Spotlight: Lumli, browser based image tools with no signup

What it is

Lumli is a studio of small, sharp utilities that live in a browser tab. In the maker's own words:

“Most online tools ask you to upload your files before you can use them. Lumli doesn't. It runs entirely in your browser, keeping your files on your own device. Edit images, manage PDFs, generate QR codes, and discover your next travel destination with no accounts, no ads, and no server uploads. Powered by WebGPU and WASM. Private by design.”

On the image side: compress, resize, crop, upscale, remove backgrounds, erase objects, extract a color palette, and convert between HEIC, AVIF, WebP, PNG, JPEG and SVG. On the document side: edit, merge, split and compress PDFs. Alongside those sit a QR code generator and a passport photo tool that crops to official spec. There is also an AI travel matcher that suggests a destination to visit.

No account. No subscription. No file picker that hands your image to a server before it will help you.

Three of the tools are Lumli's local AI, and they are where it gets clever. Background removal runs a segmentation model that decides, pixel by pixel, what counts as the subject. Object removal runs inpainting, so when you delete something a model paints in what it thinks was behind it. Upscaling runs super resolution, inventing detail that was never in the original file. None of these are filters. Each is a neural network making a judgement call about information it cannot see, and getting it right well enough to be genuinely useful.

How you remove an image background without uploading it

The normal path for all three AI tools is upload, wait for a server GPU, download the result. Lumli sends the model weights down to your browser instead, then runs the inference on your own hardware, using WebGPU where your machine supports it and WebAssembly where it does not.

The trade shows up on first load, because you are downloading a model rather than a web page. After that it is local. Install it as a progressive web app and it keeps working with the wifi off. That falls straight out of the design: there is nowhere for the file to go.

That architecture is newly practical rather than newly imagined. WebGPU reached the major browsers, task specific models shrank to a size worth shipping over a connection, and quantization got good enough to make the download tolerable. A segmentation model that only has to separate a subject from a backdrop does not need to be large, it needs to be good at one thing. Three years ago this was a conference demo. It is now a product you can bookmark.

Why the community picked it

The pitch lands in about four seconds, because everyone has felt the specific irritation of a free background remover asking for an email address before it will return a file it processed on a server you cannot see. Lumli removes that step, and the feed responded to it, marking the listing both useful and interesting.

That is the whole draw. These are tools people already reach for, and a version with no signup and nothing to upload is simply better than the one they already use.

Who it's for

Lumli is made by Makan, and it has the focus of a single-maker tool: built for one person's own needs rather than a team's workflow.

The focus is the point. If you are one person cleaning up a few images or preparing a PDF, Lumli does exactly what you need with nothing in the way. Teams that need shared accounts or an API are a different audience.

Common questions

About Spotlight

The top product each week and each month earns a Spotlight badge, decided by community reactions. It is not an editor's pick. The feed decides. The badge shows on the feed card while it is current and stays permanently on the product's listing. See past winners.