1Upload the images you want cut out — a clear subject against a distinguishable background gives the model the least to guess at.
2Wait while the segmentation model predicts a per-pixel alpha matte; there is no masking for you to do by hand.
3Check the edges in the preview, particularly hair, fur, glass and motion blur, which are where any automatic matte struggles.
4Download the result as PNG or WebP so the transparency actually survives — saving a cut-out as JPG puts the white back.
Remove Image Background FAQ
How accurate is the cut-out on difficult edges?
+
Solid subjects against a distinguishable background come out excellent. Fine hair, fur, motion blur, glass and chain-link fences are where any automatic matte struggles, and where you may still want to touch up the result by hand afterwards.
Does my source format affect the quality of the matte?
+
Yes. this format is read and written by the same streaming pipeline as everything else we support. Compression artefacts around a subject are exactly what a segmentation model has to disambiguate, so cleaner input gives a cleaner edge.
What format should the result be saved in?
+
One with a real alpha channel — PNG or WebP. Saving a cut-out as JPG throws the transparency away and composites the subject onto white, which defeats the entire operation and is the single most common thing that goes wrong here.
Can I process a whole batch of images?
+
Yes. Every image is segmented independently, so a set of product shots can be cut out in one pass — which is normally the point, since doing it by hand is the slow part of a catalogue shoot.
Will Remove Image Background lower the quality of my images?
+
That depends on what you brought. A lossless source (PNG, BMP, TIFF) round-trips exactly; a lossy one (JPG, WebP, GIF) is already the product of one quantisation pass and only re-encodes if the operation genuinely requires it.
What happens to metadata and transparency?
+
EXIF and the ICC colour profile are carried across wherever the target format supports them, and an alpha channel survives into any format that has one. Transparency is only flattened when the destination format has no way to represent it.
Does Remove Image Background cost anything?
+
No. Remove Image Background is free without an account and adds no watermark. Uploads are deleted from the workers shortly after the job completes.
Which image formats does Remove Image Background accept?
+
All of the common ones — JPG, JPEG, PNG, WebP, GIF, BMP, TIFF, SVG, ICO, PSD, JFIF and HEIC. You can mix formats freely in one batch; each file is identified from its header and routed to the right decoder.
Why does a PDF site host Remove Image Background?
+
PDF.to exists because PDF is where documents go to be shared and the worst place to keep them while they are still being worked on, so nearly every job here is about getting something into or out of that state deliberately. A document rarely arrives in the state it needs to leave in: it is too big, in the wrong order, locked, or a scan with no text in it. Remove Image Background sits on the same upload and the same account as the conversions because those jobs turn up together.
What should I do with the result once Remove Image Background is finished?
+
The converter on this site moves documents between PDF, Word, images, plain text and eBook formats in either direction, so whatever comes out of a tool here has somewhere to go next without a second upload. Running that after Remove Image Background rather than before keeps the document structure intact through the step that is most likely to flatten it.
Is Remove Image Background here the same tool the sibling sites run?
+
The engines are shared — the same document toolchain, the same workers, the same limits. What a PDF site adds is a view on when a file should stay a document and when it genuinely wants to become something else, which is the decision most tools quietly make for you. It also starts from one fact about the format this site is named after: a PDF is an object graph, not a page image, so text stays selectable and vectors stay sharp no matter what happens to the raster content inside it.
Do I need an account, and does anything get kept?
+
No account, and nothing is kept: uploads are deleted from the workers shortly after the job finishes, nothing is read and nothing is indexed. Free accounts exist for history and batch size, not for access.
