A fill is a grid
Everything on this page rests on one picture. A tatami fill is a set of parallel rows, spaced by the density figure in your software, and each row is broken into stitches no longer than the stitch length setting. At 0.4 mm spacing there are 63.5 rows in an inch. At a 4 mm stitch length there are 6.35 penetrations along each inch of row. Multiply and a square inch of that fill holds about 403 stitches.
That number is lower than most people expect, and it is worth understanding why before you argue with it. The 403 counts penetrations in a plain fill and nothing else. Real designs carry underlay under every fill, satin borders round the edges, short stitches wherever a shape turns, and objects that overlap each other. Those are what the underlay percentage and the catch-all percentage on the form are for, and they are the reason a stitched square inch of a real crest counts far above 403.
Lettering is a length, not an area
A satin column does not scale like a fill. Its stitch count depends on how far the column runs and how tightly the penetrations are spaced along it, and barely at all on how wide the column is. Fourteen inches of lettering at a 0.45 mm pitch is about 790 stitches whether the letters are a quarter inch tall or half an inch tall.
This matters for pricing. A name drop is cheap in stitches and expensive in hooping. A filled crest is the reverse. Two jobs quoted at the same price per thousand stitches can have completely different shapes of cost, and the run time page is where that shows up.
What a resize really costs
Take the default left chest — 3.5 by 2 inches, 45 percent filled, with 14 inches of satin and 6 inches of running stitch — and ask for it at 130 percent. The bounding box goes to 4.55 by 2.6. The fill and its underlay go up by a factor of 1.69, because area goes as the square of a length. The satin and the running stitch go up by 1.3, because they are lengths. The total lands between the two, and where between them is set by the mix.
The practical consequence is that you cannot quote a resize by scaling the old stitch count linearly, and you cannot scale it by the square either. Run the table. A design that is nearly all lettering barely moves; a design that is nearly all fill moves by the square, and going up 40 percent in size is very nearly doubling the machine time.
Calibrating it against your own work
The honest use of this page is not to trust the default percentages. Take a design your software has already counted, put its box size and its fill settings into the form, and change the last field until the total matches. That percentage is now yours — it encodes how your digitiser handles corners, how much pull compensation the fabric you usually run needs, and how much your objects overlap. From then on the page estimates your work rather than a generic average.
The percentage will not be the same for a cap front as for a jacket back, and it will not be the same for an outsourced file as for one you punched yourself. Keep two or three figures rather than one.
Questions people ask
How many stitches are in a square inch of embroidery?
It depends entirely on the density your fill is set to, which is why the page asks rather than tells. At 0.4 mm row spacing and a 4 mm stitch length the plain fill grid holds about 403 penetrations per square inch. Add underlay, satin edges, short stitches on curves and overlapping objects and the stitched square inch counts well above that. The reliable way to find your own figure is to divide a counted design by the area it actually covers.
Why does making a design bigger add more stitches than the size change?
Because fill is an area. Scale a design to 130 percent and every filled region grows by 1.3 in each direction, which is 1.69 times the area and 1.69 times the fill stitches. Satin columns and running stitch only grow by 1.3, because they are lengths. The whole design lands somewhere between 1.3 and 1.69, and the table on the page shows where for the mix you entered.
Does resizing in the machine change the stitch count?
Resizing a stitch file in the machine control panel usually moves the existing penetrations rather than adding any, so the count stays put and the density changes instead. Re-digitising at the new size keeps the density and changes the count. The table on this page describes the second one. If you are not sure which your setup did, stitch a sample and look at the back of it.
What should I put in the everything else field?
Whatever makes the total match a design you have already counted. It stands in for short stitches on curves, pull compensation, tie-ins and tie-offs, and objects that overlap. Ten percent is a placeholder to replace. Expect it to differ between a cap front and a jacket back, and between a file you punched and a file somebody sent you.
Can I get a stitch count from a picture?
Only as an estimate, and only if you can judge the filled share of the bounding box and know what density you would punch it at. That is the estimate this page makes. Nothing about a JPEG tells you how the underlay will be built or how many objects will overlap, so treat the answer as a quoting figure and re-check it once the file exists.