Remove an image background in your browser and check the edges

Start with a subject the tool can separate

Background removal is useful when a portrait or product needs to sit on a different design. The automatic tool has to decide which pixels belong to the subject. A single, clearly visible person or object against a distinct background gives it a more understandable task than a crowded scene.

Choose a focused image with the whole subject visible. Crop away irrelevant surrounding space if the subject occupies only a small part of the frame, while leaving its edges intact. Hair, fur, glass, translucent fabric and subtle shadows can be difficult to separate, so expect to inspect these areas rather than treating the first output as final.

Understand local processing and the first download

KitForma processes the selected photo on your device in the browser. Automatic removal downloads about 19 MB of processing files on first use, so an internet connection is needed. Later loads depend on the browser cache. Starting with manual editing skips that model download. Both modes still need the page and its code to load; neither uploads the selected photograph for processing.

The background remover accepts JPG, PNG or WebP up to 20 MB and 40 million pixels. Convert HEIC photos to JPG first. Output is a transparent PNG with its longest edge limited to 2048 pixels; larger inputs are reduced proportionally. The original file is not overwritten.

Create the cutout and inspect transparency

Leave the page open while the image is processed or edited. A busy phone may take longer than a desktop. Moving to another tool ends this editing session, so wait for the updated PNG and download a satisfactory result before continuing. Undo history is limited and does not survive leaving the page.

1. Open Image background remover and choose a supported image.
2. Run automatic background removal and allow its initial files to download, or start manual editing from the whole original image.
3. Compare the result on checkerboard, white and dark backgrounds. Zoom in, erase unwanted areas and restore parts that should remain.
4. Inspect fine edges and enclosed gaps. Wait for the latest edited PNG to be ready, then download it.

Source: 4000 × 3000 pixels
Maximum output: 2048 × 1536 pixels
Preview background: dark
Downloaded file: still transparent PNG

Refine the mask with erase and restore brushes

Choose Erase and paint over an unwanted background patch. Choose Restore to bring back the original pixels under your brush, for example a lace or finger missing from an automatic cutout. The brush changes the area you actually paint; clicking a person or object does not select its whole outline. Restore can also bring back original background pixels, so use a small brush near the boundary.

Zoom in for small corrections and use the pan control to reach another part of the image. Adjust brush size and edge softness to suit the boundary. Brush diameter refers to pixels in the exported image, while zoom only enlarges the working view. A larger view does not add detail or raise the 2048-pixel longest-edge limit.

Undo or redo a recent stroke if needed. Reset returns to the initial automatic cutout when you started with the model, or to the original image when you started manually. Restore uses the original at output resolution and preserves any transparency already in it; it cannot reveal details behind an object or make fully transparent source pixels visible. This is mask editing, not generative object removal.

Remove a connected colour patch with one click

Choose Remove region and click a plain part of the background. The tool follows adjacent source pixels whose colours are within the tolerance of your clicked colour. A disconnected patch of that same colour is left alone. This works well as a starting point for a fairly uniform backdrop; it does not identify a whole person, shoe or other object.

Start around 12% tolerance, inspect the result and Undo if matching colours in the subject were removed. Reduce tolerance before trying again. Use Restore region for a connected patch that should remain, then finish boundaries with a small brush. A transparent source pixel cannot be restored. You can also enter X and Y output-pixel coordinates and use Apply at this point without a pointing device.

Apply feather optionally softens all mask boundaries by 1–4 output pixels. Start with 1 pixel and compare white and dark previews. It can soften a jagged edge, but it can also blur fine hair; Undo restores the previous mask. It does not remove colour contamination or create missing detail.

Worked region check: two white patches separated by a red strip. Click the left white patch at low tolerance: the left patch is removed; the red strip and disconnected right white patch remain. Restore returns source pixels, including their original transparency.

Know what the preview does and how to troubleshoot

The checkerboard is only a visual indicator of transparency; it is not added to the PNG. White and dark preview backgrounds also do not change the downloaded file. They reveal pale halos, missing dark details or leftover background that may be hard to see on one surface.

If the model cannot find the subject, try a tighter crop or a clearer source photo, or start manual editing from the original and erase the unwanted regions yourself. Fine hair, fur and translucent glass may still need careful work; a restore stroke brings back source pixels rather than inventing a clean replacement edge. If the model download fails, check the connection. For memory errors, close other demanding tabs or use a smaller image; save any satisfactory result before reloading.

Keep the PNG when you need transparency. Converting it to JPEG fills the transparent area with a chosen solid color. Before placing a cutout in a product listing or design, examine the output at the intended display size and check that the actual object’s outline and details have not been unintentionally removed.