Turn a .jpeg photo or logo into editable SVG paths
Spelled .jpeg or .jpg, the file is identified by its bytes and traced once.
Convert and preview without an account. Downloading is free too — it just needs a Google sign-in. No card, no subscription.
01 · 5 colours · 193 nodes
How it works
- Drop your file on the upload area. Either spelling works — .jpeg and .jpg are treated identically, and .png and .webp are accepted too.
- Set the colour count. JPEG compression invents colours that were never in the artwork, so start near 16 and raise it only if a real detail goes missing.
- Keep SVG ticked. PDF, EPS, DXF and PNG can be produced in the same run; pick more than one and a ZIP of the set is produced as well.
- Look at the preview — a white-backed PNG, 800 px on its longest side — and adjust the colour count if the shapes are not what you wanted.
- Sign in with Google, which is free, to download the vector files.
Two files can hold the same picture and still disagree about their own name: one ends in .jpeg, the other in .jpg. No second format hides behind the shorter spelling. JPEG is the compression scheme named after the Joint Photographic Experts Group, and both suffixes point at the same kind of bytes.
The split is a leftover from MS-DOS. The FAT filesystem gave a file eight characters for its name and exactly three after the dot — the "8.3" rule — so four-letter .jpeg was clipped to .jpg, while Unix and the classic Mac never had to clip anything. Decades on, cameras, scanners and every "Export as…" dialog still pick a favourite, and one folder happily holds both spellings.
Rasterless takes either. The two sit on the same accepted-extension list as .png and .webp, so nothing needs renaming first. Nor is the suffix what decides how a file is read: once the name has cleared that list, the opening bytes are examined and the real format wins the argument — JPEG data starts FF D8 FF, PNG 89 50 4E 47, WEBP RIFF then WEBP. So an image saved out of a browser as picture.jpeg while actually carrying PNG data is decoded as the PNG it is, and traced correctly.
What actually happens
- Raster 6×6 · 8 px · PNG 77 B
- 6 nodes · 1 path
- SVG 91 B · vector, not pixels
What a .jpeg carries into the SVG besides its spelling
EXIF orientation is applied before the .jpeg is traced
A phone writes the sensor's pixels as they came and records the rotation in an EXIF tag. That tag is applied before anything else happens, so a .jpeg shot in portrait comes back upright in the SVG rather than lying on its side with width and height swapped. Files that have passed through a chat app usually arrive with the tag stripped and the pixels already rotated, which traces the same way.
Where the edge is placed inside a .jpeg's blurred ramp
Every boundary is first found on the pixel grid, then slid along its normal to the point where the colour is halfway between the two sides. A JPEG widens that ramp and adds ringing on one side of it, so the halfway crossing is still close to the original edge but wobbles by a fraction of a pixel from point to point. The tangent is averaged over a window of neighbours, which smooths most of that wobble before fitting.
Progressive, baseline, quality 95: all invisible after decoding
The tracer never sees how a .jpeg was encoded. Baseline and progressive files decode to the same pixels; a quality setting leaves no number behind, only the size of the halo and the depth of the block steps it allowed. So two files of the same logo, one saved at 95 and one at 70, trace with the same settings and differ only in how much debris the colour count has to absorb.
CMYK and greyscale .jpeg files are converted to RGB first
A .jpeg exported from print software can be CMYK, and a scan can be a single grey channel. Both are converted to RGB before clustering: grey becomes three equal channels, so its clusters split along lightness only, and CMYK goes through the plain formula rather than a colour-managed one, so expect a visible shift and recolour the paths in your editor afterwards. The geometry is unaffected either way.
Choosing the colour count
A .jpeg is where flat artwork ends up when nobody was thinking about the next step: a logo emailed as a photo attachment, a badge scanned, a label photographed off the shelf. What survives is a grid of pixels that blurs when enlarged, refuses a new brand colour, and cannot be handed to a cutter.
Tracing turns those pixels back into shapes. An SVG stores filled outlines and curves rather than a colour per pixel, so one file stays crisp on a business card and on a trade-show banner, any region is recoloured by editing a fill value, and the outlines are geometry a cutter can follow. Illustrator, Inkscape, Figma, CorelDRAW and Cricut Design Space all open SVG directly.
Be realistic about the source. A JPEG of a logo, badge or two-tone stencil traces cleanly: those images are a handful of flat areas already. A holiday snapshot will not come back as a photograph; it becomes a poster-style reduction in the shape of the scene. If an untouched PNG export still exists, start from that — a JPEG is one lossy generation away from whatever made it.
Settings we suggest
Colour count is the control that matters most here, and compression is why. JPEG discards detail in 8x8 blocks, leaving faint ringing along hard edges and blocky mottling inside areas that ought to be one flat tone. A colour-segmentation tracer cannot know those wobbles are an artefact — they are simply pixels of a slightly different colour, so each gets its own shape. Start around 16 for a flat logo or badge, drop to 2 for a clean silhouette, and climb toward the ceiling of 64 only when a real detail is missing.
Draw style: fill_shapes gives solid filled regions and suits logos and labels; stroke_shapes draws those regions as outlines; stroke_edges keeps boundaries alone, for line art and pen plotters. Shape stacking: cut-outs writes each region as its own non-overlapping shape, so recolouring one never exposes another beneath it; stacked lays them over each other. Group by colour puts one SVG group per colour. Leave the size in mm empty and the SVG carries pixel dimensions; fill it in for physical width and height, which apply to all four vector formats. Otherwise 96 pixels is 25.4 mm.
Up to 25 MB and 40 megapixels, one image per conversion. Anything above 1.5 megapixels is downscaled before tracing, so a full-resolution phone .jpeg is reduced first and the output's pixel size follows. Simple logos usually finish quickly; detailed drawings and photos can take several minutes. Each conversion has a 300-second limit; if it is not going to finish in time, the converter tells you and suggests a smaller image. The engine combines geometric tracing (colour segmentation, an anti-aliasing model and curve fitting) with machine learning (AI). It is free and adds no watermark: previewing is open to everyone, and downloading the vector files takes a free Google sign-in. Download links stop working after 1 hour; uploads and generated files are cleared by an automatic 1-day deletion rule.
Questions
Is a .jpeg file different from a .jpg file?
No. They are one format with two spellings of the same extension, and they decode identically. Renaming one to the other changes nothing inside the file, and both are accepted here.
Why do two spellings exist at all?
MS-DOS and the FAT filesystem allowed only three characters after the dot — the "8.3" filename rule — so .jpeg was shortened to .jpg. Systems without that restriction kept the longer form. Windows dropped the limit long ago; the habit on both sides did not.
My file is named photo.jpeg but is really a PNG. Will it still convert?
Yes. After the extension is checked against the accepted list, the opening bytes are read and they decide what the file actually is. A mislabelled PNG is decoded as a PNG and traced normally. Only a file whose bytes match none of the accepted raster signatures is turned away.
My photo ends in .jfif or .jpe. Can I use it?
Not under that name. The accepted list is .png, .jpg, .jpeg, .webp, .bmp, .gif and .avif, and anything else is rejected before the contents are looked at. Since .jfif and .jpe hold ordinary JPEG data, renaming to .jpeg is enough.
Can I get a transparent background out of a .jpeg?
No, and the tracer is not the reason: JPEG has no alpha channel, so every pixel is opaque and the white behind your subject becomes a filled shape like any other. Transparency survives only when the source already carries alpha, as a PNG or WEBP can.