Machine Embroidery Formats

From Wikipedia, the free encyclopedia

[edit] Machine Embroidery Formats

[edit] History

[edit] A Discussion by a pioneer in Computer Embroidery Design

"The complications with embroidery formats today are largely due to hysterical, ah, excuse me, historical reasons.

It may interest you to know that one of the first things we did at Wilcom when we started working in embroidery software was to recommend to the machinery manufacturers a universal tape format, or tape code as we called it. Their response was deafening in its silence.

In the beginning was the Jacquard...

The first embroidery design "formats" were the coding schemes by which the XY stitch movements were recorded onto punched tape used by the mechanical Jacquard mechanisms which controlled the early automatic embroidery machines. So in the beginning, there were really only 3 formats - the 46 channel "Plauen" card, also referred to as the Zahn or Vomag card system, for Schiffli machines, the 22 channel "Saurer" card for the Saurer brand schiffli machines, and the so called "68 millimeter jacquard" card used on the early multihead embroidery machines, beginning with the "Wurker" machine.

Surprisingly, all early multihead machine manufacturers used the same 68 mm jacquard card format. Including Wurker, Gross, Marco, Tajima, Barudan, Zangs, and perhaps others. In later years there were minor variations to the 68 mm card to incorporate new technical innovations such as... (wait for it) ... jump stitch, ! But basically, you could run a design card punched for one machine on pretty well all other machines. The biggest difference to watch out for was that most machines were set up mechanically to interpret the one unit of movement on a tape as 1/10 mm, but some machines were set to use it as 1/6 mm. With the +/- 40 units of movement possible on a 68 mm jacquard, this gave either a +/- 4.0 mm maximum stitch length, or a +/- 6.7 mm maximum stitch length respectively, but at the expense of slightly less accuracy on the end point placement of stitches.

An interesting thing technically with these 3 original formats is the number system used in each, which, by the way, is also a great contibutor to later divergence in tape formats. The Plauen card system used a system of "tens" and "ones" to define stitch distances. The Saurer card system used a "base 7" number system, where different hole positions on the card were valued at different multiples of 7. And the 68 mm Jacquard system used a ternary, or base 3 number system ! This is all very confusing to us humans who use base 10, and to computers which use binary, or base 2. But I'm sure there were good reasons for the strange original systems - after all they had mechanical punching (digitizing) machines to create the design tapes, and mechanical "computers" called automats to read the punched tape and control the embroidery machine.

Then came the BYTE ...

The problems really started when different machinery manufacturers started applying computers and electronics to the basic XY drive systems of embroidery machines. With electronics came that wonderful [not] new invention of 8 channel paper tape - also referred to as one inch tape. The computer industry had already virtually given up 8 channel paper tape in favour of magnetic media, but embroiderers liked being able to see the punched holes, so there you go. And after all - 8 channel paper tape was in fact 62 % smaller than 68 mm jacquard ! I'm sure the stock prices of Facit and GNT got a good boost from this.

So with electronics and computers, all the machine manufacturers were faced with designing a method of coding embroidery designs onto 8 channel paper tapes. And of course they all picked a different solution. My guess is that it was because the respective engineers at the different companies independently came up with different solutions to a common problem, which engineers are prone to do. But they were probably also trying to keep this new technology secret from their competitors. At least most of them standardized on 3 rows on the 8 channel tape per stitch, or 3 eight-bit "bytes" per stitch in computer terms.

Tajima chose an 8 channel tape code which used a direct hole-for-hole representation of the existing 68 mm jacquard coding system. They extended the code to allow stitches as long as +/- 121 units, or 12.1mm, and with the few additional holes left over, defined jump stitches and stop codes. This solution had the advantage that existing 68 mm jacquard cards could be very easily converted to or from 8 channel tape, with very simple electronics. Great compatibility, and low cost. Even today, Tajima's "tape format" is referred to as "ternary" , after the base 3 system used in 68 mm jacquard. The Eltac machine company, (the forerunner of Happy), also used the ternary tape format, with very minor differences in end-of-desing coding.

Barudan took a different approach, and abandoned the ternary coding system in favour of the binary coding system used by computers. This gave the advantage that XY stitch movements could be recorded on paper tape almost like they were stored in the computer. It also provided an increase in maximum stitch length to +/- 12.7 mm, and they had even more holes "left over" with which to define extra machine functions. So now you could not only program a "color change" command, you could even define which needle number of the multi-needle machine to change to. Great stuff. But incompatible with Tajima's.

Zangs, the forerunner to ZSK, also chose a binary coding system for their electronic machines, but alas, a different binary scheme than Barudan. But at least it was 3 bytes per stitch. And surely with computers, changing from one format to another is easy, right ? Ha. For computers maybe, but it's been a pain in the neck for the rest of us.

Ultramatic was perhaps the first (?) machinery maker to adopt electronics and 8 channel paper tape to embroidery machines. It also embraced the binary numbering system for its tape code, but, you guessed it, differently from all the others. Many in the industry may not know that Melco got its start working jointly with Ultramatic. They split from Ultramatic just before the 1980 Bobbin Show, which was where Wilcom announced itself to the world with our CED system - the world's first computer graphic digitizing system. But that is another story.

Melco took things even further with the binary coding system for stitch data. In addition to the 8 channel tape code they had worked with Ultramatic, Melco devised a new binary format that used only 2 bytes per stitch instead of the normal 3 bytes. At least for some stitches. Stitches larger than a certain value required 4 bytes per stitch. This format could be stored in the electronic machine in less RAM memory - which was very expensive at the time. Now you really did need a computer to handle this type of data coding. This format has been extended over time, and is now known as the Melco 'expanded' format, or .EXP.

When Wilcom started developing its Computer Graphics digitizing systems, we had to of course provide tape output which was compatible with the machines of the day. So we dutifully programmed all the formats, since the machinery manufacturers were not terribly interested in a common format. We understood them all, (eventually), but embroiderers struggle with them still.

Followed (eventually) by the DISK...

Melco was perhaps the first embroidery machinery maker to use computer disks in the early 1980's to read embroidery designs. Of course (!) this was before the IBM PC, and there were no standards for floppy disks at the time. Well actually there were - about a dozen or more "standards". Most popular in those early days of micro-computers were variations of the disk format used by the CP/M computer operating system. But CP/M isn't even heard of these days.

So in addition to the different "tape formats", which defined the stitch data of a design, embroiderers now had to contend with different "disk formats", plus different "file formats" for files stored on these disks. (And you call this progress, I hear you groan ?). And yes, all the machine manufacturers repeated the same strategy - they all invented their own "disk format". Some liked it so much they even invented several disk formats !

Old timers in computers would remember the rare 8 inch floppy disks (luckily never used with embroidery machines). Then 5 1/4 inch floppy disks were popular for many years in computers, but used only by Melco (?) in embroidery machines. Finally 3 1/2 inch (not so floppy) disks became popular, and these were well suited to embroidery machines. Luckily, pretty well everyone now uses a common 3 1/2 inch disks, which look the same, but unfortunately are still formatted differently, and have different types of files on them, for different brands of embroidery machines.

Tajima was the first machinery maker to adopt the IBM PC DOS format floppy disk for their embroidery machines. Some other makers have now also done this, but many formats are still "non-DOS" disks, which are still problematic with modern computer systems these days.

The "files" on the early embroidery disks were pretty much just an exact copy of the "bytes" of data normally stored on a paper tape. Compatibility with the past is very important after all. More recently, additional useful information was added to the disk files, usually in the form of "headers". Wilcom's early "tape files", for example .T01, .T03, .T05 etc, were byte images of tape data. The .T01 stored a Tajima tape image, the .T03 a Barudan format tape image, etc. These tape files included an early Wilcom innovation - the "man readable" label. I remember programming this, because I was having difficulty keeping track of which tapes were which. So I devised a system of hole patterns on the tape which formed the letters of the alphabet, and the digits. I automatically punched the design number, the description, the date, and the format right onto the tape, and left a long leader of blank tape before the start of the actual stitch data. Another innovation in those early days was to punch the tape backwards - starting at the end of the design. Then when the Facit tape punch dutifully wound the tape onto a tape spool as it was being punched, it finished with the beginning of the design on the outside of the tape spool - ready to run on the machine without re-spooling. This was possible because our digitizing system calculated the entire design on disk, and only when the design was completed was the tape punched.

Déjà vu with home embroidery machines ...

It appears that the domestic sewing machine manufacturers liked so much what they saw in industrial embroidery machines, that they decided to do it all again. Yes, they have all invented their own "stitch formats", which luckily will never be seen on punched paper tape. But they will be found in many different "file formats" on PC floppy disks, and on many different formats of "Design Cards", which use "flash memory" to store designs readable by the home embroidery machines. The tower of Babel not only lives - it grows.

Digitizing Systems and "Condensed" Files

All of the discussion above has been with embroidery machine formats. The common thing about machine formats is that they pretty much describe the same things, stitches and color changes and the like, which are pretty straightforward to represent in any of the typical stitch formats. The meaning of a stitch is clear, and it's a straightforward and very predictable matter of changing one stitch format to another. At least for machines with the same common capabilities. But digitizing systems bring in a whole new set of complications. They still have to deal with all the machine formats. But they also have their own file types to deal with. Luckily nowadays, most embroidery software systems use IBM PC compatible computers, and can exchange files on compatible DOS disks and networks. MacIntosh computers have different disk file systems, but there are some crude means to get files back and forth to PC's.

The bigger issue is that each digitizing system has its own specialized set of "CAD" files for each design. Any one digitizing system typically requires several files for each design. A "condensed" or "outline" file to hold the digitized shapes, plus a separate "expanded" or "stitch file" to hold the actual stitches (but in a different format than the machine stitch files) is the norm. Some systems also require additional files for preview icons, for text notes, and/or thread colors.

Melco and Ultramatic devised the first "condensed file" that I'm aware of. The big benefit of this type of file was that you could use the computer to easily scale the shapes to different sizes before calculating the stitches to fill in the shapes, so you could produce the same design in several sizes quickly and simply. Of course you had to still watch out for embroidery practicalities, but it was and is a great capability. Melco really popularized the Melco condensed format, .CND, because the Melco machine was the only widely used machine to be able to read (and re-size) a condensed file right in the machine.

Wilcom also devised a condensed file with our first generation digitizing system in the early 1980's. In our CED system we called it the .INP file, which stood for INPut. That is, we recorded the "inputs" that the digitizer gave to the computer, and stored them in a file. This could be "re-played" at different sizes, or at different densities , etc automatically. The CED system also had a separate "stitch file" , called the .DAT file. It stored stitches using very high accuracy numbers. It was only when we output to a machine stitch file, did we have to "round off" the stitches to the nearest 1/10 mm accuracy of the machine stitch formats.

Most digitizing systems today still have a condensed or outline file and a separate stitch file. If you want to edit the shapes of the design, you load and edit the condensed file. If you want to edit the stitches of the design, you load and edit the stitch file. Usually in a different program from the digitizing program. This can cause some problems though.

At Wilcom we took a different approach - the "single file concept" - to overcome these problems. We pioneered this concept with our DOS "ES" systems (ES-60 etc) from 1991 to 1996, using only the .ESD file for both digitizing and editing in one program. We improved the single file concept in our Windows based "Sirius" systems (ES-65 etc) introduced in 1996 with the current .EMB file, which stands for EMBroidery - (just like the magazine !). The .EMB file is unique in the industry because it contains both the "condensed or outline" data PLUS the exact stitch data, all in the ONE design file. Users can edit either the shapes/outlines or the stitches, or both - all in the same program and all in the same file. Of course things like preview icon, thread colors, text notes, and stitch parameters are also stored in the same single file.

The .EMB file brings several big advantages to embroiderers. Firstly, there is great simplicity knowing that they only have to keep track of one file - and they can come back and edit anything about the design, including re-scaling the design perfectly. Secondly, they can safely do the little "stitch edits" which fine tune a design to run smoothly in production, and not worry about losing the stitch edits when they re-open the condensed file. Nor need they worry about the stitch file getting out of step or out of date with the condensed file ! Because everything is saved automatically and can be read back from the one compact .EMB file.

A very important, but often overlooked aspect of condensed files is that they are a mirror of the digitizing system that created them. They can contain only those automatic stitch programs and stitch effects that the respective digitizing system has. So while one stitch format is generally as good as another - there can be a huge difference in the value of one condensed or outline format compared to another. It follows then that the best digitizing system will have the best condensed data format, by definition. Take for example the Melco CND file. It was great for many years, but it cannot save dozens of advanced digitizing features that advanced digitizing systems provide these days. CND has no pull compensation, no jagged effect, no motif fills, no variable density, no curve fills, etc etc etc.

Since condensed/outline type files are a mirror of the digitizing system that created / defined them, what hope does the industry have of any one software system being able to read and re-size other peoples condensed formats ? Well, all is not doom and gloom. There are 2 types of problems to overcome. The first is to be able to de-cipher the usually complicated data formats stored inside the file, so you know what the commands in the file mean in embroidery terms. This is solvable if you know the data formatting of the file. The second type of problem can be more difficult. Assuming that you have read a file and determined that a shape has for example, accordion spacing on contour stitch type, with jagged edge effects of certain values - does your digitizing software have the capability of calculating those type of stitches to fill that shape ? If not, you can't use the condensed file !

This means that it is conceivable for a MORE ADVANCED digitizing system to read the condensed/outline files of LESS ADVANCED systems, but NOT the other way around. But, even if you have a MORE ADVANCED digitizing system, that system must also understand the complex data formats inside the file, to know which features and shapes to calculate. This is an error prone activity. Condensed files are usually too complicated to reliably reverse engineer.

At Wilcom we have many advanced stitching methods in our Sirius digitizing systems that no other system has. (Our recent curve fill effects for example.) We also have cooperated with some other software suppliers to be able to read their condensed formats reliably. We read Melco .CND format designs from the detailed specifications we licenced from Melco. And we got from CadCam in England the means for our software to reliably read the .PCH condensed files used in its software which were popularized by the now defunct Gunold APS range of digitizing systems.

A third important aspect of condensed files versus expanded stitch files, is the following. With a stitch file, the stitches are essentially guaranteed to be the same original stitches no matter what software or machine is used to read the stitch data. But with a condensed file, the stitches are not stored in the file - so guess what - the stitches that result from a condensed file are completely dependent on the SOFTWARE used to read the file and to calculate the stitches. And therefore the QUALITY of the stitches from a condensed file is dependent on the software that calculates them. Of course the person who digitized the design controls the quality of the shapes, and the stitch angles etc. But it is the software which reads the file which actually calculates the stitches which fill the shapes according to the digitizers instructions.

So what's the moral of the story ? Choose the best digitizing software, and you have the most and best options open to you.

So where to Standardization ?

So is there hope for standardization in the embroidery industry ? Yes, there is, but not without significant hurdles. It would be technically easy to design a standard stitch file format, which every software and every machine could reliably and easily read and write. This is because most embroidery machines all have pretty much the same set of capabilities. And if the standard stitch format was designed as a superset of all existing (and forseeable) stitch format capabilities, machines that did not have some features could be programmed to ignore the things it didn't know about. To make such as standard happen would require the commitment of the machinery manufacturers to cooperate and support such a standard. And it would require a very good technical design of the format, so it would perform well for embroiderers so they would want to use it, and it would not be too difficult or complicated for equipment manufacturers to develop it.

What about standardization of condensed formats ? This is certainly more difficult. The very nature of the condensed format, being a mirror of its native digitizing system, make condensed files different from each other. The logic of incorporating a superset of features would the same as for stitch files. But there are many technical and commercial challenges to getting a standardized condensed format. In the meantime, embroiderers are well advised to choose the digitizing system and design file format that gives them the best stitching features, the best stitch quality, and the best prospects for long term support. Then they will have the most capabilities and options available."

- Bill Wilson (1999)