Vectorito

A lean macOS menu-bar app that turns any image into a clean SVG — and lets AI pick the settings for you.

Drop an image onto the Vectorito popover; it reads the picture, chooses the best vectorization settings, and shows a before/after slider. One click downloads the SVG. When you want control, a little analog console of rotary knobs is one tap away.

1 · Drop an image

2 · AI picks the settings

3 · Twist the knobs

Install

brew install --cask jjanisheck/tap/vectorito

Or download the latest .app from Releases. Vectorito is unsigned — if Gatekeeper blocks it on first launch, right-click the app and choose Open (or run xattr -dr com.apple.quarantine /Applications/Vectorito.app). It lives in the menu bar (the mustache icon) — there's no Dock icon.

How it works

image  →  AI reads it & picks settings  →  quantize to N flat colors  →  vtracer  →  SVG
            (OpenAI vision, optional)         (median-cut)               (bundled CLI)

• The AI (an OpenAI vision model) looks at the image and chooses the trace parameters (colors, noise, corners, color vs. B&W) and writes a one-line summary. It does not draw the vectors — it only tunes the dials.
• vtracer does the actual raster-to-vector tracing. The 0.6.4 binary is bundled, so there's nothing to install.
• A quantize-first step flattens the image to N flat colors before tracing — that's what gives a true "Colors: N" control (vtracer has no native color-count parameter).

The AI is optional: with no key set, Vectorito traces with sensible defaults.

Features

• Lives in the menu bar (mustache icon) as a compact popover — no Dock icon, no window.
• Drop or choose an image — works with files from Finder and image data from browsers, Preview, or screenshots.
• An animated "reading" phase while the AI + tracer work.
• A before/after compare slider with scroll-to-zoom (1–8×) so you can inspect how curves trace versus the original pixels.
• An analog console: rotary knobs (Colors, Noise, Corners, Tightness) + a Color/B&W rocker, edge style, and color layering. Twist to refine — the preview updates live.
• Download SVG named after the input (logo.png → logo-traced.svg).
• Per-image token-usage and estimated-cost logging.

Requirements

• macOS 13 or later, Apple Silicon.
• vtracer 0.6.4 is bundled — nothing to install.
• (Optional) An OpenAI API key to enable AI auto-tune.

Build

Requires XcodeGen (brew install xcodegen):

xcodegen generate
xcodebuild -project Vectorito.xcodeproj -scheme Vectorito \
  -configuration Debug -derivedDataPath build CODE_SIGNING_ALLOWED=NO build
open build/Build/Products/Debug/Vectorito.app

Or run xcodegen generate and open Vectorito.xcodeproj in Xcode. The project targets macOS 13+ with the App Sandbox disabled (the app launches the bundled vtracer process and reads dropped files). The Vectorito icon appears in your menu bar — click it.

The Xcode project is generated from project.yml; never hand-edit the .xcodeproj.

AI & privacy

• Set your OpenAI key in the console (gear → More options → OpenAI). It's stored in the macOS Keychain — never written to disk in plaintext.
• Your image is sent to OpenAI's API only when a key is set, and only to choose the settings. With no key, the app makes no network calls.
• The model id and the price-per-million estimates are editable in the same panel.
• Token usage and estimated cost are logged locally to ~/Library/Application Support/Vectorito/usage.jsonl.

License

Public domain — see LICENSE (Unlicense).