Background removal

Older cut-out tools relied on color: they removed everything close to a chosen tone, which worked on a green screen and failed elsewhere. Modern tools use segmentation models, neural networks trained on many images with hand-drawn outlines, that recognize what a subject is. The result is a mask, and the best versions produce an alpha matte, a soft mask with values between fully hidden and fully visible, so fine edges such as hair or the fuzz of a knitted sweater blend smoothly instead of looking sawn off.

The uses are everyday ones. Online shops put products on a plain white backdrop, a designer drops a portrait of a fictional character onto a poster, a presenter cuts a logo out of a photo, and a video maker prepares a still before turning it into a clip. Output is normally a PNG or WebP with transparency, since a JPEG cannot store it. Cleaner results come from a sharp photo with good contrast between subject and backdrop and decent resolution.

Know where it struggles. Glass, smoke, a thin chain, a subject similar in color to the backdrop, or several overlapping objects can produce a halo, a missing piece or a ragged outline, and the tool may have to guess what counts as the subject. Zoom in on the edges and place the result on a background of a different color to spot leftover fringes. Edges can also be improved afterward by enlarging the image with an upscaler, or by repairing a defect with inpainting.

Resolution and preparation matter more than people expect. Photograph the subject in even light, avoid a backdrop that repeats the subject's own colors, and keep a little space around the edges so nothing is cropped. A cut-out of a ceramic mug on a plain table will usually come out cleaner than one of a glass vase against a window.

Keep the original file as well: a cut-out is a derived image, and you may want to redo it later with a different edge softness or on a newer tool.

In the Studio, background removal is one of the tools the image generator page lists for paid plans, next to upscaling and editing by instruction. The API has a matching endpoint, POST /image/background-remove, billed in credits like the other calls. NeverNo does not claim a particular cut-out model here, and the quality depends on your photo, so check the edges of your result.

Every image is checked against NeverNo's hard limits before it is processed: nothing involving minors, and no images of real, identifiable people without their consent. The images you create are not kept in a gallery on our servers: they are stored encrypted in your browser, in the database that holds your history, and the API returns images for you to save yourself. For a fresh subject on a clean backdrop, you can also generate the image with Z-Image Turbo, the default model, in about 3 seconds, as measured on Runware, our image provider.