Convert SVG to JEF (Janome) Online

The SVG arrives as geometry and has to leave as a route the needle can follow. Every closed shape that has a fill gets covered with zigzag scanline stitches, with the angle rotated between colours so adjoining areas do not all lie in the same direction and flatten into each other. Beneath each of those fills goes an underlay, a lighter pass that stabilises the fabric so the top stitching does not pull the cloth into ripples.

Open paths and strokes get running stitches instead, spaced at roughly two millimetres. Travel between separate regions becomes a jump, and where the colour changes the file records a stop so you can rethread.

None of that is specific to Janome. It is the same engine that writes DST and PES; what JEF changes is the container and the thread table.

Everything that decides how well this stitches happens before the conversion, in the SVG. Three things are worth checking:

• You no longer have to size it yourself. The design is fitted to the hoop field before stitches are generated, so the spacing is computed on the final shape. What you must not do is rescale the JEF afterwards: that changes the density rather than the design.
• Close the shapes you want filled. An outline that looks closed on screen but has a gap in the path is read as an open path and stitches as a thin run rather than a filled area.
• Delete stray nodes and hairline shapes. A trace artefact two pixels wide is invisible on screen and becomes a real stitch on fabric.

If you are starting from a customer photo rather than a vector, PNG to SVG is the step before this one, and it is worth looking at the result before feeding it forward.

The file is built to fit the hoop, before a single stitch exists. This is the part automatic converters normally get wrong, and it decides whether what you download is usable at all. The geometry is scaled to the machine field first, and only then are stitches generated on the final shape — so row spacing and stitch length land at their true millimetre values and the density is right by construction. The alternative, which is what you get from a stitch file resized after it was written, multiplies the stitch length and the row spacing by the same factor. That is exactly the "never resize a stitch file by more than 20 percent" rule the trade warns about, and it is why this fit lives inside the encoders where no caller can skip it.

Unless you ask for something else, the target is the 4x4 in field (100 x 100 mm) — the Brother SE600, SE700 and PE535 family, by far the largest installed base — with 5 mm of clearance kept on every side, because the stated hoop size overstates the sewable field. A design fitted to 4x4 sews on every machine in the table; one fitted to 5x7 already excludes most of them.

And when the design cannot be stitched, no file is produced and you are not charged. If the short side would land under 5 mm, nothing inside it survives: satin degenerates into a running stitch, lettering closes up, and the fill silently drops the thin spans. Handing you a correctly-hooped file whose content has been erased is the worse failure, so the converter refuses instead — and says so properly: the size it reached, the hoop it was aiming at, and either the hoop that would hold it or the width to ask for. Between 5 and 10 mm you get the file together with a warning that fine detail may not read once sewn. The stitched size, in millimetres and inches, is shown on this page before you pay.

What this conversion does not do. It produces running stitches along strokes, zigzag scanline fills inside closed regions, and an underlay beneath those fills. It does not produce variable-width satin columns, and it applies no pull compensation and no per-region density tuning. On fine lettering or a narrow border, a professional digitiser will beat it. If the finish matters more than the wait, refine the file in dedicated software.

The distinction that matters commercially: this converts a clean vector into stitches. It does not digitise in the sense the trade uses the word, which involves deciding stitch types per region, sequencing to minimise trims, compensating for how fabric pulls, and tuning density to the material. Digitizing is not vectorizing goes into why those are different jobs with different price tags.

Used for what it is, it removes a wait. A flat logo that would otherwise sit in an outside digitiser’s queue overnight becomes a file you can test on a scrap in ten minutes.