Revisiting machine knit “quilting”

This will are another work in progress post for a while as I gather materials, and prior to commenting on how to create these fabrics using the Ayab interface. In 2013 I shared my first post on the topic, covering single bed quilting as a hand technique, with the aid of a punchcard to pre select needles only, and with an intro to a simple ribber repeat in a single color. It was followed by a post explaining the color separation for 2 color quilted fabrics.

one color experiments using monofilament 

In my swatches below, the red stars mark the spots where one fabric surface separates from the other. All use 2 color patterning, with a single color only appearing on the reverse side. In Brother machines to achieve this result, the ribber settings are changed manually every 2 rows. The ribber is set to knit when carrying the backing color, which is also creating the “stitching/ joining” sections, while it is set to slip when the color creating the pockets is worked. The settings on the illustration on the left create the foreground color pockets, the ribber slips for 2 rows. The settings on the right will knit every row on the ribber. Two rows of each color are knitted alternately, the same as in DBJ. Since there are as many “blister” rows as there are rib rows, the blister/ pocket design will lie flat against the backing. If the goal is a lined Jacquard, the yarn tension must be evenly balanced to produce a fabric which lies flat. The texturing of the quilted surface is produced with an uneven tension, knitting more loosely on the main bed.  The looser surface may be left as is, or wadding may be inserted between the beds as knitting progresses, while dropping the ribber part way. Commercial versions of these fabrics are sometimes referred to with the terms eight lock or interlock, and double jersey. The punchcard instructions for the first swatch, from Brother Ribber Techniques Book 

reverse of fabric (black) folded over on bottom

Passap Technique 181

here the large areas are obviously “stuffed”

Pockets  may result as part of  any larger, DBJ design, when monofilament or thinner threads may be used for the second color

Possible problem areas: stitches may be dropped along areas where fabric is joined (A), sealing side edges (C) will keep them from separating (B). The end needle on each bed must knit the opposite selection as its neighbors in order to close the selvedges. The last needle at the side opposite the color changer may require pushing needles to E manually if not selected by the KM at regular intervals. 

Blister fabrics and pintucks are cousins. Quilted fabrics are sometimes referred to as “single blister”. Both rely on one group of stitches knitting more rows than another, can occur both in single color, or multiple colors. The ribber slips while the main bed knits X number of rows in pattern, and pleats/ pockets are created, eventually sealed by knit stitches on both beds. The number of rows for which this action may be taken in Brother machines is far more limited than in Passap, where the strippers help keep fabric in place. Single color pintucks/ blisters began to be addressed in my post, which reviewed Brother ribber techniques suggestions. A multiple row blister sample executed on Passap, in turn programmed from and ancient DUO 80 magazine, eliminating hand selection on back bed, and programming it in terms of black and white squares on the front bed.

its reverse side

My E 6000 manual is annotated extensively, and my scribbles make a good argument for keeping better notes. I immersed myself in learning the machine when first purchased. The technique for solid color backing, tubular FI illustrated in the manual, is a workaround to create “quilting” “automatically”. A bit of translation and a different set up from my working notes is offered after the scanned image. The circular setting in Ayab software performs the same automatic color separation. 

With this arrangement, pouches will be formed on the white squares of the card. If you think in terms of the colors of the squares, then it is not important which color is determined as the background color by the console. The set up may be different in your copy of the manual. Knitting four consecutive rows with either color reverses the position of the colors in the pattern. Each 4 passes of the locks complete one design row. Reversing the BX <— pusher selection (manually, on right, prior to color change), will reverse the areas quilted. The first pushers on both left and right are aligned in the opposite position to rest. Floats are created between the layers by the color not knitting on the back bed, and the altered pusher position will keep them from jumping off. Depending on the size of the pockets you may want 2 pushers in that position rather than one. Always swatch before you commit to a large piece of knit. The identical design methods as those offered for Japanese machines and the associated fabrics may be reproduced by entering the separations as a pattern and then in turn entering technique 129. The PDF suggests a method for using Japanese designs as drawn on punchcard knitting machine in an E6000. My scribbles from my Passap manual. The console and manual recommendations are only guidelines. Any and all technique lock settings may be altered to suit planned fabric, and if the machine offers beeps and reminders for the factory program, simply disregard them.

Double bed work requires at least 80 stitches by 100 rows for gauge testing for finished garments. Any machine: for blankets or warm jackets, the pockets may be filled with padding every X # of rows, the front bed may be dropped to make this easier. Mono filament and /or fine wire may produce structures which have objects “dropped” into pockets ie marbles, sequins, sponges that  will expand into the space when wet with fluid, etc.



Revisiting GIMP in knit design

If my only goal is to create illustrations for my posts rather than to produce programmable designs for download to any knitting machine, my go-to program in the past was Excel. Depending on my charting needs, now that Excel is no longer available to me, I am using Mac Numbers or Pages.

My computer is an iMac, with OS High Sierra installed. Last December my 910 was altered to work with Ayab, as mentioned in other posts, via a USB connection to the Mac. For generating files for download I like to work primarily in GIMP (Photoshop is no longer available to me as well). It is freeware. The latest stable Mac Version is version 2.8.22.  Windows users have an option to download  a later release . Another freeware option is Paintbrush , a smaller, user friendly program.

I began sharing my GIMP explorations in mid 2013. Some topics included charting 1, 2 , 3, color reductions for 2 color portraits , generated mazes 1  and 2 , lace mesh , lace mesh and superimposing , 2 color drop stitch lace

I am using one of my drop stitch designs throughout this post to illustrate steps taken for altering the original by using options such as scaling and tiling. It is intended as a partial quick start guide, not a complete tutorial. The program’s manual provides keyboard shortcuts etc. for anyone wanting to explore more options to obtain similar or other results. That said, the manual is not recently updated or Mac specific, and Mac vs Windows appearance of windows and menus differ.

To alter and set pixel per stitch options double click on pencil tool,  choose square single pixel highlighted in brush shape menu

to draw in single pixels, choose show grid, snap to

There are limitations as to how low the magnification may be set before the grid is visible or disappears. I often work in 1,000 magnification. The magnification value numbers, seen at the bottom of the charts in progress,  may be typed in or use arrows to scroll up and down in value 

If your document size is larger, it is sometimes quicker to work in larger pixel size blocks. Here one meets limitations

GIMP large square brushes: numbers indicate desired pixel countand if you are using Paintbrush instead, the “square” is lost

To scale images in height or width without color distortion, change color mode

Image, mode, select Indexed
choose Scale Imagehighlight number you wish to keep constant, whether width or height
to break aspect ratio and control one of the 2 values, click on chain like symbol on right, it will appear “broken”change the alternate value to the desired number

Click on scale image. If the intent is to continue editing by using the pencil tool, the image mode needs to be returned to RGB before proceeding with editing

You can verify finished image size by choosing scale image again, at that point the chain like symbol on the right of the numbers is “whole” again, awaiting further editing choices

double height double width
 scale may be performed again on the last image created 
the appearance if the image mode is set to RBG, and not indexed prior to scaling

scaling down by half for mylars or for machines that can alter motif to double wide: 

click to the right of the highlighted number, the “chain” symbol on right remains intact replace the width value, click to the right of the height value and the number will automatically change

click on scale; remember to change image mode prior to doing so if needed; on the left is the image scaled in RGB, on the right in 1 bit palette

If the image has numbers that have odd numbers in stitch width, the software makes the choice, in assigning values, since no half pixels can be rendered.  For design details see previous post 

Tiling is useful to check how repeats line up when applied to the all over surface of a knit, choosing to alter the horizontal repeat in width or vertical in height, adding borders, and other variations. Ayab software in particular, requires the individual repeat to be programmed for the number of needles you wish to knit in pattern on the main bed. Depending on the size of the original motif the tiling function may be used on nearly any scale, with or without a visible grid

To keep aspect ratio, chain like symbol on right of width and height values should be kept whole (first image). It will appear “broken” on side of values from screen grab after the image was tiled, when by pulling up the scale option again, to verify pixel counts for the newly created repeat (images 2-4)

to color invert a single repeat or a tiled one  

A Facebook Ayab Group share and tip by Claudia Scarpa  brought up a discussion on using Gimp to rescale knits with the intent of estimating distortion in the design as a result of stitch gauge. In many knits aspect ratio is forgiving. In representational knits it can start to matter significantly. I played with numbers from 3 different gauges, getting back to a simple circle square first, then following up with a much larger image. Starting with pixels per inch setting in scale menu:

doing a bit of math: if gauge is 5 sts, 7 rows, divide 7 by 5 = 1.4,  6 divided by 4 = 1.5; use the quotient (division result) to modify the respective X and Y values

keeping the X value constant, multiply the Y value by 1.4 for 5 X 7 gauge
in the view menu, uncheck dot to dot prior to selecting scale

keeping the Y value constant, multiply the X value by 1.5 for the 6X4 gauge, uncheck dot to dot if it is available in the view menuTo get rid of the math, simply switch to pixels /mm and plug in your gauge. The starting image:

change X and Y resolution to match your stitch and row gauge, breaking “chain link” on right to disable aspect ratio. In order to show the altered scaled image, the dot per dot feature in the view menu needs to be unchecked prior to selecting scale

knitting gauge 5 stitches, 7 rows per inch  4 stitches, 6 rows  per inch and going much larger, using Rocco again: the original version scaled for stitch gauge of 33 X 40

a large geometric .bmp edited with same process

Claudia Scarpa has created a document on “how to edit a picture with GIMP for “Oknitme software” that she has given permission for me to share here: 03-ENG-gimp. The tutorial has clear instructions for anyone wishing to explore photo editing  for use in knits with GIMP, and the Oknitme online tool helps punchcard knitters participate in the process.

If freehand drawing suits your needs, you may find it easier to work with view settings shown below, and adjust pencil pixel size to suit



Drop stitch lace using Ayab software

Some notes on how tubular software color separations such as the one automated in the ayab circular setting may be found in previous post

In an effort to respond to a request I have had via my blog, I am sharing information on this topic as I have time to explore it. My first attempt when up and running with the Ayab software, was to to reproduce an earlier sample I had created as a color separation originally intended for a hand knit shadow knit experiment.

a chart from that blog posttry_drop_stitchthe resulting hand knit, on purl side
IMG_0823the color separated sample knit pre ayab the sample knit using the Ayab circular setting

Patterns predawn for shadow knitting, appear to be one published source for interesting 2 color drop stitch variations. What about geometric shapes or developing your own designs? Designs created using this technique lengthen considerably when off the machine. Both color sets of stitches become elongated as they are dropped, and that should be a consideration in planning your design. If your goal is a circle, the actual shape programmed may have to be closer to an egg laid horizontally rather than that of a “true” circle. The fabric also widens considerably when blocked and off the machine, making cast on and bind off methods considerations another necessity. Design repeats may be drawn in Paintbrush or GIMP (both freeware), or Photoshop. I have been a longtime GIMP user, and prefer to use it in tiling repeats as opposed to the copy and paste features in Paintbrush to accommodate the Ayab requirement of programming the repeats horizontally in the width of your piece.

In my previous how to posts on designing your own 2 color drop stitch lace, part of the color separation required elongating the design X 2. Starting side in Ayab for needle selection always needs to be from left to right, and elongation of the motif is not required when using the Ayab circular option.

This was my first working repeat, A = repeat charted out. B = the working bitmap or png, etc. (which would be the only requirement for the mylar). C = the image tiled for the chosen number of stitches (again, Ayab requires the repeat be programmed for the width of your knit piece). D = the image elongated, not usable for this fabric, it results on too much elongation. If you would like a knit border on either side, that can be achieved by having extra stitch(es) in work on the ribber

The results up to the point in which I had a yarn caught in brushes and stitches dropping on the left of the needle bed:

Drop stitch lace has been referred to over the years in other terms as well, such as release stitch, drive lace, and summer fair isle in Passapese. Passap knitters will recognize the results from this first method are akin to those produced using Technique 185.

Getting started: stitches intended to be dropped may be created on either bed. If the ribber is used to create loops, then the technique is a manual one. Using the main bed in Japanese machines to program dropped stitches increases accuracy and ease.

If you are swatching and testing, a permanent cast on is not necessary. The broken toe cast on is one of the two quickest on the Japanese machines, usable on on either bed. It is fondly called that, because if ribber cast on comb and weights are in the wrong place so that the wrong loops are dropped, everything falls to the floor, and likely on your toes. There is an online video by Diana Sullivan that shows its use for a tighter cast on row in 1X1 rib, but the use here is for a different purpose.

In producing this fabric you are technically knitting an “every needle rib”. Cast on a fairly tight zig zag row. The ribber comb wire needs to be placed so that it holds down the stitches on the bed on which you need to keep them. The principle and results are akin to  the first row knit when you use a single bed cast on comb, and the second pass, with the first knit row anchors open loops before your continue to knit. Any loops not secured by the comb will result in dropped stitches. Any fabric, any time, when 2 stitches are empty side by side, stitches are not formed and the yarn is dropped off them creating a float or ladder.  The red line indicates the ribber wire on top of the ribber loops in the zig zag row, placed so released knitting will be left on the ribber. Black lines your zig zag yarn loops, blue dots the teeth of your ribber cast on comb. You can check placement by dropping just a few loops on the main bed before hanging your weights on the comb. 

zig zag row showing placement of cast on comb teeth, on each side of main bed needles
with wire in place, anchoring ribber stitches testing out dropping a few stitches all stitches now on ribber in preparation for dropping stitches created on main bed 

It is possible to also use a wired cast on comb for an open stitch cast on on the top bed only. Remove wire from comb. Bring the comb up and between needles to be used, and  re-insert wire. Needles and latches will need to travel easily under the wire when the first knit row takes place. 

The knit carriage will not clear the comb properly because of the location of its brushes, etc. For the “cast on” row, exchange the sinker plate on your knit carriage for the one normally used with the ribber. The first photo below shows the approximate location for the comb during the first row knit. Needles are centered between the teeth, the teeth themselves line up with gate pegs. The comb needs to be manually held in place, since there is no opposing bed in use to help balance it. Working with the ribber up would ease the process in wider pieces of knit. The ribber sinker plate has no brushes or wheels to anchor knitting on the knitting bed. Any rows knit single bed using it, will need to have needles brought out to hold position prior to knitting each row for all stitches to be formed properly

the comb in positiona pass is made slowly with the ribber sinker plate in place the comb is droppedbring all needles out to hold position knit one more row, returning to starting positionchange sinker plate on knit carriage if needed, proceed with knitting

To use the same method with ribber in place: hold the appropriate ribber comb with the bump(s) up facing you, so that the teeth line up with gate pegs as shown above, and so the needles can come through the gaps. Leave the wire in, hold the bump(s) against the ribber, and tilt the comb against knit bed. Hold the comb high enough to take the carriages across to opposite side. Move carriages to other side, drop the comb and weigh it. If continuing on the top bed only drop the ribber, switch sinker plates, and continue to knit.

“bumps”: Brother comb 

For other purposes and an edge similar to a “weaving cast on” executed on Japanese machines use EON for the “cast on row”, then bring into work and add the rest of the needles prior to knitting the second row.

Use a cast on comb appropriate for your knitting machine’s gauge ie 4.5mm, 5mm, etc., brand is not relevant, only tooth spacing is. It is possible to cut ribber cast on combs into different widths for use when knitting is planned on fewer stitches than those accommodated by their available commercial widths.

As for dropping those loops that will form the long stitches, one can do so “manually” with improvised tools. For more “automatic” dropping of stitches using knit carriage in Brother patterning, one may punch a card or draw a mylar with a method akin to color separation. A pass of the KH carriage across the knit is made with no yarn in feeder, “color 2” is actually “no yarn/empty” from left to right,  while establishing the proper needle selection on its return. The ribber would need to be reset to slip, or the ribber carriage separated from the knit one for the 2 passes to and from the color changer. This is the “scariest” option by far, more error prone. It is not applicable when using the circular Ayab setting in creating the fabric. Without a specific “tool”, all stitches can be brought to E and back to B with a ruler, piece of garter bar, ribber cast on comb, or other handy ‘toy’. Dropping stitches is done while carriages are on the left, after the return to the color changer side. It is possible to modify the Studio accessory used to drop stitches

full altered_500

For 2 color drop stitch, the main bed is set to slip in both directions. Because not all needles on the main bed are used for patterning on every row, the KC II setting on the change knob is used, eliminating end needle selection on the 910. The ribber is set to knit every row

Ayab: begin  your design repeat on your first row, choose its circular setting in machine type pull down menu on right
First design row is preselected left to right
Main bed is set to slip <— —> , change knob on KC II (end needle selection is cancelled)
Ribber is set to knit <— —> for the duration
COL: as you go from the left to right, needles are preselected on the top bed, they will knit, picking up loops that you will in turn drop on the subsequent passes of the knit carriage from right to left.

COR, the KC knits on preselected needles as it moves to the left. Clear the color changer, set up your next color. Drop the stitches knit on that last pass

It may be necessary to push those loops down between the beds before you next pass, remember to pull down on your knitting periodically,  visually check needle alignment on the main bed (all needles in B in work area)

*With new color move to right, preselecting the next row of loops
Knit right to left, picking up loops on preselected needles, change colors, drop stitches,** and repeat * to ** steps in 2 row rotations

so I want circles, here is my test pattern more like eggs, the black squares in shape appear as drawn on the purl sideshapes are reversed as drawn on knit side 

It is possible using the circular setting to drop only one of the 2 colors, whether background or shape. I began by dropping the white ground. I used to encourage students to develop a tune/ repeat in their head when regular actions needed to be taken. For me, in terms of yarn color,  it was “white, knit, drop”,  “brown =  erase (push back to B), go back”. 1: white travels to right, needles are pre selected; 2: white travels to left, picking up loops 3: on left, change color, drop stitches. For brown: 1. travel to right, needles are preselected 2: on the right, before traveling back to left push all selected needles back to B. Only the ribber knits on the way back to left, so brown will have knit 2 rows with no dropped stitches. I ran out of brown yarn, started over with the blue and white, planning on having the shape drop the stitches. There is a difference in the fabric width with the change in distribution of stitches. I stopped knitting not due to any mis patterning, but because I encountered another Ayab behavior that may be well known to punchcard knitters. Due to a yarn mast issue I moved the knit carriage back to disentangle the yarn, and lo and behold the pattern advanced a row.  Punchcard machines will advance a row with any movement of the carriage outside the edges of the knit. This was never an issue in the unaltered 910. At that point I stopped knitting. 

Not fond of stripes? prefer one color? the sample below was worked on 40 stitches in width, using the repeat charted for 56 stitches. Here decisions are made at the design phase of your repeat. For single color drop stitch use an image double length, and single setting in the Ayab software. The process is the same: *preselect stitches left to right, knit on selected needles right to left, drop loops just picked up traveling to left**, repeat from * to **. Settings are the same as for the 2 color drop stitch, but the elongation depending on the number of stitches dropped is not as noticeable. The texture in my swatch is diminished after a quick press, the yarn is an acrylic blend. The charted repeat illustrated is wider, but I worked it on only the center 40 stitches. As always in slip stitch, the black squares knit and they represent the stitches that are dropped. If you wish to create the long stitches on the ground (white squares),  reverse use Ayab Action Invert prior to knitting

the chart as viewed and explained at the top of the post

Sources of inspiration from studio publications vary, patterns designed for pile knitting make for suitable one color drop stitch. A partial punchcard repeat

from an electronic collection

and a punchcard pattern book, where markings emulate eyelets, usable only for single color knitting 

Note that in #2 card directly above, there is a solid row at the very start that is a design row (third all punched). In Ayab again, first row preselection is left to right, you will be picking up loops on preselected needles going from right to left and then dropping them. That first design row needs to have punched holes or black squares/pixels in it. The color separation is essentially done for you in the source image. Do not use circular in Ayab, but rather, use “single” setting and follow instructions for creating the fabric as described  above,  with no color changes. The blank punchcard rows match the no selection rows in self drawn color separations,

If double bed work is daunting, for a different stitch, worked single bed, that may cause interesting distortions in all knit, single bed fabrics see block stitch post.

The self drawn design repeats for 2 color drop stitch may be offset as well, resulting in colors being dropped alternately.  The design shape needs to be created in 4 row blocks in order for the yarn to make it back and forth to the color changer with both colors to complete one design row. The second pairs of rows in each 4 row block is  “erased”. In this instance as well, rows marked with black squares will pick up loops on the main bed, which are in turn dropped to create the long stitches. The second design repeat is off set to try to get sections of it to create loops to be dropped as well. The final motif must be a multiple of 8 rows in height if it is to be used as Ayab’s “infinite” repeat in length.

Color changes are indicated in the vertical strip in the center of the design. This was my starting idea

Elongation to 4 rows per block, erasing second pair of rows for each color on right. The repeat on the right may be used if only one color is to be dropped; the 2 blank rows in the design field represent stitches that will knit only on the ribber for 2 rows. No stitches are dropped in color(s) used in sequential rows of blank squares in all “white” areas of chart, including multiple rows above and below the design shape. Color changes every 2 rows continue, creating continuous stripes. shapes staggered, visually checking for placement of alternately dropped stitches 

To accommodate the Ayab preselection for the first row to be knit from left to right, move the last blank row in the design to the first row position. As the carriages travel from left to right and back to the color changer, the stitches will knit 2 rows only on the ribber. Continue knitting in steps as described earlier in post, changing colors every 2 rows. On a larger knit ground such shapes may be arranged to suit. This was my working repeat, but I used a third fewer stitches in the swatch than in chart. Note that the images will be reversed on the knit side, so if preferred, use Action Mirror to flip the image horizontally prior to knitting it

The swatch has been quickly pressed, so texture is flattened out, but I am reminded a bit of shadow knits when viewing its purl side

In an unaltered 910 with the ability to double the width of the programmed repeat, mylar users are not excluded from exploring a similar fabric. The repeat above may be rescaled to half the width,  drawn that way, and then use the twice as wide built in feature. Gimp does an “interesting” thing when scaling this design to half width, note the right side of each repeat is an odd number of squares, the left side an even. The repeat may be used as is or redrawn, adding or eliminating black squares if symmetry in each shape matters.  Paintbrush produces the same image, mirrored.

The explanation: further analysis of the original design reveals the fact that some of the pixel numbers in the design black square blocks are uneven in width. In this instance 3.5 is half of 7, and half pixels cannot be rendered, so the software assigns the split to 4 and 3.  


Woven Lace: a brother punchcard to electronic

This is actually a reworking of a previous post. I usually sit on any post for a while, returning to it, editing multiple times out of view, and publish when satisfied with occasional return visits. Thanks to comments from another Ravelry member, I realized after my first “quick” publication on 589 that I needed to take another look at my thinking on this particular card from the very start.  I chose to stay public during editing to show that no matter the level of skill or time spent knitting, sorting out issues for any particular design or fabric can take time, sometimes obstinacy, and that at times the simplest route is taking good notes during the actual process for the most direct results in terms of clarity.

I have been knitting a long time. Sometimes things seem so obvious to me as I work, that I do not take notes. I still have a swatch from my teaching days that fell into that category, and that I have never been able to duplicate. With the 910 and the limited availability of space on mylars, I sometimes erase too quickly, and now that my charts are being created in a new and not yet habitual and familiar to me programs (Numbers, in high Sierra and Pages) wonderful, “surprise” variables can happen. So tips to self: keep step by step notes to return to as a reference, choose order of steps top down or bottom up and stick to one or the other, try not to scrawl randomly and everywhere on any one sheet of paper (never mind keeping it legible), don’t recycle papers with such notes when you think you are done, and watch those autosaves and revert options in Numbers and Pages. Then there’re the added factors of occasional WordPress crashes during saves, taking what one assumes as familiar for granted, and of simply putting even single, lone black square in the wrong place on a mylar sheet. I found with this series merely editing information in the software was not enough. Some of the errors became easier to see when matching the software theory to actually taking notes the old fashioned way with each carriage pass on the 910.

Some information on punchcards, their use, and pattern repeats may be of help to any of you who have not used a card before. The previous post included a reference to this punchcard from a Brother punchcard book in its combining weaving and lace section. I began with the assumption that knit carriage would operate from right, and lace carriage from left, their traditional placement in most lace knitting

its supposed related swatch

Analyzing the card, sorting out possible repeats follows, since mylars or bitmaps for download often only require a single repeat of the pattern. If you are not used to doing this, sometimes beginning with recognizable vertical ones first is a bit easier.

the apparent width and card height repeat

Going from lots of dots to far fewer ones can be dizzying. The punchcard multiple vertical repeat on the left is double checked to insure that all marks are in the correct placement on my chart. Black dots for EON needle selection, red ones for lace transfers. Here things get checked off twice, particularly for lace holes. Lace cards are the hardest to copy and place pixels or black squares accurately, simply because their markings are so few. I usually begin on the left hand side of punchcards to  isolate my repeats. The number markings in the center image reflect those found on the far right of Brother punchcards. The #1 on the factory cards represents where the card reader teeth are reading the holes on the interior of the machine, not at the operator’s eye level on the machine’s exterior. That is also the reason why in any fabric, needle selection does not match expected design row. On the far right,  a single repeat in height is isolated further. The same sort of check should be done when punching cards from published images. Blowing up the source and printing can help with accuracy. 

Operating carriages for even number row sequences is is the most convenient. The assumption on the basis of the arrows in card 589 is that each carriage makes 2 passes, operating in a continuous loop. That simply did not work for me in terms of producing 2 rows of lace followed by 2 more in any other pattern when operating LC from left, and KC from right.

If punching a card, verify your final punching by holding black paper behind the punched holes. In this instance there is an obvious mistake on the third row of holes. I am at the end of a very long punchcard roll. Some of the new rolls are wonderfully sturdy material, but the roll can retain a curl that may make it easy for the card to roll back into the reader unless joined into a tube (resulting in patterning errors), and making it hard to feed its starting rows  into “punching machines”. The rolls are marketed for Brother use, and numbered separated into “standard” lengths with blank segments between them.  Those markings may however,  be for Studio machines use (seen here, at row 5),  with row one on the right actually being 2 rows too low for Brother, so as punched LC  first preselection row would need to happen on row 3.589 begins with lace transfers. A second option, is to begin with a weaving pattern, with the KC on the left for the first selection row, and beginning your pattern reading 2 more rows up from the lace starting line in the punchcard. Markings on the side of the image above have nothing to do with actual starting rows. Pencil lines are outlining individual repeats, have no other reference. Numbers and other necessary marks would require adding by hand. If hesitant, #1 marking on right may be double checked by overlaying a factory pre punched card over your own.

Beginning the mylar conversion : my adjusted repeat checking squares vs holes again

The end needle selection needs to be cancelled on lace rows to avoid transfers or dropped stitches on the edges of the piece. If a needle is selected, manually push it back to B. If patterning ie tuck, slip, or FI are used on alternate groups of rows, then end needle selection is preferred. In weaving,  2 side by side stitches drop a float, so keeping the EON selection on both sides gives a better edge. Invariably, some operator involvement is needed altering end needle position, no matter the setting for it on either carriage.

Even with a lot of knitting and design experience results are not immediate. I ran into issues when I first tried to knit on the 910 with a later, “final” repeat selection being correct, but the technique failing, resulting in a loopy mess. In weaving, weaving brushes need to be moving freely, so check them, unscrew them, remove any fluff, and air knit, making certain they are down and turning at the same rate in both directions. Tuck wheels if movable, should be in the forward position, lined up with all other wheels or brushes on your sinker plate. To isolate the problems further: test lace with your knit carriage selecting,  but no cam buttons pushed in. The KC will knit for 2 rows across LC transfers, creating only the lace pattern involved. Your mylar markings get checked, also offering an opportunity to sort out why stitches may drop, and if you indeed have patience to combine techniques for more than a few rows. It is easy enough with an EON needle pusher to test both weaving yarn and needed ground yarn tension. If that is successful as well, then issues occurring with patterning may be from other causes. They were in my case. I had to switch sinker plates in my knit carriage to get weaving to work properly. I failed, however, at combining the final repeat with tucking, even with the KC set to tuck immediately after the transfers, and to knit every needle in the opposite direction.  I tested the carriage, sinker plate, and patterning with EON tuck only with different designs, and had no problem. At such a point I would abandon that fabric combination with the particular mylar repeat.

Returning to actions on the punchcard: the preselection row can be confusing in any translations. The last row in the card takes care of lining up repeats for us as it is rolled continuously in the drum, but there can be oddness to the eye when only a small repeat is singled out. Looking at the image of the repeats tiled allows one to choose a different starting row, rather than lace markings. The direction of transfers raises the quest to yet another level. This was my progression in editing and moving starting row for beginning the fabric with a weave start rather than a lace one: the numbers in my charts reflect carriage passes and direction, not completed row counts as they might appear on KM counter

Pass 1: COL, KC, N for knit row, slip <—> for free pass that selects only, move carriage—>
Pass 2COR, KC, EON pre selected, set card to advance normally, lay in weaving yarn, KC will move  <— will weave first row, preselect  second weaving row
Pass 3: COL, KC weaves second row EON, preselects firs row of lace on its way to right —>
Pass 4: COL , LC transfers to right, repeats previous row’s selection, moves to —>
Pass 5COR, LC no transfer happens on empty needle selected for the second time by the previous row, selects for the next row of transfers, moves <—
Pass 6: COL, LC transfers pre selected row to right, preselects first EON row for weaving, moves —>, release it, return it to left
Pass 7COR, KC, EON, lay in weaving yarn, KC will move  <— will weave first row, preselect  second weaving row
Pass 8: COL, KC weaves second row EON, preselects firs row of lace on its way to right —>
Pass 9: COL , LC transfers to right, repeats previous row’s selection moves to —>
Pass 10COR, LC no transfer happens on empty needle selected for the second time by the previous row, selects for the next row of transfers, moves <—
Pass 11: COL, LC transfers pre selected row to right, preselects first EON row for weaving, moves —>, release it, return it to left.  This row matches design row 1, and is starting the repeat sequence again
Passes 2-11 complete 10 rows of knitting, as well as the first vertical repeat as drawn

My chart has had multiple drafts, which included using software to insert a row to allow for that repeated selection as the carriages switched sides. The goal is to keep the 2 lace transfers to the right, with LC beginning each sequence on the left side, ending on the right, matching the punchcard movements listed above.  Keeping things as simple as possible is something I at times forget to do. After several drafts, here is a simpler way to look at things. In D black squares with dots indicate stitch that will be transferred on the next pass of the LC, the yellow squares the location of the eyelets. All pattern repeats with the KC preselecting for the first row of knitting on the left, with the change knob set to either KC I or KC II

the 10 row repeat and its mylar companion

In actual knitting of this stitch combination,  since needles preselect for the next row knit, once the pattern is set up correctly, it is easy to recognize when carriage changes are required by looking at the number of needles selected. EON rows are for weaving, starting on right. Few needles selected are for lace, starting on left. Lace transfers happen on the stitches selected the previous row, in the direction in which the carriage is moving, here transfers are all to the right. Grey squares indicate repeated selections, black squares with yellow dots indicate stitches transferred to the right on the next pass of the LC. The mylar repeat would only require the black squares

Below the repeat in the chart is used, knit first on my punchcard KM, then on my 910, but it misses the mark in terms of matching the swatch in the pattern book

This is the result I was still trying to get back to. The first lace pattern test swatch: gotta love dropped stitches in lace!

no tucking allowed in combo with lace, but not a carriage or mylar problemweaving test: fails were from a problem sinker plate weaving lever in one direction only, hand technique success was with change of sinker plate lace and weave with 2 different weight yarns and, hallelujah!

Taking another look at the original punchcard and those arrows on the left hand side, a detail I had originally missed. Both carriages operate from the right side, for 2 rows each, switching their place on the machine bed. Arrows for design row 1 begin above red line. Transfers are all first to left, then to right. The card advances a single row for each carriage pass. With carriages operating on the same side, the punchcard advances one row for each carriage pass. Operating the LC from the left, and releasing it when it is on the right as in the previous fabric produces the desired results. The blank row on rows 3 and 8 allow the LC to travel to right, making selection for the transfer to left on the next row

the  10 row repeat and its mylar companion

the actions

the fabric, again!

Reworking the repeat in order to use the LC for four passes, beginning and ending on the left hand side of the machine, its usual home

the now 12 row repeat and its mylar companion

the actions the fabric again, in lace only repeat 

For reproducing the fabric using Ayab software: please see


Tubular machine knit fabrics: work in progress

This will be another of my WIP (work in progress) posts. I will add information and edit as I have time.

I recently came across this topic in yet another forum, so thought I would share some of my thoughts on it. The technique involves different patterns on 2 opposing beds. Table for 2 offered one option for programming 2 different  knitting machine beds to achieve reversible DBJ or true tubular fair isle. I will be addressing Brother settings for the moment. Things to consider: the piece will be double thickness and 1.5 times the finished weight of one knit in single bed FI. FI technically knits 2 colors for each design row, the ribber at least one for the first piece here, so thinner yarn is probably best. If the main bed is creating a 2 color slip stitch, whether one color at a time (DBJ) or 2 colors together as in traditional FI, the fabric will be shorter and thinner because it gets pulled in by the shorter floats of yarn in areas that do not form knit stitches with that color. Generally, in FI it is good to have end needle selection on, so that the second color gets caught at the edge, and the design does not separate from the rest of the knit at the sides. I knit my sample using KCI on the knit carriage. The ribber is going to knit one row for every 2 passes to complete each design row by knit carriage, so some adjustments in tension both beds need to be made for each side of the tube to be balanced. If the goal is to have the tube open at the bottom, I like to start in waste yarn, and begin tubular while leaving a long end of the first color used so “bind off” can happen by sewing or crocheting the open stitches when the piece is done. The same can be done at the top of the piece, and both ends will match. Setting the needles point to point during tubular knitting will diminish any gap at sides of the knit if tube is made in the traditional set up. Since only one bed knits at a time, needled will not come in contact even if directly opposite.

This fabric shares features with quilted fabrics, except there are no areas where the fabric gets joined to make pockets (needles on both beds knitting on any one pass of the carriages), the goal here is a continuously open tube. Minimal manipulation of carriage settings can be achieved in color separating for the specific fabric. My previous posts on quilting: 1 and 2 , and on color separations for DBJ 

It is always best to start with a simple design that allows one to recognize needle selection. Long verticals can separate at their edges if knit single bed, but I went for a simple pattern of squares to test out my premise and winged it. Thinking color 1 knit left to right, slip right to left and ribber slips when main bed knits right to left, knits when main bed slips left to right. The all blank rows indicate the main bed slipping the width of the involved needle bed. Same applies after any other consecutive color change. My mylar repeat, free drawn End the circular cast on/ waste knitting start  and begun with opposite part buttons pushed in with COR; set change knob to select pattern (KC I), to prepare for move to left toward color changer. Main bed only will knit. With main bed set to knit <–, needles for the first row of pattern are selected, but all stitches knit. The ribber needs to slip <– completing the circle. If in doubt, push up both part levers on ribber.

the start of the tube, needle selection for first pattern row 

after the color change, with COL, change settings once more. The main bed will now knit to right and slips to left based on needle selection, while the ribber now slips to R, knits to L  (the goal here is simple stripe)

the start of a tube is evident if the ribber is dropped a bit the tube dropped off the machinethe FI created on the main bedthe pattern on the reverse, created by ribber

It helps to analyze the carriage actions on knitting the repeats. These charts illustrate the 2 colors knitting alternately with each pair of carriage passes. With any knitting motif, it is worth testing repeats consecutively in height and width to get a sense of what will take place using them on the finished knit. Here 2 lengthwise repeats are shown. The pattern could be planned to start the repeat with the area outlined in red on the right, and a full block would occur at the start of the tube. A bit of editing would make all rectangles equal in height.

So now that I have some blocks, what about other patterns and shapes? ……

In a previous post with exercises on DBJ, 2 separations of the triangular shapes below were offered. Testing the idea of using pre designed jacquard separations for knitting tubular fair isle, I chose to use the 2 alternatives presented. The related information is available in PDF format for punchcard and electronic machines

Here we have the motifs, and 2 methods of color separation for DBJ, charted, drawing each color design row only once

The middle image shows the method for separation achieved automatically in Brother by using the KRC button when using only 2 colors per row. It causes the least (if any) elongation of the motif. Each design row knits only once, programmed as is. If the KRC button is used, none of the other functions including double length are allowed. The separation on the right is the most common color separation, actually knits each design row twice, which elongates the motif. As drawn, it would need to be programmed double length. If the knitting is representational, adjustment may be needed in altering its height prior to downloading it for knitting. I used both separations above  drawn on an old mylar for my second tubular FI attempt, sticking with my thin yarn, and working in small swatches while testing technique ideas. My initial premise was to knit each row twice using double length. The preparation was as already discussed. After the color change and resetting the carriages COL, the first row slips on the ribber, and the main bed knits first design row in the alternate color, seen here in white. As in single bed FI, floats are created in that color between needles knitting it  with COR, before  traveling back to left for the color change, return all main bed needles back to B position with a ruler, ribber comb, or any preferred tool, so that only the ribber only will knit, keeping the knit tubular as you travel back to the color changer.

COL: change colors and repeat. Here are my trial tubes, again in too thin a yarn, but proofing the concept. My mylar was ancient, marked in #2 pencil on its reverse, and I can see a couple of spots that need darkening, but that is not a concern for this exercise

Adding color and expanding the charts to match the B/W versionseach row elongated X 2what knits when setting machine for tubular; in the swatches the blank rows were achieved by manually pushing needles back to B as shown abovethis shows what happens when the first separation is collapsed into what actually gets knit by the carriages in this technique
and the result in repeat in B/W like my swatch

Both elongated repeats are reduced to black and white squares below, with blank rows included to let the carriage do the work, without having to push any needles back to B position, KC set to slip <–  –>. I often work out my separations in color, then fill the colors in with black and white. It is easier for me to draw the black only squares on the mylar or to enter them as pixels, than trying to do that from a colored graph. I am now charting in Mac Numbers, no longer have access to Microsoft office on my new Mac.

DBJ patterns predawn for punchcard machines or provided as charts for electronics may all be used for the FI pattern. The first separation method will yield a surprise. The second method if each row is illustrated only once, needs elongation. In separations where the work is done for you, producing a result like the elongated swatch on the right, the pattern is ready to go.

Creating the separations by hand is time consuming, having software separate the motif, eliminating that design time is quite helpful and more error proof.

The ribber needle selection can be altered to produce patterning with hand needle selection, or with changing the slip lever positions as frequently as every row. That is a post for another day. That said, some things to consider: my own DBJ scarves often involve more than 1200 rows of knitting. The number of carriage or lock passes involved to achieve the same length in increased exponentially should I choose to knit the same designs tubular. If any needles on both brother beds wind up in work at the same time prior to the next row knit, the fabric will seal closed in those areas (another range lovely fabrics, but not the plan here). For the ribber to knit in any other pattern than an easy stripe, one needs to either select needles by hand or add changes in levers possibly as often as every row. If multiple rows are knit on the ribber and then in turn on the main bed, one is actually creating a tapestry technique and there will be small slits at the sides of the knit, depending on the number of rows knit on either bed. It is truly helpful if the software in use or the machine’s console are able to recall last row knit and to take you back to that spot if knitting is interrupted or put off for another day. That is one of the very convenient features of the Passap system providing the battery is still holding the charge. The latter also performs the same function and returns to correct pattern row if one has to unravel rows and the number of unraveled rows is entered into the console. If long, non repetitive in length designs are in use and need to be downloaded in segments (wincrea has such limits), a warning noise when the end of the pattern repeat is reached would be nice to have. In my opinion it’s fun to achieve things because we can, then it becomes a personal choice as to whether the process is worth it, and under what circumstances to use it.  My own accuracy when a lot of hand manipulation or other details are in use, has faded with my increasing age and decreasing attention span. Others can tackle the same with great success, such as seen in some of the wonderful all hand transferred complex machine knit laces found on Ravelry, and then there was the man who used to tour machine knitting seminars with a MK sweater he had made for himself in fine yarn that incorporated over 3,800 (yes, thousands) cables, weaving in and out of each other and in turn in and out of crossing diamonds in a fine yarn (no, no one I know of actually counted them, the count was accepted on faith).

The Passap console has built in color separations for tubular knitting. As with Brother, the simplest fabrics are in one color. For one color fabrics CX/CX (the equivalent result in Brother happens by using opposite part buttons on either bed) is used on both locks.

For 2 color fabric the DM 80 front lock (carriage term equivalent) is set on HX <— while if using the E 6000 one may select KT (knitting technique) 243, with the front bed set to LX (slip). I have had very limited experience with the DM80 decades ago, can only speak for the E6000. The front bed is programmed as for any other pattern. With back lock set to CX (no pattern) or HX (pusher selection for pattern) the function is fixed, the front bed knits to left and slips to right, the back bed slips to left and knits to right to create the tube

Horizontal stripe is the simplest to execute: set the back bed on CX, or put all back bed pushers in work position and set back lock to HX, no arrow key. Pushers up knit, pushers down slip. (in Brother set the appropriate ribber lever to slip in one direction)

Vertical Stripe: “automatic”, bring pushers under every back needle into one up, one down, alternating in work and in rest. Set back lock to HX, left arrow key <—, changing color every 2 rows. The same needle selection is in the same location with each color change, creating the stripe

Bird’s eye backing, manual tech involved/ lock setting change involved:  set back bed pushers alternating one up, one down, thus alternating in work, out of work. Knit 2 rows with the lock on HX, no arrow key, and 2 rows with the locks on HX <—.

Solid color backing is possible: set the front lock is set to LX <— (slip <—). When Tech 243 is in use, after the set up rows are completed, the console gives the prompt for setting up the pushers, all in work, and for HX <—on the back bed. The result is that for 2 rows in the lining color, stitches knit alternately on each bed. The front bed then knits alone in turn for 2 passes;  the color yarn in use creates a float the width of the knit as the locks return to the right, making it the least acceptable variation, and an unbalanced knit. The set up is different for both quilting (BX <—/LX)  or solid color backed DBJ from tubular FI created in this manner.

My Passap manuals have been well used, often technique numbers are surrounded by notes in my scrawl. Techniques perform the color separations for the specific fabric, the accompanying diagrams and directions are suggestions for swatches provided as well in the alternate accompanying manual. Lock settings can be set to suit i.e. tuck substituted for slip, etc. The separation is fixed, but all lock settings are in the hands of the knitter.

2 color circular tucked designs are produced using techniques 162-165, with the frequently unfamiliar settings using OX/DX. As with many other techniques, some may be used with stitch patterns, some not, but if you understand what is happening front bed patterning could be converted to automated pusher or needle selection by entering a planned color separation as a pattern. OX is a combination, tubular tuck setting, paired with DX. OX is a combination of KX (tuck in pattern) and CX (circular). DX is the tubular tuck setting for the back bed. OX and DX will knit needles with a pusher (needle selected on Brother) in work (selected needle), and tuck on needles with no pusher in work (non selected needle) moving from right to left. From left to right the front bed is not knit any stitches (contributing to making the fabric tubular). The back bed knits knits only from left to right, the second side of the tube. E6000 technique 185 with locks set to N/OX will produce tubular FI with long stitch.  For three and four colors where the colors are separated at one row per color  use technique 252, locks set to N/OX. Superimposing patterns for three and four color techniques require entering card reader techniques. With my cable set up, when I was experimenting with reader techniques eons ago, I believe I was successful downloading the pattern for the design from the PC, and entering the card reader technique as a second pattern, via the card reader. I have never tried doing so with additional software, and shy away from any situation where I have to enter commands to get software to work.  Out of the group, 162 is the only one that can be combined with stitch pattern. The console gives cues as when to change lock settings, with some thought similar fabrics could be knit on the Brother KM. The techniques, and my notes, which at this point would need some self interpretation. The UX scribble is in reference to a fabric that slips over needles with pushers selected down from right to left, then tucks over needles with pushers selected down from left to right. Needles with pushers up will always knit. In Brotherese the pattern is knit with opposite function cam buttons  in use slip <–, tuck –>. It will not produce a tubular knit. Worked out patterns for such fabrics were included in some of the Brother punchcard pattern books. 
For tucked 2 color fair isle use technique 185 with lock on N/OX. 

What that means to anyone knitting on Brother machines?

Double Jacquard and color separations: some previous posts, links, hints

A recent forum discussion on DBJ on Ravelry led to my looking back on some of my previous posts. Some of the features in both excel and numbers changed over the years, but most basics remain. Program specific or a “software” general search in previous posts touch on GIMP, other programs and other fabric design choices.

This is a copy of one of one of my handouts when teaching DBJ eons agoanother former handout, intended for Passap knitters

A lace mesh series: GIMP, superimposing, Brother 910

I like placing motifs, grounds, and borders myself whenever possible, in any knitting technique, rather than relying on adding or combining them via the built in KM software.  It is simply my strong personal preference in designing and gives me additional controls over patterning. However, the ability to superimpose is a convenient feature, available on multiple machines and worth mentioning. I have used it more frequently in operating my Passap, than my Brother 910. That said, it would appear to be an easy feature to use for programming shapes onto mesh grounds. For illustration purposes I am using a 24 stitch repeat. Electronics width and height potential depends on use of mylar or PC download. The “rose” is not a workable, resolved repeat. When electronics were first introduced, at one point Kathleen Kinder wrote a book on electronic knitting (? 1984), exploring the full potential of managing designs by combining settings and flipping buttons for both the Brother 910 and the Studio 560.

A bit of review on machine buttons and functions for the Brother pattern case for those unfamiliar with it.  M= Memory: each of the tiny red spots on the garment representations lights up, as specs on motif are entered or reviewed. CE= cancel entry: corrects programmed numbers or cancels the red error light when it flashes. CR= cancel row: press in a number, say 2 on the panel, and the card moves back 2 rows. If you press the button and no number is entered, the error light flashes and the card stops advancing. This is the same as locking the punchcard to repeat a pattern row. RR= row return, brings the card back to the set line. This is routinely done before shutting off the machine when knitting is complete or to remove the mylar for editing.  CF= card forward. The mylar returns to programmed design row 1.  Numbers pressed in using CR or CF do not change those programmed using the M button.

When the pattern selector button on the 910 is down, the pattern is centered on green #1. A reminder: Brother has 2 needle #1 positions, one on each side of 0. When the pattern has an even number of stitches, it will be centered with half that number of stitches divided evenly on each side of 0. For 24 stitches, the pattern’s limits are yellow 12 and green 12. If the repeat is an odd number of units wide, the center stitch then will be placed on green #1 (right of 0). If the repeat is 25 stitches wide, then the pattern limits are yellow 12 and green 13. If using a 24 stitch repeat, the machine automatically knows that the first needle position (FNP) for the pattern is yellow 12. When the pattern selector is in the upper position (motif A) and the middle position (motifs A & B) we can choose the FNP, and with A & B the number of stitches separating them. One way to produce filet lace as is to program A & B motifs. The lace mesh is the A motif, the “rose” the stocking stitch motif. The A motif can be on the left or right of the mylar, but it is always the taller of the 2. In most cases it is the dominant pattern. The starting row for the combined motif is shared.

In lace knitting transfers and resulting eyelets are programmed as black squares. Superimposed solid patterns in stocking stitch occur in unmarked areas of the mylar or downloaded image, so they need to be “white squares”. In order to get the 2 to meet, the mesh repeat in the height required (column A in illustration), and a single width is drawn or programmed in reverse (column B), with the result being  read as (column C) when it is programmed as the base for a complete overlap. The basic mesh becomes the A motif in programming the 910, the stocking stitch “shape” is the B motif.  If end needles are selected during knitting, they need to be pushed back to B manually, or use the orange L cams. It is possible in addition to mark the “L window” on mylar, but the mesh repeat is so regular you may not find that necessary.The “rose’ would also need to be drawn in black, and positioned or programmed, with first placement resulting on the image seen below right. Color reverse is then used converting white squares to black, and lace knitting could proceed based on the black markings resulting in eyelets.  As is already noted, there is no guarantee the image placement on mesh the will yield a precise shape or the best possible results for motif edges, its definition, and its segments’ outline In terms of placement: if the all over mesh is programmed centered on G1, and the motif is positioned with FNP other than G1, any simple, extra rows of mesh prior to starting the all over pattern (below green line), will need to be programmed separately with adjustments also in FNP to match the superimposed segment. The programmed repeat for the mesh “rose” below would begin immediately above the blue line, and the extra mesh rows at the top would provide the transition to the start of the rose once again. The height of pattern seen in the B column in the first illustration may be adjusted accordingly. 

Color reverse (button #6) will provide “black squares” for creation of transfers to create eyelets. The mesh and superimposed design need to share the same starting row. Image above reversed, shows extra white at the bottom of chart that needs to be eliminated. Brother lace starts on a selection row (black squares), ends with 2 blank rows at the top of the pattern (white squares). This is reversed in Studio knitting.

What happens when starting row placement is the same for both motifs, and color reverse is used: the first row lace carriage selection is good to go. The height of the mesh above the rose may be adjusted to suit. Trying to place multiple roses in different locations on the finished piece using this method is more than my brain wants to even consider. Punchcard knitters are not completely out of the superimpose loop. If you are so inclined, areas on mesh punchcard ground may be taped over to test the repeat. Again, this works best for simple shapes. Tape may be shifted or trimmed as needed. If intended for extended use, trace holes over a blank card, punch the final version, and proceed. Images on punchcard machines are reversed on the stockinette side. If direction matters, flip the card over horizontally, mark and punch, or if the card is already punched, work with lace carriage on right, knit carriage on left. No worries about multiple programs or a mix of starting points, etc. Know the rules for where to begin and end for lace knitting.

Sharp single stitch points are not attainable. In the illustration below, the yellow indicates the “desired point”, red squares indicate the additional stitches knit, adding to the intended shape as the alternating directions transfers are completed. The one stitch start, is actually converted to 3 stitches, 3 to 5, etc .

A lace mesh series: using GIMP

Eons ago I owned a BitKnitter for my Passap machine, and to this day I miss it and some of the other Cochenille software that is no longer available, especially when working with multiple colors for color separations of any sort. When using it for downloading the resulting motif, picking the same exact color for each “square” on the grid was a necessity for accuracy. If “only” black was involved, the same guide applied. Working in GIMP one can set the color mode to fewer colors to start with, reducing the palette. This can limit your editing ability sometimes, and switching between modes may be required. I tend to go autopilot when I work on color separations for any purpose and work with built in colors rather than custom.

It is helpful to have previous experience with any program as well as with basic knitting  before tackling and combining large scale designs and  fabrics such as lace. This is not a step by step how to use GIMP (search for my other posts on topic), merely an illustration of my experiment in working with superimposing an image onto a mesh grid.

To start with: a 1 X 1 grid is set up. “Black squares” become single pixels in downloads. For the single stitch 1 X 1 grid to be visible, I prefer to work in at least a 500X magnification. Because I do not own a hacked machine I am unable to provide a corresponding test swatch to the final chart.

The resulting image may be tiled to produce a large enough mesh ground, the grid needs to be turned off for this step

The tiled image will appear on a different part of the screen, shown in both versions below, I realized reviewing the above that the width was for more pixels than could be programmed on a 200 needle machine, I resized its width to 120 pixels

here the grid is removed for further processing 

the goal is to retain the black squares for the end bmp, so the ground is filled with a color that will later be removed gridded again if and when needed 

Today’s image is a letter (20 pixels by 24), produced in a new file with text from one of my built in computer fonts. I happened to be working in RGB mode the same image as it would appear in indexed mode Here some edges “clean up” has been worked with grid on, then removed for tripling the motif length (now 72 pixels) and pasting on the “mesh” ground 

Switching to indexed colors (6) if you have worked in RGB to this point, it may make cleaning up of edges easier.  I flood filled the A to insure only one red was in use. Below the A is selected (fuzzy selection tool) and copied for paste onto the “mesh”. Magnification reduction to 500X makes the whole file more easily viewed on my mac

the A pasted in placewhite color fill leaving only black squares of letter and ground

Superimposing may be used in some machines, but eliminates the choice of editing when one is “not happy with edges” until after sampling the knit, has several parameters and limitations. Above image gridded for editing:
feeling better, with the exception of the left side bottom of right leg of Aone last bit of clean up, switching between magnification  as needed happier grid removed, image magnification reduction to 100X for export as bmp and download to KM

When downloading a large image that constitutes the width of your piece, programming an extra square or 2 vertically for the height of the pattern in “white” on each side will insure that stitches in those areas are not selected for transfer by the lace carriage, and a knit border will be created.

Things become more complicated with more complex shapes. This is part of a filet crochet chart, and a rose now becomes my goal

use selection editor or fuzzy selectpaste and move to “best” spot not quite a rose

Back to the drawing board: the mesh and knitting method are altered. For a possible knitting method option see  The ground “mesh” is now composed of every other stitch, alternating every other row. The “rose is not lengthened. “Stitches” need to be “cleaned up” to approximate a more recognizable shape.

 When satisfied, export in format for download.


Return to circles, knit “pies” 3

Elizabeth Zimmermann provided guidelines for circular shawls in her books and publications, including “Knitting Workshop”. For a basic pi shawl (p. 112, Schoolhouse Press, 1984) the assumption is that each section is twice as deep as the previous section and has twice as many stitches. Below CO row represents cast on stitches if the work is to begin from the center out, Column A the row count on which the increases take place, column B the number of rows knit just prior to increase row (A-1), and C the number of rows available for any planned repeat (A-2), thse are constants. The columns directly below each cast on (CO) number (orange) counts represent the number of stitches when increases are complete. The stitch count doubles when the number of rounds has doubled.Mary Thomas’s Book of Knitting Patterns, Dover 1972, p.p. 245-247 provides guidelines for circular medallions. She calls her first a “disc” medallion. In executing it the aim is to scatter increases so they are less visible and do not form spokes. Four stitches are cast on, with 4 stitches increased in the total count every other row. The number of stitches between M1s increases by one on every other row. My chart happens to read from left to right. As with any knitting in the round, the process may be reversed, starting at the circumference and moving toward center. I personally like charts to help visualize results, and have revised her counts in the illustration below so increases are at the same rate, but placed a bit differently within the rows. On rows with even numbers between decreases, start row with half that number of knit stitches before the first increase. Because one is knitting in the round, with knit side facing, all rows are knit. If the work were knit on 2 needles, knitting every row would produce garter stitch.what happens if increases “line up”For her circular “radiant” medallion after the first 2 rows increases are made every 4th round. My chart is renumbered excluding the first 2 rows, so the increase rounds would occur on numbers divisible by 4, making it easier for tracking them. Each “building” round increases the number of stitches by 16.

In her “target” circular medallion, the building increases are arranged in concentric circles. Each increase row begins with a M1. Once RC 20 is reached, a stitch is added between increases on each increase round. This chart reflects the knitting progress, but not the shape. STS column on right reflects the total number of stitches after increases have been made. Each building round after RC 6 increases the count by 32. Formulas for more, varied geometry based medallions are also offered in the book.  I finally “discovered” actually using formulas in Excel! The video that clearly and quickly helped me learn how to: at the flow in table form for the first 2 medallions

These formulas do not take into account changes in gauge, or stitch type within bands. For similar shapes to be achieved in machine knitting, the number of transfers would be prohibitive. In order to achieve similar shapes one begins with the radius of the finished circle and the shapes in the family may be knit sideways, using holding.

Handknitters can work with 4 double pointed needles, one or 2 (or more) circular needles, and crocheters can follow similar shaping methods. The advantage to long circulars is less bunching up as the work grows, and if you like working flat or want to try the garment on while shaping it, you can use more than one long needle, making the piece or the try on manageable. Working from the top down when knitting such shapes may give one more control over the size of the finished piece i.e.. on length of body and sleeves, height between bands, extending a yoke into a shoulderette or cape. Stitch pattern size and repeats add to the math calculations. Garter stitch is the only hand knit stitch that approaches a square gauge, could be used in combination with patterned bands.

The charted patterns above rely on M1 to increases. Yarn overs may be used for decorative holes at increase points. If preferred, the hole may be diminished by twisting the stitch when picking it up on the next round.

When knitting in stripes, the “jog” at the color change in knitting can be eliminated by slipping the old color purl-wise, and starting to knit the second stitch. TECHknitting provides more alternatives in her posts: For a method using yarn ends and a needle when yarn is cut

For shawl shapes and their geometry using YO increases, see the posts and publications by Holly Chayes.

To start it all from center out: I am used to doing the magic loop cast on with a crochet hook, and then moving on from there, Kitty Falol shows it worked with DPNs.

Return to circles, knit “pies” 2 (round yokes and more)

Just as other knitwear styles have varied in style, ease and fit over the years, round yoke sweaters have also done so. Yokes can be wide or narrow, in patterned or textured stitches, and in varied proximity to the neckline. This is not generally a tailored style. Ease in knits can be calculated on the basis of fashion or personal preference. With some familiarity with slopers measurements may however, be adjusted in this style as in any other sweater. Neckline measurements do not reflect the measurements achieved after adding finishes i.e. turtle or round. Depending on the size of the yoke, shaping can begin at the armhole level bind off (seen in the early hand knitting directions in the 70s), while smaller yoke shaping can begin at whatever point is desired, extending to the neckline, or simply to create a design band. The shaping is created by decreases if the garment is knit from the bottom up, and with increases if worked from the top down. In most styles the same number of rows are worked from the armhole bind off or held section to where yoke sections meet. At that point if hand knitting on circulars the 4 sections: i.e. left sleeve, front yoke, right sleeve, and back yoke may be picked up and joined for completing the yoke. My illustrations have been created using Mac’s Pages lines and shapes.  They are not to scale.

Beginning to visualize process: yokes are generally superimposed on raglan shaping

they form part of a flat circle; here is how they might appear in a partially seamed cardigan without front bands. They may be created in varying widths or patterns,

and in a pull over with shaping in back that raises the rear neckline. Some of the early patterns were executed with front/ back and both sleeves sharing equal measurements and slopes

separate the elements: the yoke 

the front and back can be begin to consider shaping at breasts, waist, and those wedges under where the yoke “circle” meets the sweater may be short rowed on each side with the intent of achieving a much better personal fit


Hand knitters are probably familiar with Elizabeth Zimmermann and her daughter, Meg Swansen. Handknitting with Meg Swansen 1995, and Elizabeth Zimmermann’s Knitting workshop 1981, Knitting around 1989, Knitting without tears 1971 are classical references that include information on yoke creation, including these guidelines

Other authors suggest 1/4 of total body and sleeve measurement in stitches (excluding armhole) for a tighter neckline (turtle neck), one third for a more open style (crewneck). Original circumference/ body measurement should include any ease. Though decreases for the yoke: first halfway up 25%(one out of every 4), second 3/4 up 33% (one out of every 3), and last 1/2 inch before full depth is reached 40% (two out of every 5) are the most common, they can be placed and varied to suit your own design.

In drafting your own patterns and partnering with someone, a tape measure or string can be laid on shoulder line, etc. for an idea as to preferred placement and measurement. Necklines finished by bound off stitches whether machine or hand knit, do not stretch, so measuring your head with something that does not as well gives you a guideline. Yokes end in open stitches, so the thing to consider there is what method is used to finish the neckline, and its own stretch factor after bind off. Hand knitters have the added benefit of splitting the work on 2 circulars and trying on the sweater or its pieces on while in progress to double check fit.

Japanese designers began to publish patterns that often included yokes that were constructed on the top of the drop shoulder line, with the back yoke placed higher than on the front. Such yokes also began to be represented in stepped figures showing decreases. In the round calculations are gauge based, not relying on the pi formula.

modified raglan for higher placement of yoke on back of body

full pattern with traditional full cap sleeve

pieces meeting at dropped shoulder line: dotted line represents back collar placement, note difference in height between back panel on lower back, and front panel on lower right

a sample diagram from a Japanese magazine 

Yoke shaping may be indicated in stacked format. The final count and frequency of decreases is shown, publishers may vary in language. On the first row here 4 sts – 2X means there is a group of 4 stitches followed by a decrease 2 times, then 5 stitches followed by a decrease 23 times, etc.

Two online calculators are available to help with DIY:  1. the Yoke-U-Lator, and 2. for Lopi, Icelandic styles. The image below is a screenshot from the latter website, indicating a sample possible result.

Jessica Tromp offers free circular knitting patterns with round yoke, dimensions in inches and ounces.

There are endless possibilities for combining math formulas, gauge, and pi. There are many ways to do decreases. With planning so that much more frequent intervals happen between decrease rounds, the decreases themselves can be fabricated to line up in line, and the resulting texture creates the interest in the sweater as opposed to any color patterning (i.e. along white lines)

“pie wedges” may be placed on neckline, yokes, sweater parts, various silhouettes and garment pieces, or full shapes (red dots outline possible dolman sleeve)from a Japanese magazine a hint of detail that must be calculatedand the pie may be oriented in different locations on any one piece 

from a Japanese knitting magazine, an idea for long sleeve and side details merging with and becoming part of a circular yokeFor some of the math  calculations please see:

“Decreases” in rib sometimes can be achieved through changes in needle size if hand knitting, or tension changes on the machine. The yoke in machine knitting would need to be split in 2 parts or knit sideways. Plain colored rows between bands of FI may appear noticeably lighter in weight, so using a 1X1 one color FI pattern or double strand of one of the pattern colors may improve the look.

Before transferring stitches on the machine in the single color rows, make your transfers. The lace carriage may be used after selecting appropriate needles and putting them in position. Knit the following row before removing knitting on waste yarn or garter bar. If stitches are tight for garter bar use sometimes the row after transfers may be knit at a looser tension to facilitate the process, and the difference may not be noticeable when knitting at “normal tension” is resumed. The carriage should be set to plain knit for row prior to and after transfers. It may be easier to work toward the center from each side when returning stitches to the needle bed. In order to match the pattern at the shoulder seams or when motifs need to stack in position on separate bands, the stitches need to be rehung at specific positions on the needle bed that take into consideration the size of the repeat and its location within the stitch count. Also, take into account the seam allowance. One stitch extra on each of the meeting seam sides will allow the end needle selection stitch or an extra patterning needle to be hidden within a full stitch join. Working on machines that preselect needles or pushers makes tracking a bit easier. It is possible to combine knitting pieces in both directions. For example, knit yoke up toward neck, join shoulders, and then pick up appropriate stitches to knit body and sleeves from the top down. Top down makes any adjustments in length easier prior to finishing the sweater. Short rowing in garment segments underneath the yokes makes for a better fit at bust line and upper back.

Calculators to help with all that math: online

for purchase:

Going low tech: if gauge works out to whole numbers, shapes can be plotted out on square grid graph graph paper (or grid created within software to suit) where each square represents one stitch, one row. Draw connecting lines, follow outline, filling in squares (or removing them) as the edge moves a whole unit (k2 tog). Here the goal is to go from 39 stitches to 5 over 60 rows. Color bands could be added and planned between decreases, which should occur on single color rows. Once a gauge is obtained, charting on graph paper or within programs can be boiled down to connecting dots and following outlines as above. Some simple breakdowns for outlines of garment pieces/ shapes 

more inspiration from an old Japanese magazine

visualizing a peplum