Skip to content
DYNAMB

Transparent Background Maker (PNG)

Turn any photo, logo or product shot into a transparent PNG — the file format that drops cleanly onto any design, listing or slide. The AI segmentation runs in your browser (nothing is uploaded), keeps the original resolution, and exports true alpha-channel transparency, not a fake checkerboard.

  • No upload — files stay on your device
  • Works offline once loaded
  • Free forever, no signup, no watermark
Drop a photo here, or tap to choose {{ copy.dropHint }}

This works best on a desktop — it will run here too, just more slowly.

Your chosen photo

One download before the first removal

{{ consentSub }}

{{ segmenting ? 'Removing the background…' : stage }}

{{ progressPct }}% — {{ progressMB }}

Original Original photo
Background removed
Photo with the background removed
Background:
  • PNG size {{ resultSizeLabel }}
  • Pixels {{ resultWidth }}×{{ resultHeight }}

{{ copy.doneNote }}

How to use this tool

  1. Add the image

    Drop in a product photo, portrait or logo. Anything your browser can open — JPG, PNG or WebP — works.

  2. Pick a model and approve the one-time download

    For logos and product shots choose Accurate (~170 MB, cached forever after); Fast (~25 MB) is the portrait-oriented lightweight option. The subject is lifted off the background automatically.

  3. Download the transparent PNG

    PNG keeps the alpha channel, so the background stays transparent wherever you place it. Prefer a flat colour? Pick one before downloading.

Questions people ask

How do I make a picture with a transparent background?

Add the picture above and the AI separates the subject from its background automatically. Download as PNG — the format that stores transparency — and the background stays see-through in any editor, marketplace listing or document.

Why does my transparent image show a white background after saving?

It was probably saved as JPG, which cannot store transparency at all. Always keep transparent images as PNG (or WebP). This tool exports PNG precisely for that reason.

Does it work for logos and product photos, not just people?

Yes — pick the Accurate model for that. The Fast model is a portrait specialist and is honestly unreliable on non-people subjects, which is exactly why the Accurate tier (a general-purpose segmentation model) exists. Very fine detail like wispy hair or semi-transparent glass remains the hard case for any segmentation AI; the preview shows you the exact result before you download.

Does this tool upload my photos?

No. The photo is processed by an AI model running inside your own browser — this site has no upload endpoint for it at all. The only thing ever downloaded is the AI model itself, once, from the Hugging Face CDN; after that it is cached on your device.

Is there a limit on how many images I can process?

No limits and no watermark. Your device does the computing, so there are no server costs for us to meter — remove backgrounds from as many images as you like, free.

Why do I have to download an AI model first?

The AI has to live somewhere, and here it lives on your device instead of on a server. You pick the size: the Fast model is roughly 28–46 MB including the runtime, the Accurate one about 190 MB. Either download happens once and is what makes it possible for your photos to stay private — the second visit loads it straight from your browser cache.

Which AI model does this use, and can I use the output commercially?

Two open-source models, your choice: Fast is MODNet (Xenova/modnet, Apache-2.0), a small portrait matting model that is at its best on photos of people; Accurate is ISNet general-use (the DIS segmentation model, Apache-2.0, run via the MIT-licensed ONNX Runtime), which also handles products, logos, animals and other objects. Both licences permit commercial use, so your cutouts are yours for shops, ads or client work.

Related tools

Free because it costs us nothing to run: your browser does the work, not our servers. Made by Dynamb Technologies — we build software for businesses.