Recently (2024) I have begun to review my lace edgings once more.
The repeats are planned so that the first row of lace transfers happens to the right rather than the traditional left. This fact allows for
1: consistent knitting with 2 passes of each carriage and fewer passes of wheels and brushes on the yarn being used
2: end needle selection being canceled on the knit carriage, and turned on on the lace carriage, with no manual adjustments in end needle positions necessary at any point.
PNGs are provided, they may need to be mirrored depending on your machine model and download software. When copying the pngs they may be changed to RGB mode by default, use any program to switch them back to indexed, BW, 2 colors and save in the new format prior to attempting to knit them.
Edgings may be also produced beginning with the first row of LC transfers to the left, but the designs will be longer, requiring extra blank rows to place the respective carriages properly on their respective sides, a topic that can be addressed in a future post.
Looking back at my post from 2018, written while using the older version of Ayab software and working on a 910, another including an edging written this month(7/20).
The mirrored repeat
its 22X48 png 
Most of the more recent proof of concept swatches have been knit on a 930 using img2track for downloads.
A reminder: although I make an effort to include punchcard knitters in my writings, the full repeats as given and used for my test swatches and knit on an electronic are not suitable for use on punchcard machines as they are.
Electronic machines advance the pattern a design row with each carriage pass. Punchcard machines do not advance the card as the second carriage comes into use from the opposite side, repeating the action produced on the previous row.
For a discussion of some of the required design differences when using cards please see the post: Doilies: Lace meets hold and goes round, where short rows are combined with lace transfers, as in this sample. 
When using the slip stitch setting and changing the number of needles in work, or having ladders created by having needles out of work in the A position, the knit carriage end needle selection needs to be canceled (KCII).
Many early published trim repeats were intended for use with hand transfers combined with plain knit rows. Such designs may be translated for use on electronics and automated with this method.
At first I used a punchcard machine knit (KC) carriage with an added magnet as my KC, then on the 930 I found I could use one of my 910 carriages but not the other, the second fit the bed too snugly to make it usable.
The arrows indicate the direction of the adjustable slots in the LC mechanisms.
In a working later model LC, end needle selection is in work (W) with the slots in the horizontal position, out of work (OW) in the vertical position.
The 910 LC is also missing its magnet (left), gluing on a magnet in the position shown below made it usable again on the electronic 
Reworking some of the former swatches in no particular order, beginning with this one:
The 2018 repeat is charted on the left, the amended 24 stitches wide repeat is in the center, and the mirrored image for download to my 930 is on the right. 
Having the repeat 24 stitch width allows for positioning the knitting other than on the very center of the machine following the markings on any 4.5 mm punchcard machine needle tape. If this is done in electronics, the design should then be programmed for all-over patterning, not for single-motif.
Splitting the bottom 2 all-black rows of pixels/punched holes and moving the one row to the very top of the design is another of the new changes.
I prefer to plot out these repeats at the start of the decreasing angle. The first row then sets up the needle bed preselection on the widest stitch count of the pattern. The original repeat was charted using Numbers and scaled in Gimp to produce a downloadable file.
The method as it was worked out in chart form, and symbols used
The resulting, more successful swatch 
revisited 2024, the repeat PNG 
with added eyelets formed with all transfers in the same direction along the straight edge 
the proof of concept swatch: 
2018 failed attempt at a continuous pattern
Charting it out anew: on the right is the amended repeat adding 4 more rows to the top of the original
The resulting swatches:
A= the LC switched to fine lace “accidentally” for a few rows.
B= on the very edge eyelets are single on decreasing angles, and double on increasing ones.
C= Though the very edge has those differences, the number of knit stitches between the inner shape on the left and the outer one on the right is fixed.
D= the lines created by the transfer sequences to create the diamond shapes are far more successful.
These charts illustrate the above repeat on the left. On the right, the edge transfers are imagined reduced to single eyelets on the increasing angle. In turn they would result in a wider area of knitting in the zigzag shape between the curved edge and the diamond shape. Another possible solution follows, using expanded graphs with extra LC passes 
Returning to the above repeats after a break in 2024, a new design, 24X48 is more resolved, columns of knit may be added along the straight edge, with or without planned eyelets, and the shape at the center of the XX also offers enough room for forming a DIY smaller shape. 
“A Machine Knitter’s Guide to Creating Fabrics” by Susanna Lewis (1981) is the ultimate resource for punchcard knitting for knitters with any amount of experience. On page 223 she offers a repeat for a lace trim. It is shown on the left below, after being reorganized to start on the full 24 stitch width row of the edging. In the center chart, rows of black pixels are added, and on the right, the full electronic repeat is shown mirrored for download to the 930
2018, a 26X70 stitches wide repeat beginning at the narrowest part of the trim
2020, a 24 X70 stitch version beginning at the widest part 
The curved edge is noticeably different and more uniform than in edgings where the increasing angle is formed only by a single stitch being transferred to the next needle on the left, creating an eyelet immediately to its right.
The above design, as well as the Brother one published and shared at the bottom of the 2018 post, add two more passes of the LC for stepped decrease/ increase shaping. Three needle positions are involved in each sequence. The decreasing angle will have a single edge stitch, followed by a right-hand transfer with 2 stitches on that needle, and an eyelet to their right after the four LC passes are completed. The increasing angle will be formed by two single edge stitches with an eyelet to their right as well after those four LC passes are completed. 
Moving on to the Brother published chart at the bottom of the 2018 post: the original is on the left, readjusted for planning to begin with the decreasing curve in the center, and charted for download on the right (mirrored in turn for use on the 930). The original was said to be 68 rows high but proved to be 72. It is also 18 stitches at the widest point, not 24. I did not follow the publication’s directions for alternating between fine and normal lace either, simply left the LC set to N.
the png for the repeat as drawn, 18X72 pixels
The test swatch
A fellow Ravelry member reminded me of Tessa Lorant’s lace publications. I had forgotten I actually owned this single one from 1981.
Upon examining it, I rediscovered her patterns, many for hand knitting with accompanying written row by row instructions and charts, others at the back of the pub, with punchcard machine repeats.
The 24 stitch card designs provided were for use on Knitmaster/Studio 260 and 360 machines. The repeats are typical of lace often referred to as “simple”, a specialty of carriages that are capable of transferring and knitting in single carriage passes. For some ways to use such cards on Brother models please see post, or search subsequent shares.
The increases and decreases are achieved through row tracking and hand manipulation. Many of the edgings pictured are very open, and the suggestion is made that they be starched. Using different yarns and working in trims that contain more stocking stitches in their body make for more practical use nowadays.
Transfers are made in the same direction, which may factor into the results biasing. The edges in some, in addition, are shaped with multiple increases and decreases. If knitting long strips, a small piece of scrap yarn with weight on it may help keep the cast on stitches knitting and transferring properly.
I am not sharing any of her published directions, only providing ideas for analyzing and converting some of the punchcard repeats.
The first is from page 48, marked up for identifying transfers to the right (later decreases, magenta line), and transfers to the left (later increases, cyan line). A template with solid black pairs of rows to indicate slip stitch knitting sequences is a good place to start and can be expanded to suit the repeat. The yellow squares on the right indicate a beginning plan for increases and decreases. They and black squares to their left would be eliminated from the final design. 
Working one chart through to knitting: the bottom left image shows part of the original card. Magenta squares indicate transfers to the right and cyan ones to the left. The straight edge border with the larger number of eyelets was a bit fiddly to knit, so the second option is also offered and tested.
The repeats were mirrored for actual knitting on my 930, the 24X24 pngs: 
I knit to the left after all the needles were preselected at the end of the decreases and then cast on over the empty needles on the left, bringing them out to hold so they would be knit for the second row as the knit carriage returned to the right side. A garter carriage weight seemed to be enough to help anchor down the newly formed stitches during subsequent transfers. 
Making a pattern design more one’s own: part of the original design from p. 52, split so the pattern may begin on the widest part of the repeat, punchcard “holes” marked for left and right transfers
Here it was charted out in Numbers with the intent to produce the 2 stitches stitch edge along the outer curve and a less open one on the straight vertical side 
This sort of knitting can be a bit fiddly. I believe if I were to produce any such trims in great lengths and had the option, I would choose to dedicate a punchcard KM to the project so it would be easier to step away from and return to it as the spirit moved me.
If any loops are formed (A) and rows are not unraveled to correct the situation, it is best not to tug hard with the work on the machine. Stitch sizes on several needles may be altered, and the movement of the pairs of eyelets may become distorted. Because the knit carriage must be moved to the far right to be disengaged from the belt, this will tend to pull down extra yarn from the tension mast. A very gentle tug at the start of the first knit row should keep loops from forming.
The condition of the latches and needles is important. In stocking stitch knitting one of the signs of a “bad needle” is stitches tucking repeatedly on a needle that is not intended to be patterning in any way. The stitch in the hook is joined by a loop that does not knit off properly. It helps to start in a light color and a “friendly” yarn. Clearly identifiable stitch formation helps to develop an understanding of its structure. Eyeing the row of transfers upon their completion helps pick up improperly transferred or dropped stitches, saves runs that result if the latter go unnoticed before continuing to knit. At times there may appear to be a needle problem with a loop sitting over a needle or a dropped stitch and attempts to form “properly” knit stitches in transfer rows will create simple stockinette. Eyelets will not form properly, and stitches will want to get longer and not maintain their shapes. B and C point to the resulting differences in structures. 
The last Tessa sample, from page 60, brings up the topic of large eyelet lace, also revisited in a recent post. I thought the 24-stitch repeat too narrow, I expanded it to 32 and added eyelets on the straight edge.
I began by splitting the card in order to begin the pattern at its widest point and decreased the number of eyelets in the shapes to allow for automatic shaping. In these punchcards, rows with no transfers are part of the program, are free of any holes. Black pixels are used to program slip stitch all knit rows.
More transitions in planning, a 32X32 repeat:
two of the rejected swatches 
Consideration needs to be made in terms of whether some small ladders are a design element or not, hooking up “floats” helps reduce or eliminate them on decreasing angles on both sides of the shape, but the resulting eyelets appear a bit larger and different than the rest.
When at first I attempted 2 single eyelets following each other at the intersection of the shaps I needed to cast on after a single float was created between the center needles and followed by the first all knit row from right to left. That turned out to be unnecessary in the final swatch with the number of eyelets increases immediately after the single one in the final repeat. Using it will place a transfer left and a transfer right, holding down the loops in the 2 hooks affected on the second knit pass to the right. Leaving the bottom float alone echoes the remaining fabric.
The image on the left illustrates the result of not casting on on that center eyelet if it is to be repeated. The other photos show picking up floats and hooking them up in needles not receiving any transfers. 
The Tessa trims are knit in fine yarns. This shows a portion the inspiration large eyelet trim, note the appearance of the space between the two center eyelets is different
My version with the floats hooked up and onto the needles with no multiple stitches already on them at the center eyelet as the pattern shifts:
Sometimes less is more, here extra hookups are not used, making the trim quicker and easier to knit. At this point, I am experiencing sticky latches, not improved by light oiling, likely to be fatigued from constant use of the center 24 stitches. The needles do not look obviously in need of replacement. One alternative is to swap and shift the center 24 needles to a less used part of the needle bed.
Another image of the Tessa sample in repeat followed by my own test swatch 
At the top of the piece I did have to deal with closing the eyelets to facilitate 2 knit rows and binding off. I prefer to work from left to right, dealing with the right side of the loops first: drop loop on the right side, use a tool to enter the float created from its back 
twist tool and yarn clockwise
lift the e-wrap onto the empty needle on the right 
Repeat the process with the loop on the left of that needle pair, and do so across the bed before continuing to knit. The wraps will create full stitches on the next pass of the knit carriage.
An example working with a randomly selected pattern from Stitchworld, #156: 
A: the pattern is divided and shifted to plan the trim beginning with decreases rather than increases
B: with the superimposed table grid in numbers, the eyelet transfer squares were traced, adding all black rows for slip stitch selection of all knit stitches in between each transfer sequence
C: increases and decreases are planned by removing black squares
The image was then scaled in GIMP, and downloaded with img2track.
The straight-edge eyelet detail alternates the direction of transfers to the right and in turn to the left in each segment.
On the curved edge, increases and decreases were performed manually before each 2 knit rows. Empty needles need to remain in work positions throughout.
I chose to use the fully fashioned method for shaping, moving 2 stitches to the left for decreases, and two to the right for increases, then picking up from below the adjacent stitch to their left to fill in the empty needle. This was slower than having planned needle selection for eyelets do the work of shaping.
A very similar charting process may be used to construct circular “doilies”, where the slip stitch rows are used as a substitute for hand-selecting needles to holding/ short row positions. I have revised my 2011 post on lace meets hold and goes round/, and plan on following that up with one using edging repeats to achieve that type of shape.
single complete transfers 
Multiple transfers may be made either as a hand technique or expanded for use in electronics. Because single stitches are moved with each carriage pass, pattern repeats can become quite long, with few punched holes or black pixels 
Use a smooth yarn that has some stretch and does not break easily. Because the yarn will be transferred to and from or in addition also being shared between needles in fine lace, some extra yarn may be needed for proper stitch formation. In overall meshes begin testing using a tension at least one whole number higher than when using the same yarn for stocking stitch. Too loose a tension can result in dropped stitches or loops getting hung up on gate pegs, too tight and the stitches will not knit off properly or drop, or the yarn may even break. When eyelets are few, tension adjustments may not be needed.
A comparison of hand-to-machine stitch symbols with illustrations and more information: 
The two rows at the bottom of the card reflect the overlap when punchcard snaps are in use to keep the pattern continuous. 
Column identification at the bottom of the chart:
Spacing out the zigzags, another 24X30 repeat. This is the minimum repeat for electronic KMs as well, knit stitch spacing (white squares) can be planned to suit 
The transfers of the stitches by the LC while using the knit carriage set to slip in both directions to create the knit rows, and will automatically create increases and decreases along the left edge. Due to this fact, there will be one less eyelet in each transferred row than the number of pixels/punched holes in its corresponding pattern row.
The pattern repeat on the left below is as I drew it and intended it, on the right, it is mirrored for use to knit it on my 930
The first preselection row is from right to left, the knit is centered with 10 stitches on each side of 0.
The above fact allows for planning transfers in both directions while still keeping the routine of 2 passes for each carriage to and from their original home.
The fabric was capable of changing considerably in look and width that could be encouraged to remain more permanent with blocking. There is a visible asymmetry, with one side of the chevron actually containing an extra eyelet. Still trying to retain the 24 stitches in width design constraint, I began to work with simply counting eyelet transfers matching in number, guessing rather than planning. Tiling the repeats can help get a sense of how things line up horizontally and vertically. Electronic repeats can be minimal unless one is programming the pattern as a single motif that includes edge knit stitches etc when downloading via cable to an electronic or using the Ayab interface. 
A and B continue to produce uneven numbers of eyelets on each half of the resulting “V” shapes. The greater the number of row repeats of eyelets in a single direction, the more the resulting bias. If asymmetry is the goal, then this may meet the need. I knit most of the swatches on 48 stitches, with more needles added on the right if needed to ensure the edge stitch will be a knit one. Larger shapes require wider tests. The photo is rotated to reduce its length on the page 
The added eyelets in pattern C produce an interesting change from a sort of V shape to more of a W, but the fabric is still unbalanced 
Back to the drawing board: a different mesh, with eyelets in alternating numbers, resulting in a more balanced fabric. Here the charted repeat is shown X2, side by side. 
Using electronics one may expand or reduce the number of stitches in the repeat to reach an estimated equal number of eyelets 
With any mesh, the number of knit rows may be varied between each pair of transfers. My swatch was knit using a 26 stitch repeat width 
Exploring a smaller repeat, also 24 stitches wide 
The white yarn used is the same cotton as in the previous sample, knit at tension 9, the chenille is rayon, 1450 yards per pound, the eyelet patterns are stable and distinct, 
Another variation 
As often can happen in machine knitting, on a different day, the same yarns, same tension, same operator, and stitches drop, get hung up on gatepegs, and perform other unwanted actions.
The blue yarn refused to knit off properly, so the different added colors were tried to see if I fared any better using them.
and has a very different appearance when lightly touched with my now fiber burning iron 
During knitting, these fabrics will appear to be producing straight color stripes 
Taking the guesswork out of repeats, one may begin with or use published cards as they are, this is from a Japanese magazine, 
Here the cast on and bind off match very closely. I used the looser cast-on method described by the “answer lady” in the 
A very different chevron in an advanced technique combining tuck stitches with lace and fine lace patterning 
A variation of Card C, with 2 rows blank after each pair of transfers throughout was used in borders in some of my lace shawls including these, made in 2011. 
The knit side of the fabric is the most interesting. The swatches were first photographed as they were immediately after their removal from the knitting machine. I also tried to photograph them at an angle to show the protrusions from the surface. Pattern A: 
When using card A, the shapes alternate vertically between all mesh ones with all knit rows at the outer edge of the folds. Using Card B, all protrusions have a fixed number of knit stitch rows at the outer fold. Card C produces an all-mesh fabric that made me think of ocean waves somehow. Turned sideways, and pressed to set the folds, the resulting fabric could provide a springboard for a host of other, different ideas: 
A larger swatch started on waste yarn, with 2 rows of knitting at the bottom and at the top before binding off. The design is card B, with 4 rows added to each half of the design. I am still working with the constraints of a punchcard machine, and for the moment, of retaining symmetry in the width of the “scales”. The working chart, turned counterclockwise to save space 
the above arrangement will, in turn, need to be mirrored for the lace pattern. My sample was knit using 2/15 wool at tension 7, using needles 36 left to 35 right 
The knitting in progress 
Often the question of what fine lace is and how it is made comes up. Fine lace is a fabric against which I have a personal bias. It seems to involve a lot of work for the result unless one is using a light-colored yarn with a smooth surface to show the subtle and at times hard to observe surface changes. It fares better visually when mixed with eyelets. The yarn is shared between the needle that would normally be left empty and the one with the formed stitch already on it to its right or left depending on the direction of the transfer. If the shared yarn is dropped instead of staying in the hook of the transfer needle, as the next 2 rows are knit there will be eyelets on the ground lacking them anywhere else. Card C, knit using only the fine lace setting on the LC, looks very different from the previous fabrics, both knit and purl sides are shown, with a couple of spots where the yarn was not shared by 2 adjacent needles, forming eyelets
Now evaluating the possibility of mimicking my 
Planning out a repeat in chart form: 
Knitting began with working in a 2/15 wool, tension 6, using a 7 prong tool with all prongs engaged to transfer groups of stitches onto their new positions on the needle bed. The first two images review picking up from the row below to fill in needles emptied by transfers. It is one of the machine-knit equivalents for the M1 direction in hand-knitting patterns. As a result, the number of stitches being knit remains constant throughout the piece. 
The stitch structure so far appeared too loose to me, so I then switched to a 2/8 wool, at tension 7 with the intent to test knitting as tightly as possible while still facilitating the multiple moves of stitch groups on each row. The resulting fabric was stiff and wanted to curl strongly to the purl side, needed pins to help it lie flatter. The shapes refused to stay poked out to the knit side.
On electronic machines, one does not have the same limitations in terms of the width of the repeat maxing out at 24. For example, the number of eyelets on either side of the block components of the repeats may differ, the greater the number of eyelets, the wider the spacing between the 3D shapes. Graduations may be planned in height and width of repeat blocks across the full needle bed, limited only by patience in designing, the capacity to download, and the tolerance for both the yarn and the operator to complete wide, long pieces using the technique.
and in a staggered one, tested at the top of the swatch 
The “scales” were more like twisty bumps, with subtle variations vertically in some of the stitch groups. That said the results were dramatically different when color changes were added to the pattern knitting in the shorter repeat. Blocking may make an even more marked difference depending on whether the mesh (lightly steamed) or the tighter knit (vertically pulled, should be washed) produces the preferred effect. The looser stitches at the top outside edge ie at the top left corner are the result of a stitch that got away from me. The cast on was a bit too loose, I knit 4 rows at the bottom after the cast on and 4 rows at the top of the repeat before binding off. 
Adding a second color and reversing directions of shapes brings lots of yarn ends and its “price to pay”
Some handknit large scale inspiration to begin my revisit to MKing them: 
The chart is actually rotated 90 degrees counterclockwise, could serve as inspiration for an electronic pattern.
Automated holding sequences may be planned for single or multiple stitches in width, as well as for single and multiple rows in height. For the new initial test, which proved to need editing, this was my repeat
A tiny test in too thin a yarn 
I am knitting on a 930, the image needs to be mirrored in order for it to appear in the direction I intend on the knit side. The above repeat did not work properly when knitting a whole row of shapes. With some patience, a final, edited, and mirrored repeat was developed that enabled a completed a full row of shapes using the slip setting and holding, and starting with working it from right to left. Sometimes differences are subtle, especially in designing using single-pixel units. The new repeat proved to also work for rows of shapes in the reverse direction after horizontal mirroring and restarting the pattern or design row 1. When working from right to left, the initial preselection row is from left to right, while when working from left to right, the first preselection row is from right to left. After a full row of repeats is completed, the pattern is rolled back to row 1 and mirrored. Punchcard knitters could turn the card over and start again on the proper row. I used contrast color knit rows initially in between rows of shapes to help me note transitions more clearly. The “leaf” is not pointy enough for me, but at times what was not planned may lead to a pleasing result of a different sort.
To knit: cast on with a multiple of 12 stitches on each side of the center 24 on the needle bed
Moving on to a wider version, using 24 stitches in width to allow for using the pattern on a punchcard: this repeats works both as-is and mirrored, the groups of stitches moved in and out of work is now half of the new design repeat = 12. 
The lovely mess in the swatch happened when I stopped paying attention to everything but what was happening on the needle bed and missed the tangle of yarns in my yarn mast. There is enough knitting, however, to note that the repeat is sound and that the edges on both sides are formed by the narrowest part of the shapes in each direction. One way to solve that is by casting on and binding off along tops and bottoms of shapes as seen in the yellow and green swatch at the top of the post. 
Planning things out to release those edges as seen at the top of the post
Problems to solve: maintaining an even number of rows in-between shapes and a straight edge along both sides. The latter could happen with triangles prior to knitting full shapes at either or both ends, the first sample failed on the left side due to both triangles being knit in the same sequence; 
Begin on waste yarn, decide on the color of the cast on, and any additional knit rows prior to beginning in the pattern. Each of the side triangles is shaped using manual holding techniques over 12 stitches. If starting on the right, the first preselection row needs to be made moving from left to right as above on the first 24 stitches. With knit carriage set to both slip in both directions and holding with COL make certain the first 24 stitches on the right in B position make a free pass to the right.
Observations: there are elongated loops on the knit side of the fabric, likely created with slipped stitches. There appear to be eyelets on the purl side. The total number of stitches remains constant throughout the knit. The row repeat spacing is fairly close, so at least to start with I tried single repeats to achieve the look, was not pleased with any of the initial results. That led me to an online search for what would visually appear to have similar qualities in handknit samples. Knittingfool.com is an extensive resource and, to my eye, this slip stitch pattern, 
as do “oats” found in a 1984 handknitting publication 
and then transitioned to translating any repeats for use in machine knitting, keeping in mind that in hand knitting actions are made as the row is being knit, while in machine knitting they take place prior to returning the carriage to the opposite side thus knitting that row. This was my first repeat after replacing cells with squares representing knit stitches with black and white planned pixels for electronic download.
The slip stitch setting is used in both directions throughout. White squares represent areas where needles on the bed are skipped/slipped, not being selected forward and thus knit. This happens for 3 rows, resulting in the required elongated stitches. On the 4th pattern row, the group of 3 not selected needles is where the stitch transfers and crossings occur. Any cabling, eyelet fill-in, etc, needs to be performed prior to knitting that row and moving the carriage to the opposite side. The work is always done with purl side facing, so matching the direction of patterning to a hand-knit may also require mirroring of crossings, depending on your knitting machine model. The goal is to have the K3tog with the long loops in front of the single knit stitch in the center position.
Move the stored original center stitch back onto that center position, there will now be 3 stitch loops held on the single needle 
repeat the process across the bed, bring all the needles used out to hold position prior to knitting the next row 
The yarn used was 3/8 wool at loosest tension possible, the result is subtle as any spaces between stitches get pretty well filled in. Moving on to denser patterning 
The elongated stitch crossings now happen every 4 rows, but across two needles, not one. I used a tightly twisted rayon for the test swatch thinking they might be more visible, but the openness in the stitch formation because of the fixed spacing on the metal bed confuses the texture a bit. Simply leaving the empty needle out of work and continuing to knit (1) created too large an eyelet to my taste. Trying to pick up the third slipped loop (2) had the same effect. The best result was obtained by picking up from the row below on each side of the three stitches that are removed and then returned to the needle bed (3).
e wrapping the third row of floats can be tested by inserting a single eye tool front to back, twisting either counter or clockwise and rehanging on an empty needle, thus casting on an “extra” stitch 
the appearance on the knit side 
Charting the actions for a hand-knit version: the top images illustrate the moves on the purl side while on the machine, below it those on the knit side when hand knitting the pattern
Plotting out borders and repeats for a small test including make-1 increases in order to keep the total stitch count constant. I do little hand knitting nowadays, so when doing so I add as much information as possible including some stitch counts until the pattern is established and I can visually follow it more easily. 
The resulting test swatch 
This is an illustration of the method I used to “make one”, in my swatch I kept the direction constant
The 910 in my default setting produced the “image’ as drawn on the knit side of the resulting fabric. The post was written prior to my tiling the repeats as a matter of routine to check their alignments. Doing so would have shown a couple of missing pixels, and pointed to any other errors in filling in mylar squares.
All transfers were made in the same direction.
As always, ideas need error-proofing and refining, easier done in a chart if possible prior to any actual knitting.
There are a number of changes to make if it is necessary to get the pattern to reverse direction in alternate rows of completed shapes in a brick layout.
Attempts to visualize holding can happen in spreadsheets, documents, image processing canvases, or even simply on graph paper, moving/ “drawing” back and forth across the cells mimicking carriage movements and marking them accordingly.
The 930 automatically mirrors downloaded motifs, if direction matters, as in these samples, either mirror the design horizontally prior to saving the png for download to the machine, or use the mirror button selection on the operation panel after the download.
while here the repeat for all full-size shells is planned, 
Marking up the needle bed with water-soluble markers or pencils helps track placements of repeats across the desired number of needles in work: dark lines indicate placement beginning with the mark for half a repeat to maintain straight side edges on the finished piece. Red lines mark the placement of the stitches when they are moved to the left in order to knit the full shells across the bed. 
Scaling the image to render it a bit more legible:
The machine will be set for slip stitch in both directions and holding. End needle selection is canceled.
I used two repeats side by side to eliminate having to consider and choose the position option on the 930 needle bed, resulting in having the pattern centered in each 24 stitch fixed segment of needle selection.
I was too aggressive with clipping yarn ends on the left side, especially while in the process of changing colors, not ever a good idea. Automating the pattern fully on electronic models using only slip stitch patterning is possible. The length of such patterns grows exponentially in proportion to the size of the repeats. Reviewing errors in the beginning concept 
An attempt to visualize the placement of the shape variations in the finished piece using the shell motifs beginning with the shapes created in order to create a straight side edge
COR: knit a base row in color one from right to left
Preserving the 3D texture relies on using yarns with “memory”, ie wool and avoiding aggressive blocking. Using thinner yarns makes the stitch formation more evident. Hard pressing, in this case, knit using acrylic yarns, flattens the fabric considerably, and often, permanently. Both the hold/slip (top) and fully automated swatches (bottom) are shown.
2023: a screen grab from the shell charts was cropped to its outline, opened in ArahPaint, and using the program’s tool “
A return to the original 14 stitch repeat, illustrating a way to begin editing for an extra row in width at the bottom of the shape and ending on 2 stitches rather than a single stitch at the top
A Prada sweater using similar shapes 
If you are interested in any large size clamshells, and intarsia appeals to you whether in hand or machine knitting, 
the common illustration for shaping triangular shawls using such motifs 
Increments in height need to happen at sequences of 2 rows each, so the design was then doubled in height, resulting in a scaled image now 33 stitches by 46 rows in height, with a planned horizontal repeat X2 = 66. Note: the sidebar offers start and end needles are given for pattern placement on the needle bed. Sampling may occur on fewer stitches than that. Since the number of repeats programmed to add up to an even number and center alignment is chosen, the number of needles is even on each side of 0. 
In my second series of swatches, I decided to try for a smaller “circular” shape, with the repeat now measuring 15 wide by 20 high, and a planned horizontal repeat X3 = 45. If centered, the software places the odd number of needles on the right-hand side of 0. 
The wider horizontal band of all knit stitches was due to operator error, happened when I pushed back preselection an extra time, resulting in the ribber only knitting extra rows. For the sake of added clarity, I have added color to the chart below, assigning yellow and grey to all-white design areas in the pattern. The black squares are what I choose to drop. For illustration purposes, this is only a segment of the repeat. 
drop the loops, return needles to B position. At this point, since all needles are in B a modified stitch ditcher may be used for 2 passes, dropping the loops on the first pass and returning the whole series back to B on the second. 
COL: push all preselected needles back to B, as you knit back to the right the next group of white squares (yellow) in the next design row will be preselected
COL: knit to the right in order to form loops on the main bed, continue for the desired number of repeats and end as suggested for the two-color version.
and its accompanying punchcard repeat:
UKI is no longer available. I recently acquired some 
Here the white yarn shows position beneath the pin 
and scotch tape in place to adjust the amount of pressure exerted on the elastic. 
Thread lace produces a single set of floats (in this case the elastic), the white border is not visible in the repeat below, it is the same as the above fabric, with colors reversed. I prefer the single float backing. In addition, for these fabrics the elastic is placed in the A feeder, the yarn in B
A second thread lace variation:
The multiple folds and creases, as opposed to smooth blister in all the above, are very interesting to me.
The spots, where the elastic and yarn knit together, are compressed, so the results are quite different than what might be expected from studying the chart 
A quick, imperfect sample using a fine cotton and a single strand of the elastic, each with its own upper tension disk adjustments.
Here the same pattern was executed on the same number of stitches with the white wool used above, but the elastic was plied with a 2/24 acrylic yarn and knit as DBJ with the blue, stretchy combination creating the 
I was curious to try a pattern previously tested in an all wool
Working with a similar arrangement I decided to try having patterning on the top bed, with the aim to tuck on the single needles in work in the repeat (bottom line of black squares). Not all ideas are an immediate success. Wanting to see the effect on the single-ply wool I began with that at the tension that appeared to be required for knitting it along with the elastic and stitches were far too loose.
I had more success and a bit better definition of the fabric by knitting with both strands together from the bottom up. In an effort to attempt to have better color distribution I used the plating feeder and had some issues with stitches not knitting off properly. The result was not significantly different than that obtained by simply feeding both yarns together. When using plating feeders both on the single and the double beds one of the yarns may have a tendency to jump out, to prevent that from happening here is one “hack”.
The same yarn as above, also knit in tuck setting, showing the difference in size and dimension between slip setting on left, tuck on right 
As with any knitting, keeping an eye on what your yarn is doing still matters. Such as this will lead to a series of circumstances that may bring your project to a far earlier end than planned ;-(
A very last effort at attempting the scales with elastic and wool knit at a far looser tension that in previous tests, ended when elastic broke as a result of above, IMO an unremarkable fabric 
For more swatches and information see post on 
The FI results are shown on the left, the first dbj test with lili buttons engaged and both carriages set to slip in both directions appear to the right. Note the difference in width and height of the resulting knit. The floats in FI are overlong, and there are some separation and lengthening of the stitches along the diagonal edges of the houndstooth shape 
The program does provide prompts as to which color should be picked up next, and clues as to where one is in the repeat, they are found at the bottom of the Ayab window. I happened to grab content with color B upcoming both times. 
When preparing images for download, they should be indexed to 3 colors 
It takes a while when knitting double bed knits before one can assess whether the results are correct in terms of patterning since the knit is hidden for some length as it drops between the beds. Because the Ayab prompts for color changes were altered sometimes when the carriage on the right, sometimes on the left, I dropped the first sample off after slightly more than one repeat, found its patterning to be correct, and that the prompts when on the color changer side for the following row to be knit from left to right were reliable, as shown in the knit swatches. In turn, I used an indexed greyscale image. The lettering on the right indicates where colors were placed following the give instructions, rather than where they were assigned in the original design. The manual states that the color sequence for the separation is: white, grey, black, gray, however, if the prompts for changing colors as given are followed, which is very valuable in tracking them, the knitting occurs in reverse order. 
Knitting mode options for DBJ are listed below the single bed one, 
The main bed is set to slip in both directions, the ribber as well, with lili buttons engaged. The number of colors is not altered automatically by Ayab, the change needs to be made manually. 
This was the set up in the color changer 
Planned possible use of the middle color twice may be seen in this image