How to convert an image to SVG
A raster image, a JPG or PNG, is a fixed grid of pixels. Blow it up past its own resolution and it goes soft. An SVG describes the same picture as shapes and coordinates instead, so it scales to any size, a favicon or a billboard, without losing a single edge. Getting from one to the other is called vectorizing. It works very well on some images and badly on others, and it's easy to tell which you have before you start.
Why a logo needs to be a vector eventually
A logo drawn once as a JPG looks fine on the page it was made for and rough everywhere else, a business card, a large-format banner, a favicon shrunk down to sixteen pixels. Each of those needs a different pixel size, and a raster file only really has one size it looks right at.
An SVG sidesteps the problem entirely, because it isn't a grid of pixels being stretched, it's a set of shapes being redrawn at whatever size is asked for. The same file works as a sixteen pixel favicon and a two metre banner, because there's no fixed resolution to run out of.
What traces well, and what doesn't
Flat graphic art traces cleanly: logos, icons, simple illustrations, anything built from a small number of solid colours with clear edges between them. The tracer's whole job is finding those edges and describing them as shapes, and clean, high-contrast edges are exactly what it's looking for.
A photograph is the opposite case. It shades gradually between millions of colours with no hard edges anywhere, and forcing that into flat vector shapes produces a blocky, posterized result rather than anything resembling the original. If what you're starting with is a photo rather than flat art, this isn't the tool for that job, and no amount of adjusting the detail level fixes it.
Choosing a detail level
Low folds similar shades together the hardest and keeps the fewest colours, the right pick for a logo built from two or three flat colours. Medium is the default and suits most logos. High holds on to subtle shades, which helps when the shading matters, but it also keeps texture and faint background patterns you may not want.
Start at Medium and look at what comes back. A colour that got merged into its neighbour means go up a level. Stray specks or extra shades mean go down.
Preparing an image so it traces cleanly
Size matters more than any setting. Find the largest original you can, the file from whoever designed the logo, or the version on a company website rather than a social media thumbnail. Text and thin lines need several pixels of height in the source to survive. Anything only a few pixels tall has already blurred into its background.
If a small copy really is all you have, the tool's "Sharpen small images with AI first" option helps. On images up to 512 pixels it redraws the picture at four times the size with an AI model before tracing, which is often enough to bring tiny lettering back. It uses your graphics card through WebGPU, and the model is a one-time 36 MB download. The upscaling guide explains how that kind of AI enlargement works.
Pick PNG over JPG when you have the choice. JPG compression smears faint colour noise around every edge, and while the tracer cleans most of it up, it's better not to start with it. A PNG with a transparent background also comes out with a transparent background.
Crop tight around the logo first with the crop tool, because a border or a strip of leftover background becomes a shape of its own. A screenshot works fine as long as it was taken with the logo shown large, not zoomed out.
How to convert an image to SVG in your browser
Open the image to SVG tool, pick a detail level, and drop your image. It's decoded, traced and handed back as a downloadable SVG file in the same step.
Images over 1200 pixels on the long edge are scaled down first. Tracing runs pixel by pixel on your device, and anything bigger adds time without adding accuracy. Nothing about this needs a server, disconnect from the internet after the page loads and it still works.
What the output actually looks like next to a hand-drawn vector
A designer tracing a logo by hand in vector software decides exactly where each curve sits, and produces a small number of deliberate anchor points. An automatic tracer instead follows whatever pixel edges the source image actually has, which usually means more anchor points and small irregularities a human wouldn't have drawn in.
For getting a rough logo into a scalable format quickly, that's a reasonable trade. For a brand asset that's going to print at a large size or needs to be edited cleanly afterward, a properly hand-traced vector from a designer will still look better and be easier to work with.
Editing the result is simpler than it sounds, though. Every colour comes out as one shape, so changing a colour in Inkscape or Illustrator is a single fill change rather than a hunt through hundreds of fragments.
A quick way to tell if vectorizing will actually help
Look at the source image and count how many distinct flat colours it really has. A handful of solid colours with sharp edges between them, that's a good candidate. Smooth gradients, shadows, or anything photographic, that's not going to trace into something usable, no matter which detail level gets picked.
Quick questions
Can I vectorize a photo?
Not usefully. Photos shade gradually between colours with no hard edges, and tracing flattens that into blocky, posterized shapes. This tool is built for logos, icons and flat graphic art, not photographs.
What's the difference between the detail levels?
Low merges similar shades for the fewest colours and the smallest file. Medium is the default for most logos. High keeps subtle shading and texture, producing a larger file.
Why did the small text in my logo come out broken?
It was probably only a few pixels tall in the source, which blurs it into the surrounding colour before tracing starts. A bigger copy of the logo is the real fix. Keeping the AI sharpening option ticked often rescues small text on images up to 512 pixels, since it redraws the edges before the tracer sees them.
Is a PNG or a JPG better for tracing?
PNG. JPG compression adds colour noise around edges, and a transparent PNG background stays transparent in the finished SVG.
Is my image uploaded anywhere to convert it?
No. Everything happens in your browser. Disconnect from the internet after the page loads and it still works.
Why does the tool shrink my image before tracing it?
Tracing runs pixel by pixel on your device, and a full-resolution source adds processing time without improving the result, since it all gets flattened into solid shapes regardless. Images are capped at 1200 pixels on the long edge first.
Will the SVG look exactly like a professionally designed logo file?
Probably not as clean. An automatic tracer follows pixel edges rather than making a designer's judgment calls, so it can add more anchor points than a hand-built vector would. It gets you something scalable quickly. A proper redesign is still a separate job.
Can I convert several images at once?
Yes, drop a batch and each one is traced at the same detail level and returned as its own SVG file.
What formats can I convert to SVG?
Any format this site's other tools accept: JPG, PNG, WebP, AVIF, HEIC, BMP and GIF. The output is always SVG.
Why does this matter for a favicon?
A favicon gets shrunk down to as little as sixteen pixels. Starting from a vector rather than a raster image means that shrink happens without the soft, blurred edges a bitmap would show at that size.
Converters
The converter on this page handles every conversion this guide talks about.