The quick answer
This tool turns one flat design into separate, editable parts. Words become real text in the same place, size, color and weight. Photos stay sharp images, logos and flat shapes become vector curves, and the background is saved on its own. Everything comes back as one layered PDF that CorelDRAW opens with every part selectable and Canva imports as editable elements, plus a layered SVG and every piece as a file.
How to ungroup an image or PDF
- Drop a JPG, PNG, WebP or PDF, including a flattened PDF exported as a single picture.
- Keep “Make text editable” ticked so words are read as text. For a PDF, enter the page number.
- Select Ungroup and compare the separated preview with the original.
- Download the editable PDF and open it in CorelDRAW (File › Import) or upload it to Canva. Use the SVG for Inkscape or Illustrator.
What becomes what
Each piece is sorted by what it looks like. A piece with many tones, such as a portrait, stays a bitmap so it is not distorted. A flat piece that reads as words becomes live text; a flat shape with words on it, such as a button or label pill, keeps its shape as a vector and gets the words on top as text. Other flat pieces, such as logos, icons and QR codes, are traced into vector curves. Words on one line are joined into a single text line.
How the editable text is matched
Text is read with Tesseract, an open-source text recognition engine that runs in your browser. Its position comes from the ink on the page, its size from the height of the letters, its color from the middle of the strokes and its weight from stroke thickness. The PDF uses Helvetica or Helvetica Bold stretched to the original line width, so swap in your brand font in CorelDRAW or Canva after importing.
PDFs that already have text
When a PDF still contains real text, images and drawings, those are taken directly instead of guessed. Text keeps its exact wording, color and weight, and the SVG names the original font, such as Montserrat, so an editor with that font installed uses it. Embedded photos are saved at their stored resolution and drawings are traced into a Graphics layer.
Where it stops
Text recognition reads English letters and numbers; Urdu or other scripts are kept as shapes. Very small, decorative, curved or heavily outlined lettering may be kept as a shape instead of text, and every recognized line should be checked for spelling. Pieces that touch each other come out as one piece, and the background behind removed objects is filled in from the surrounding color, so check it before reusing it on its own.
Private by design
The conversion runs in your browser with JavaScript, the open-source ImageTracer library and, for PDFs, Mozilla’s open-source PDF.js renderer. Files are not uploaded, stored or logged. Read the privacy policy for details about ordinary website requests.
By Shahab Uddin · Updated