JPG
SVG
SVG
ЗАГРУЗИТЬ
ФАЙЛ
Ожидание файла...
Превью
Ожидание файла...
Tool guide
You start from a JPG — a camera shot, a messenger download, a stock photo — and you end with an SVG made of curves, computed entirely inside the tab. The source matters: JPEG is lossy, so contrasty edges carry ringing and 8x8 block noise, and the tracer dutifully turns that noise into extra shapes. A signboard photographed on a phone therefore needs far fewer colours than a clean export from a design app. See also: vectorise a logo starting from PNG, prepare a contrasty black and white before tracing, keep a raster copy as WebP, the other direction: rasterise an SVG to JPG at a chosen size.
The format itself. JPEG encodes an image in 8x8 pixel blocks, so sharp boundaries pick up ringing: faint halos and patches in slightly different shades. To the algorithm those are genuine colour regions and get outlined just like the letters of a logo. Drop the colour count to 4-6 and the neighbouring shades collapse into one, taking the junk shapes with them.
It becomes one large filled rectangle sitting under everything else. JPEG has no alpha channel at all, so transparent pixels simply cannot exist in the source and the background is just another colour. In Illustrator or Inkscape you select that bottom object with a single click and delete it, leaving the mark on an empty canvas.
As a rough base yes, as a finished file almost never. A photograph carries perspective, uneven lighting, glare and relief shadows, and all of that flows into the paths together with the actual shape. The sensible workflow is to trace the shot for approximate proportions and then redraw the mark by hand with clean curves on top of it.
Colours 2 and detail 4. Line art has only two meaningful tones, ink and paper; everything else is scanner noise and the yellow cast of the sheet. Two colours cut those out immediately, while high detail keeps thin strokes from rounding off or breaking apart. If lines still fall apart, raise the contrast of the scan before uploading it.
The weight comes from the two sliders, not from the megapixels. The frame is scaled down on its longest side before tracing, so a 12 MP and a 2 MP source end up broadly similar. At 6 colours and detail 2 expect a few hundred kilobytes; at 24 colours and detail 5 the file easily passes several megabytes and outgrows the original JPG.
It draws no curves whatsoever: the photograph is placed inside the SVG file untouched and only the extension changes. That helps when an upload form accepts .svg exclusively yet the picture has to stay a photograph. You can spot the difference by the file weight and by the fact that zooming still blurs it exactly like a JPG.
Before: shop-sign.jpg, 900×900, phone snapshot of a sign, recompressed down to 120 KB by the messenger
After: Colours 5, detail 2 — four solid outlines for the letters and the panel; edge dirt disappeared along with the extra shades
Before: sunset.jpg, 1920×1280, smooth sky gradients and soft shadows across the field
After: Colours 12, detail 3 — the sky breaks into bands of flat tone, giving a poster rather than a faithful copy of the frame
Your rating and feedback help decide what to improve next.