Background Remover — Remove an Image Background to a Transparent PNG
Part of Image tools: See all Image tools.
What is Background Remover?
Background Remover cuts the subject out of a photo and returns a PNG with a transparent background. It uses an AI segmentation model (BiRefNet, portrait version) that is tuned for people, so headshots and portraits give the cleanest edges; products, pets and other single, clear subjects usually work too, but check the edges. Upload one image for a single PNG, or up to 20 for a ZIP. The output keeps the original size unless the long side is over 2048 px, in which case it is scaled to 2048 px. Phone photos are turned upright first. Images are uploaded over HTTPS, processed for that request and then discarded.
How to use Background Remover
- Open Background Remover and upload a JPG, PNG or WebP image, or select several (up to 20).
- Click Remove Background & Download. One image takes roughly 10 to 30 seconds; a batch takes longer.
- One image downloads as name_nobg.png, with a before and after preview. Several images download as removed_backgrounds.zip.
- Zoom in on the edges (hair, fingers, product corners) before you use the cut-out. For a white background instead of transparency, convert the PNG to JPEG with Image Converter.
Why use this tool?
Product listings, profile pictures and slides often need the subject without the room behind it. Doing that by hand takes a long time; this tool does it in one step and returns a real transparent PNG you can place on any background. It is not a photo editor: there is no brush to fix the mask, no background colour picker and no shadow tool.
When it matters: Background removal is great for product photos, portraits, and logos where you need clean transparency. Download the PNG output and place it on your design canvas.
If the edges look uneven, try again with a different input image (higher resolution usually helps) and ensure your subject is well lit.
What "removing the background" actually means
When you remove the background from an image, you are not erasing pixels in the everyday sense — you are deciding, for every single pixel, whether it belongs to the subject you want to keep or the background you want gone, and then making the background pixels transparent. That decision is called segmentation, and it is genuinely hard because the line between subject and background is rarely as clean as it looks to your eye. A person's hair is thousands of fine strands with background showing between them; the fuzzy edge of a sweater is part subject, part background blended together; a glass or a wisp of smoke is partly see-through. A good background remover is judged almost entirely on how it handles these awkward edges, because the flat middle of a subject is trivial and the edges are where the work — and the failures — happen.
Why modern removal works without a green screen
For decades, cleanly separating a subject from its background required either painstaking manual masking — tracing the outline by hand, pixel by pixel — or a controlled studio setup with a green screen, where the background was a single known colour the software could key out. What changed is machine learning. Today's background removers are trained on enormous collections of images where humans have already marked the subject, so the model learns the general shape of "a person", "a product", "an animal" and can find that shape against backgrounds it has never seen before. This is why you can drop in a casual phone photo taken against a messy room and get a clean cut-out: the tool is not matching a colour, it is recognising the subject. The practical upshot is that the cleaner and more distinct your subject is from its surroundings, the better the result — strong separation between subject and background gives the model an easy decision.
The two jobs people use it for
Almost all background removal serves one of two ends. The first is e-commerce and product imagery: marketplaces from Amazon to Etsy expect product photos on a clean white background, and a consistent cut-out across a catalogue makes a shop look professional rather than thrown together. The second is profiles, graphics, and composites: lifting a person or object out so it can be placed on a new background, dropped into a presentation, turned into a sticker or logo, or given a transparent backdrop for a website header. These two jobs want different outputs — the shop owner usually wants the subject on solid white, while the designer wants true transparency to composite later — which is why the format you export to matters as much as the cut itself.
Transparency is why you export to PNG, not JPEG
This is the single most common mistake, so it is worth stating plainly: if you remove a background and then save as JPEG, the transparency is destroyed and the empty area fills with solid colour — almost always white. JPEG has no concept of a transparent pixel; it simply cannot store one. To keep a genuinely transparent background you must export to a format with an alpha channel — PNG is the universal choice, and WebP also supports it. Only choose JPEG if you have deliberately placed the subject on a solid background you are happy to bake in. If your cut-out looks perfect in the tool but arrives with a white box around it when you use it elsewhere, the cause is virtually always that it was saved as JPEG somewhere in the chain. Reach for our image converter if you need to move between formats while preserving that transparency.
Getting the cleanest cut: what helps the model
You can dramatically improve results by giving the remover an easy image. Contrast helps most: a subject that stands out from its background in colour and brightness is far easier to separate than one that blends in — a person in a dark coat against a dark wall is a hard case, the same person against a light wall is easy. Even, diffuse lighting avoids harsh shadows that the model may mistake for part of the subject or leave clinging to the edges. Sharp focus on the subject gives crisp edges to cut along, whereas a motion-blurred or out-of-focus edge has no clean line to follow. And resolution matters: a larger, clearer image gives the model more detail to work with at the edges, so start from the best original you have rather than a small thumbnail (this tool works at up to 2048 pixels on the long side and scales bigger photos down to that). None of this requires a studio — it just means choosing the better of two photos when you have a choice.
Where automatic removal still struggles
Honesty about the limits saves frustration. Fine hair and fur remain the classic hard case — automatic tools have improved enormously but can still leave a slight halo or lose a few flyaway strands, and for a hero image you may want to touch up the edge by hand afterwards. Transparent and semi-transparent objects — glass, sunglasses, veils, smoke — confuse the subject-or-background question because the answer is "both", and results vary. Busy edges where the subject's colour nearly matches the background can produce a wavering outline. And multiple subjects or ambiguous scenes — is the bag the person is holding part of the subject? — force the model to guess at your intent. The model this tool uses is tuned for portraits, so people give the cleanest edges, and products or pets usually work but deserve a closer look. For a clear single subject against a distinct background, automatic removal is now very good; for these edge cases, treat the automatic cut as a strong starting point rather than a guaranteed finish.
Fitting it into your workflow
Background removal is usually a step, not the whole task. A typical product-photo sequence is: remove the background, place the subject on clean white (converting the transparent PNG to JPEG with the image converter does exactly that) or keep it transparent, then resize to the marketplace's required dimensions and compress so the listing loads fast. For a transparent graphic destined for a website, remove the background, keep PNG transparency, and resize to the display size. Either way, do the background removal early — while you still have the full-quality original — and save the resizing and compression for last, so every later step works from the cleanest possible cut. Done in that order, you get a sharp, correctly-transparent subject that drops neatly onto whatever background comes next.
Frequently asked questions
- What kinds of photos work best?
- One clear subject that stands out from the background, in focus and evenly lit. The model is trained mainly on people, so portraits work best. Products and pets usually work; glass, smoke, fine fur and a subject the same colour as the background are the hard cases.
- What format and size is the result?
- A PNG with a transparent background, the same size as your image. If the long side is over 2048 px the image is scaled down to 2048 px first to keep processing time reasonable.
- How do I get a white background instead of transparent?
- Open the PNG in Image Converter and convert it to JPEG. JPEG cannot store transparency, so transparent areas become white.
- Can I remove backgrounds from several images at once?
- Yes. Select up to 20 images and they come back as one ZIP of PNG files. Each image takes about 10 to 30 seconds, so a full batch can take several minutes. Images that fail are left out of the ZIP.
- Will my phone photo come out sideways?
- No. The orientation your phone stored in the photo is applied before the background is removed, so the PNG is upright.
- Is my image stored?
- No. It is uploaded over HTTPS, processed for your request and discarded. There is no account and no gallery. Only upload photos you have the right to edit.
Step-by-step guides
Also try
Related tools that work well with this one: