laid underneath, so the cut-out lifts off the page. Blur sets how soft its edge reads, Offset how far it's displaced, and Angle which way it falls (0° right, 90° straight down) — the angle is hidden at an offset of zero, where the shadow sits dead behind the subject and has no direction to point in. Opacity and a color finish it.

With the outline on, the shadow is cast by the outlined shape; with it off, by the artwork's own silhouette.

The canvas grows to fit whichever of the two is reaching furthest out, then crops tight to the finished result — so the output is always snug to its own outline and visible shadow. A practical consequence: it looks identical whether Trim to content is on or off, because the outline defines the bounds, not any surrounding margin.

Because the outline and shadow add pixels, a stylized image can end up larger than the original. The output-bounds badge on the preview always shows the true final size.

Add a background

Turn on Add background to composite the image over a backdrop, filling every transparent pixel the earlier stages produced — the inverse of keying white out. Pick a solid color from the built-in color picker (HSB, alpha, and hex), a gradient, or an image to sit behind the subject.

An opaque background covers all the transparency underneath, so a Save can stay in the original format (see Saving). A translucent background still leaves real transparency, and the save will fall back to PNG.

Resize & canvas

Resize Image and Adjust Canvas are independent stages — turn on either, both, or neither.

Resize Image

Scale the output to an exact width × height. An aspect lock keeps the proportions intact so you set one dimension and the other follows — turn it off to stretch to a specific size on purpose. The width and height fields respond to the arrow keys — press and hold to keep nudging, hold ⇧ for steps of 10.

Adjust Canvas

Resize the canvas without scaling the image inside it. Pick an origin from a 3×3 grid — the corner, edge, or center the image stays anchored to as the canvas grows or shrinks around it — then set the new size as an absolute pixel value or a size relative to the current canvas, with an optional fixed aspect ratio. Growing the canvas pads transparently; shrinking it crops from whichever sides move in. This is also how you add a transparent margin: a relative size with a positive amount and the origin set to centre pads all four sides evenly, and any other origin pads one side at a time — it replaces the separate Add padding stage earlier versions had.

Because an even border is the common case, Relative mode also shows a Padding slider that grows the canvas by the same amount on width and height at once, so you don't type the same number into both fields. The fields stay editable underneath it for uneven padding — or a negative value, which crops in rather than pads.

Add text

Turn on Add text to caption the result: a title in a band of space added above the image, a caption in one below it — the way a matted print is captioned. Both fields are optional and independent, so a title with no caption adds space above only, and an empty field costs no space at all.

This is deliberately not a free placement. Putting type over the image is what Add a watermark already does; Add text sets type around it, outside the artwork.

How it looks

A shared color swatch, a font style — Sans Serif, Serif or Round — and an alignment for both bands. There's no weight control on purpose: titles are always set bold and captions regular, which is what makes them two fields rather than one. Each band gets its own size slider, shown once that band has text in it, as a percentage of the output's longer side so the type reads at a consistent scale whatever the final dimensions are. A padding slider sets the room around the type inside its band, and how far in from the left and right edges it wraps.

Each band grows to exactly the type it holds, and a line too long for the width wraps onto another line rather than running off the edge — the band grows to match. Newlines you type are kept. The output's width never changes; only its height.

Add text runs in two halves either side of Add a background: the bands are opened up before the background stage, so a fill, gradient, or image covers them along with everything else, and the type is drawn after, on top of it. A caption therefore sits on the same backdrop as the picture instead of floating on transparency. With no background on, the bands stay transparent — which means the save falls back to PNG (see Saving).

Add a watermark

Turn on Add watermark to stamp a text mark or a logo image over the finished output. A segmented control at the top switches between Text and Image mode.

Text mode

Type into the Watermark text field and pick a text color from the same built-in color picker used elsewhere in the app. Nothing is stamped until you type something.

Image mode

Choose a logo file to stamp instead of text. Once chosen, a thumbnail and filename appear with buttons to swap it for a different image or remove it.

Position, size, opacity & margin

A 3×3 grid of dots stands in for the canvas — click a dot to anchor the mark to that corner, edge, or dead center. A Size slider scales the mark as a percentage of the canvas's longer side, capped at 30% for text (any bigger stops being legible) and 100% for a logo image. An Opacity slider fades it. For every anchor except dead center — which has no edge to sit away from — a Margin slider sets how far it sits from the edge it's anchored to.

The watermark runs dead last in the pipeline, after Resize & canvas and everything above it. Its size and margin are fractions of the canvas's actual longer side, so they track the dimensions you're about to save rather than a pre-resize working image.

Unlike every other stage, the watermark's style is remembered globally: whatever text, color, logo, position, size, opacity, and margin you last used seeds every newly added image, including after relaunching ImageBuddy. The Add watermark switch itself still starts off on a new image, the same opt-in convention as every other stage — only the style carries forward, not whether it's on.

Vectorize to SVG

ImageBuddy can trace a raster image into scalable vector paths and hand you an SVG. Three modes cover the common cases:

Mode	What it produces
Silhouette	A single filled shape of the subject's outline — the whole cut-out as one solid path.
Outline	The subject's edge as a stroked path rather than a filled fill.
Full color	Every flat color region traced into its own path, reconstructing the artwork in vectors.

Once traced, you can export the SVG to a file or copy it straight to the clipboard to paste into a design tool or a web page. Vectorizing shines on logos, icons, and flat illustrations, where the paths stay clean and small.

Vectorize detects the artwork's background color the same way Trim & crop does, and discards it before tracing — including matching-color regions fully enclosed inside the artwork — so background never bleeds into the result as a filled path. This runs independently of whatever Remove background mode is set.

Split collages

If the image you drop is actually several pictures laid out on one background — a contact sheet or a collage — ImageBuddy notices. It finds the near-white or transparent gutters between the pieces and offers to split the collage into separate files, cropping each piece out and saving them to a folder you pick. This is a one-click operation, separate from the trim / key / stylize pipeline.

Saving

Two save buttons sit at the bottom of the settings sidebar:

Save Over Original — write the processed image back in place of the file you dropped.
Save As New… — write to a new file and leave the original untouched.

ImageBuddy preserves the source file's format when it can. But keying, stylizing, padding, and Add text's bands all introduce transparency, and formats like JPEG and BMP can't hold an alpha channel. When the output has real transparency, ImageBuddy falls back to PNG — a Save Over Original on a JPEG writes a sibling .png and re-points the item at it, so nothing lossy silently flattens your cut-out.

Choosing a format

Save as New…'s panel carries its own format controls. A Format menu offers PNG, JPEG, TIFF and HEIC, plus whatever the original was, so keeping the format you opened is always available. Picking a lossy container reveals a Quality slider. The menu lists only the formats your Mac can genuinely write, so nothing silently falls back to something else behind your back.

The default selection is whatever ImageBuddy would have chosen on its own — the source's format, or PNG when the settings added transparency it couldn't hold — so you can ignore the controls entirely and get the old behavior.

Choose a format that can't hold an alpha channel while the image actually has transparency, and the panel says so before you commit: Transparency will be flattened in this format. It's judged on the image about to be written, so a cut-out you opened as a transparent PNG is caught even with every stage switched off.

The exception is an opaque Add background fill: because it covers every transparent pixel, there's no alpha to preserve, and the save stays in the original format. A translucent background still triggers the PNG fallback.

Copy it, or drag it out

You don't have to save to a file at all. ⌘C copies the finished image to the clipboard — as PNG, so a cut-out keeps its transparency — ready to paste into a chat, a slide, or a design app. You can also drag straight out of the preview into another app or a Finder window: the drop arrives as a real file, named the way Save as New… would name it. Both carry exactly what a save would write, including the Vectorize trace when that stage is on (there they hand over the SVG).

The batch queue

Drop several images at once — or a whole folder — and ImageBuddy queues them in a sidebar on the left, so you can work through them one at a time. Each image remembers its own settings as you switch between them, so nudging the trim tolerance on one image doesn't touch the others. The queue drawer appears automatically once more than one image is loaded.

To apply the same settings to the whole queue instead, turn on Sync all, shown in the sidebar once more than one image is loaded. With it on, changing a setting updates every queued image at once, matching ImageBuddy's original behavior before per-image settings existed. It only covers pipeline settings — exclusion marks are painted pixel data, not a setting, so they never sync and a new image never inherits another's marks.

One more thing stays with its own image even under Sync all: Add text's title and caption. How the type is set — font style, color, alignment, sizes, padding — follows the batch as you'd expect, but the words don't, since one image's caption is never the right caption for the next.

Languages

ImageBuddy is localized into English, Deutsch, Español, Français, Italiano, 日本語, 한국어, Português (Brasil), Русский, and 简体中文. It picks the right language automatically based on your System Settings → General → Language & Region preferences.

To pin ImageBuddy to a specific language without changing system preferences, open System Settings → General → Language & Region → Applications, click the + button, and pick ImageBuddy.

Support

Questions or bug reports? Email ben.dansby@gmail.com and you'll get a reply, usually within a day or two. For privacy questions, see the privacy policy.