Organic Contour Vector Engine
Original Source
Click or drop an image here
Vector Output
Vector preview will appear here.
The Organic Contour Vector Engine turns a PNG or JPG into clean, scalable SVG vector artwork — without uploading anything, creating an account, or paying for a subscription.
Load an image into the box above, choose how much detail and how many colours you want, and press Generate SVG. The result appears beside the original so you can compare the two directly, and you can download it as a scalable SVG or as a high-resolution PNG. Everything below explains what is actually happening when you do that, and how to steer it.
A raster image — any PNG, JPG, or photograph — is a grid of coloured pixels. It has a fixed resolution baked in at the moment it was saved. Enlarge it and there is no more information to draw on, so the software interpolates, and edges that were sharp turn soft. Print a 400-pixel logo at poster size and you get a blurry rectangle, because the detail was never recorded in the first place.
A vector image stores something entirely different: a list of shapes, each with an outline and a fill colour, described mathematically. There is no resolution in the file at all. When you open it, the shapes are drawn fresh at whatever size is asked for, so the same file is razor-sharp on a business card and on the side of a building.
Vectorizing — also called image tracing — is the process of looking at a grid of pixels and working out what shapes would reproduce it. That is a genuinely hard problem, because photographs are made of continuous gradients and vectors are made of flat regions with hard boundaries. Every tracing engine is a set of decisions about where to put those boundaries, and the controls above are how you influence those decisions.
The conversion runs in four stages, and knowing them makes the controls much easier to reason about.
Colour quantization. The image is first reduced to a limited palette — the Color Quantization setting, anywhere from 4 to 64 tones. This is the step that decides how many distinct flat colours the finished vector can contain, and it has more influence on the character of the result than anything else.
Region finding. Every area of contiguous same-colour pixels is flood-filled and identified as a candidate region. A quantized photograph produces an enormous number of these, most of them a few pixels across.
Healing. Tiny stray regions — the specks left behind by sensor noise, JPEG compression, and anti-aliased edges — are merged into their neighbours. The Polygon Fill Size control sets how aggressive this is. It is the difference between a clean flat field and a field covered in confetti, and also, pushed too far, the difference between a smooth gradient and a hard band.
Shape assembly. The surviving regions are turned into solid vector polygons using a leak-proof scanline assembly pass — leak-proof meaning that adjacent shapes share exact boundaries, so no hairline gaps of background show through between them. In Photo / Art mode the contours keep their organic, textured character; in Flat Graphic mode they are traced and simplified into clean geometric edges instead.
The fastest route to a good result is to set the Source Type correctly and then adjust Color Quantization. Those two carry most of the outcome. Reasonable starting points:
| Your image | Source Type | Quality | Tones |
|---|---|---|---|
| Logo, sign, or text | Flat Graphic | 7–9 | 4–8 |
| Portrait | Photo / Art | 7–8 | 32–48 |
| Landscape or sky | Photo / Art | 7–8 | 32–48 |
| Painting or illustration | Photo / Art | 6–7 | 16–24 |
| Bold poster / screen-print look | Photo / Art | 5–6 | 4–8 |
Generate, look at the result at full size, change one setting, generate again. Conversion is fast and nothing is destructive, so iterating is much quicker than trying to reason out the perfect settings in advance. The instructions page documents every control in detail.
Take the SVG if the artwork will be resized, edited, recoloured, or printed large. It is the master file: every shape in it stays selectable and editable in Illustrator, Inkscape, Affinity Designer, or Figma, and it stays sharp at any scale.
Take the PNG if the artwork is finished and just needs to be seen — a social post, a slide, an email, a site that will not accept SVG uploads. Our PNG is written at twice the original image's resolution, so it has real headroom, and any Paint or Canvas texture is baked in exactly as you saw it. One thing that surprises people: vector files are not automatically smaller. A complex photographic trace can run to tens of thousands of polygons and produce an SVG several times larger than the PNG. There is a full comparison here, with a decision table by job.
No. The engine is JavaScript running inside your own browser tab. The image you load is read from your disk into that tab and never leaves your computer — there is no upload step and no server-side processing. You can verify it by disconnecting from the internet after the page loads; the tool keeps working. The privacy policy covers this and the advertising cookies in full.
There is no fixed limit, but there is a practical one: the work happens on your machine, so a very large image at a high Quality Level takes longer and uses more memory. If generation is slow, lower the Quality Level a notch or scale the source image down before loading it.
Almost always too few tones, a Polygon Fill Size set too high, or Flat Graphic mode used on a photograph. Each has a different fix — the troubleshooting guide works through all five causes.
Because an SVG's size is driven by shape count, not by dimensions. A high-quality photographic trace with many tones legitimately contains thousands of polygons. Lower the Quality Level, reduce the tone count, or raise the Polygon Fill Size — all three reduce shapes, and shapes are the file.
As far as this tool is concerned, yes — you keep all rights to what you put in and what comes out, and no attribution is required. But tracing does not create rights: if you did not have permission to use the source image, you do not have permission to use the vector of it either. That matters most with logos, which are trademarks regardless of what format they are in.
It runs in any current desktop browser — Chrome, Safari, Firefox, Edge. Phones and tablets work for smaller images, but conversion is processor-intensive, so a large photo at a high Quality Level will be noticeably slower on a phone than on a laptop.
No. There is no sign-up, no login, no free trial, and no watermark on the output. The site is supported by the ads on the page.
Instructions — every control, what it
does, and the tips that matter.
Guides — longer pieces on vectorizing logos,
fixing blotchy traces, and choosing an export format.
About — how the engine is built and why.