BGVanish / GuidesLOADING & TROUBLESHOOTING

Background remover loading slowly? What to check

A slow first result can have a different cause from a stalled image-processing job. Start with the stage shown in the editor, then check the connection, file and browser before changing anything else.

By BGVanish · Updated

The quick answer

Check whether the editor is preparing, downloading or removing. First use loads about 18 MB; watch the received-byte counter, keep the tab open, and use Cancel plus the example to isolate an image-specific problem.

  1. Read the stage and check whether received bytes are increasing.
  2. Keep the browser foregrounded on a stable connection.
  3. If stuck or errored, Cancel and test the built-in example.
  4. Check format, 10 MB and 40 MP limits before retrying a compatible resized copy.
Try the remover or its example

Photos are processed on your device. The first run downloads the processing engine; speed depends on your device. Exports are up to 2,048 pixels.

Read the current stage before retrying

Prepare reads and decodes your selected photo. Load downloads the removal engine and model. Remove runs the model and creates the cutout. These steps use different resources, so a slow download is not the same problem as slow processing on the device.

First use loads about 18 MB of processing files. If the received-byte counter is increasing, the download is progressing. The elapsed timer reports how long the job has been running; it is not an estimate of time remaining. There is no fixed completion time because connection and device speed vary.

If loading is slow, check the connection

Keep the page open on a stable connection. A restricted network, interrupted connection, or browser setting that prevents loading the processing files can stop the engine from becoming available. Do not repeatedly refresh a download that is still progressing.

If the page reports an error or the counter stops progressing for a prolonged period, use Cancel and try the built-in example after checking your connection. If the example also fails, the issue is less likely to be your photo. A normal up-to-date browser is a useful alternative when an in-app browser interrupts file loading. Cancel affects the active job; it does not erase files already saved to your device.

If preparing the image fails, check the source

Use a real JPG, PNG or WebP file up to 10 MB and 40 megapixels. A phone HEIC image, damaged download, or file whose extension was changed without conversion may not decode correctly. Export a compatible copy from an image app instead of renaming the original.

Very large camera images may exceed the pixel limit even when their compressed file size is small. Resize a copy while keeping the original. BGVanish limits the longest output side to 2,048 pixels, so loading a far larger source will not produce an unlimited-resolution result. A smaller input may reduce preparation and memory pressure, but it does not change the fixed model inference size.

If removal is slow, check device and tab activity

The model runs on your device, so an older phone or busy browser can take longer. Close unrelated heavy tabs if appropriate and keep the editor in the foreground. Switching apps can suspend a mobile tab. Automatic removal quality and processing time are separate issues: retrying does not guarantee a cleaner edge.

If a job remains stuck, Cancel before trying again. Reload only when needed because it discards the current image and unsaved corrections. Once a result appears, inspect it and save a PNG or JPG before leaving the page. If the same photo repeatedly fails but the example succeeds, try a compatible resized copy and note the exact error message rather than assuming the website is offline.

Common questions

Is the timer a countdown?

No. It shows elapsed time, not an estimated finish time.

Will a smaller photo always make removal faster?

It can reduce preparation and memory use, but model inference uses a fixed processing size. Download speed and device speed still matter.

Give it a try.

A clean result starts with your next photo.

Try the remover or its example