WORK IN PROGRESS
I created large-scale nonrepetitive image garments very early in my knitting career using Cochenille Bitknitter and Commodore computers linked to a Passap E6000. Over time my focus changed considerably, with any production knitting moving onto accessories as I began to make items for sale in galleries and in shows, most often single bed on a Brother 910. If knitting is a primary source of income, one needs to consider production time management, material costs, and what the local market will bear in terms of pricing.
A post, written in 2018, began to explore two-color-dbj-non-repetitive-images-electronic-kms/. At that time I did not have a machine model capable of using img2track.
An orphaned 930 entered my life, and with rare exceptions, over the past few years, my blog sample swatches have been knit using img2track, which I have found easy, reliable, with any programming errors due to the operator issues including learning the differences from 910 programming and remembering to actually use them.
No matter how long any of us have been knitting, there can be many aaargh moments both in everyday knitting and when exploring new techniques.
I have a supply of lovely 2/48 cash-wool in royal blue, black, and grey. Three strands worked predictably on my punchcard machine in a series of my spiky scarves, shown in progress on the machine. 
Nearly all my previous dbj pieces have been knit on a Passap E6000. The 930 experience for such repeats is new to me. With some help from Tanya Cunnigham in reviewing steps required when using img2track, I returned to cellular automata repeat saved years ago.
I encountered problems with the triple strands of blue not feeding evenly, here both colors were picked up by the changer accidentally, I realized the issue, trying to unravel the row of knitting produced this 
That provided an opportunity to decide I preferred the reverse color placement as well as wanting a thicker ply for the white, resulting in twice the fun with 2 colors, and another scrapped sample 
Switching the white to a single-ply thicker yarn made its stitch formation far more manageable. The blue however seemed to have a single strand of the three with a propensity for catching on gate pegs. I tried tension adjustments, the usual tips in managing static. At about row 1,000 out of 1288 rows, I realized I had an issue with both yarns being caught on gate pegs. In trying to lift the stitches off, the yarn broke but gave with no immediate visible clues, the dropped stitches and a lovely hole, as a result, appeared when knitting had progressed far enough below the current knit rows. On the left, the work is shown still on the machine, while on the right, it is off the machine, and in the process of a patch job with a temporary accessory and stitch holder in place. 
I was able to achieve a reasonable repair on the knit side, but the birdseye pattern on the reverse is a bit scrambled. 
For folks that are not familiar with electronics and are curious, the 930 has the smallest memory of the later Brother electronic models. My pattern repeat is 74 stitches wide by 644 rows in height. The user manual explains: the KH-930 takes just a few seconds to load the track because the memory holds only 2 KB of data (about 13000 stitches). Later models have a much larger memory (32 KB). The KH-940 and KH-950i require 42 seconds to load a track. The KH-965i and KH-970 load only the requested pattern, so the loading time depends on the size of the pattern. img2track indicates progress as the data is sent to the KM. When the pattern has finished loading, the KM should beep, and show the green READY light and a 1 in the display (for row 1). The program automatically chooses Selector 2 for a single image and centers it. You may change this by using the normal pattern selecting process on the knitting machine, choose Selector 1 for all-over patterning, or use Selector 2 and choose a different location on the needle bed to center the pattern. If your pattern was divided into more than one track, you will have to load successive tracks when completing the previous track, specific instructions are given for programming subsequent tracks. My pattern was broken down into 4 tracks.
The cable used for downloads to the machine is used externally, no alterations to the machine’s hardware are required as when using Ayab on the 910. The pattern is stored in the machine, so the computer needs to be awake only during downloads, not constantly as in programs that use knit-from-screen.
Each track for 2 color DBJ using the KRC built-in separation is entered in numerical order as a new pattern with first-row preselection from the left to the right and first row knit from right to left toward the color changer. If the repeat is not planned for the number of needles in use, any position or change to all-over design needs to be re-entered, and the KRC button must also be set again.
Cam button settings are set according to the chosen dbj variations for either or both beds. End needle selection is usually canceled. In some patterns using it can create an interesting beaded edge on either side, which is worth testing on small samples to determine one’s preference.
I like to plan my pieces beginning with the dark color, plan my repeats with the deliberate placement of both colors, and any scaling in the pattern BMP prior to download, using Gimp. I also prefer to have color 1 as the dark, color 2 as the light. The default in the Japanese DBJ separation uses the light color, white squares, as color 1. Out of habit I color reverse my images so my first preselected row from right to left can just knit my black squares rather than the white, and I can continue my motifs as I intended while having machine prompts for each color also match.
Pausing knitting is easy as long as the needle selection is not disturbed. Ending with COR avoids any confusion about which color should be used next. Starting outside the set mark, turn the machine back on, and simply continue in pattern with appropriate color changes.
Tanya Cunningham manages and moderates the membership, settings, and posts for the Img2track – For Machine Knitters group on Facebook.
These were her tips and reminders to me for handling pauses in knitting immediately after the following track in the sequence is first downloaded: let’s say that either some needles got pushed in or repositioned, or for whatever reason, you don’t have certainty that the last row of needle selection is reliable, and you want to “re-select” the last row before you knit it, the last track you knitted should still be in the memory, even though you’ve completed that part of the pattern.
1. Push all needles back to Pos B.
2. remove the yarn from the feeder, disconnect the K carriage from the R carriage.
3. Turn Change/Selector knob from KCII to N (NOTE, this will cause your PART buttons to de-select)
4. press BOTH PART buttons.
5. Move K carriage to right. No needles will knit, since all are in POS B, and both PART buttons are depressed, AND no needles will select, since you’ve moved the change/selector knob to N, and the memo will not record any advancement of row. (However, if you’re using your mechanical row counter, it WILL record a row, and one on the way back so plan to either disable the ribber arm or plan to turn the counter back 2 rows)
6. Now you will have to re-select the last row of needles for the track you’ve most recently knitted. First, verify that KRC is activated. Now, you will have to push the up/down arrow buttons to select the very last row of the pattern which will be an even number, and color 1. Depending on whether the carriage was moved in such a way as to activate the sensor enough to cause the memo to advance, you may be able to simply use the row that’s showing, but even beginning the movement of the carriage may advance it. To be sure, what I do is to advance (in this case to Row 1 Color 1), and then back up one row, using the arrow buttons.
7. Move your Change/Selector knob to KCII, be sure to move outside the turn mark. Verify KRC; memo says last row, color 1; both PART buttons in. Now, move your carriage right-to-left, to select the last row of the previous track.
8. Load color 1 in the carriage, load next track into the machine, KRC selected.
9. Now, as you knit to the right, you will be knitting the last row of the previous track, and selecting needles for the first row of the next track. Carry on.
In terms of generating cellular automata math-based patterns, the Wolfram website is a great place to explore repeats. A player, temporarily unavailable to Mac users is presently available, allows for the download of interactive demos in .cdf format. In terms of knitting any of the repeats, the most suitable appear to be ones that are generated in black and white to start with. Not all are, and at times changing the mode to bitmapped in programs such as Gimp can produce a glitched effect. Though the latter may be interesting and desirable to some, I prefer clean lines and diagonals along with identifiable shifts in the scale of any triangular components.
I am often amazed at the speed with which time passes, previous related posts: 2015/12/09/cellular-automata-charts-for-knitting-etc/
Previously knit repeats 2017/09/11/my-new-knitting-projects/
Weaving drafts may also serve as inspiration for knitting repeats. Posts with related content: 2015/11/28/weaving-drafts-as-inspiration-for-other-textile-techniques/, and 2018/07/02/numbers-to-gimp-to-create-images-for-electronic-download/. These images are extracted from a draft for an advancing twill. One may explore segment placement and color reversals easily using programs such as Gimp. 
There is also potential for exchanging colors to get a sense of how the pattern might appear in different colorways 
My planned test repeat is 76 stitches wide by 556 rows high.
translated for knitting, patterning may be used on either bed, with the opposite bed set to knit every stitch. If patterning is on the ribber, have the first and last needles in work on the main bed. 
The top bed is set to tuck in both directions, the fabric is shown lightly stretched 
and could conceivably be used to create a ruffled edge when followed by narrower stitch types ie. every needle rib or Milano variants 
Changing settings: with the knit carriage set to slip in one direction, knit in the other, the ribber knitting every row
Needles may be taken out of work to create pleats in every needle ribs, alternating their placement between the 2 beds. Knit stitches stabilize tuck ones. Here every other needle is taken out of work on only one bed. The tuck loops are more visible in any open spaces between the vertical rows of ribbed stitches. The ribber will knit every row on the same needles aligning directly above each other, thus ruling out full pitch.
The needles are set up so that the first and last are in use on the ribber, ensuring that a knit stitch will be on the side of any tuck stitch selected on the top bed, 
on the top bed the every other needle tuck selection reverses as the direction of the knit carriage does, needles brought forward knit, the ones skipped hold tuck loops in their needle hooks, 
the resulting fabric 
Choosing spacing between needles in work on the main bed 
A working repeat with 4 circular rows, 6 full rows knit 
My first swatch in the technique was in response to a Ravelry member share describing the stitch pattern used in a finished garment 
followed by float formation when only the top bed knits 
The resulting grid may be used as a guide for hand techniques off the machine in isolated areas or all over 
Needle arrangements may be varied to accommodate thicker yarns, or alter the texture by varying both the number of knit rows and circular ones 
Because the top bed needles are in pairs, the EON selection remains evident, each of the first 2 passes completes one row of knitting.
When only the top bed knits, floats are formed between the needles in work 
Half Milano with tucked long stitch: the main bed needles will tuck one row, then slip one row. The ribber needles will first knit a row, then slip a row while the second bed knits both rows. Then the ribber needles knit one row then slip a row while the second bed knits two rows. 
I have had no success with trying to execute my interpretation of the directions without getting floats on the top bed, and the photo in Machine Knitter’s Guide to Creating Fabrics is not clear enough to distinguish if there indeed are floats on the surface of the fabric. The experimental repeat 
the tuck row is formed with the knit pass to the right, no preselection 
the row of stitches is skipped on the way to the left, preselecting for the next first knit row, tuck loops visible on top of the hooks of the selected needles 
the process is repeated. The resulting fabric: 
In DIY the potential for exploration is endless. As always, if needles are out of work on the main bed, end needle selection is canceled. Here slip setting in both directions is used, along with needles out of work on both beds, the 1X1 needle repeat is programmed.
the result has floats on the purl side, a slightly pleated look on the knit side. 
After both no preselection row bring alternate needles out to hold using any EON needle tool. In my case, COL, I began with needle 1 on the left on row 3, alternating beginning with needle 2 on the left on row 6. The number of rows in between hand techs is an odd one, so sides for the hand tech will alter as well. This shows the tuck loops formed EON as the carriages travel to the right.
Prior to resuming pattern knitting, needles with loops on them need to be returned to the B position, maintaining the EON needle preselection for the next pass 
Repeat the process when design row 6 is reached, beginning with needle 2 on the left. The texture appears on the purl side of the piece, shown on the left
This repeat uses 6 passes as well. The tuck loops on the top bed occur above slipped stitches in the previous pass, which may be replicated, but the real problem is that every third row on the ribber is also having to tuck on specific needles with no knit stitches between them. The tuck loops happen directly above knit stitches formed in the previous row if the tucking lever is changed manually from R to P on those rows. That is a lot to juggle, a no. 
Here the eon tuck would fall on slipped stitches in the previous pass, so that is doable. It is possible to fool the lili selection into believing there are extra needles in work on each side of the ribber, which can “make” the first stitch on either side knit rather than slip or tuck. The method is used in creating a 
This pattern repeat is a short one, the changes are more regular on the ribber. The hand tech tuck stitches could be formed for 2 rows with all stitches getting worked back to B position in between those preselected for the next pattern row. The lili buttons need to get fooled after alternating pairs of rows, the start of several nos for me.
Planning for interlock:
when the lili buttons are used with the machine set to slip in both directions. Studio or Knitmaster knitters would require an RJ1 ribber carriage. I refer to the marks on the Brother ribber needle tape as blanks and dashes. When lili buttons are in use, one must work on an even number of needles. Pairs of both are required. The starting placement on either of the 2 marks does not matter, the first needle on the carriage side on the left will always slip, with the last stitch away from the carriage always being knit. This observable if one “air knits”, slowly moving the ribber carriage from side to side. As in any pattern knitting, needles selected to D or E will knit, the ones lined up in B position will slip or tuck. 
Imagining the sequences required to complete a single row of knitting. I prefer to start with a knit stitch on the left, top bed, end needle selection is canceled 
When testing new ideas, particularly in rib setups, I often begin on a small number of needles and in a way that feels “safe”. This was my first test, with the ribber set at half-pitch. There is an obvious difference in rib before the slip stitch segment and the every-needle, EN, segment. 
Using a very thin space-dyed cotton, tension 4 on both beds, produced a soft, drapey, squishy fabric. Because once the cast on and needle set up are complete, the knitting occurs on every other needle in alternate directions, full-pitch is used without any problem even though every needle is in use on both beds. 
after only the top bed knits with slip stitch floats behind every skipped stitch, which, in turn, will appear longer on the knit face
Ponte di Roma
the modified version of the repeat used in my test swatch
Pairs of rows of interlock are followed by pairs of circular rows. The repeat was changed and the second color was used to define the difference in stitch formation between the two groups, color changes were made every four rows. 
Using a single pair of carriages and having to constantly change cam settings appears far too complicated to manage for producing any length of fabric. My hack for making things easier and faster was to resort to knitting with 4 carriages. One pair was set for interlock operating from the left, the other set so as to achieve tubular knit. The ribber carriage is set to knit from right to left and to slip from left to right when the main bed knits. The knit carriage is set to slip in both directions so that the pattern selection remains continuous. It will slip all stitches while moving to the left on rows with no needle preselection, knit on all preselected needles on its return to the right.
My initial test was to knit twice as wide a repeat, thinking it might be easier to identify stitch formation, working on 4 X an odd number of stitches, in this case, 36, beginning and ending with knitting on the top bed, 
single row pockets are formed. 
With both part buttons pushed in, in every needle rib, the needles in D will knit, the ones in B get skipped completely. As the carriage passes to the right, the needles for the next row are preselected, come forward holding only the skipped stitches in the row just knit, while the alternate needles are set up to be skipped on the next pass to the left 
After the two passes one full row of knitting is completed on the patterning bed. The needle preselection is now on the next row of the repeat, in this case, row 1.
The process repeats throughout. Two passes complete a single row of knitting. The carriage actions for completed passes 1 and 2 
The close-up begins to make the tuck loops a bit more visible. The swatch was knit at half-pitch. Using full pitch in patterns that allow for it, diminishes the appearance of “ladders” between rows of stitches, indicated by cyan arrows. 
Manually the tucking lever position can be changed from R to P for a single row or more of tucks and then returned to the R placement. Switching between the 2 stitch types on the ribber is far easier and less complicated than changing cam buttons on the main bed, however, there are no rows where only tucked loops are created without knit stitches beside each of them if the lili buttons are in use. Using the slip-tuck Brother selection on the main bed allows for slip and tuck settings to be used at the same time in the same row as long as opposite cam buttons are in use. Some of the old punchcard books include patterns referred to as “lace-like” using the setting. The starting side makes a difference in results as to whether one leads with a slip stitch float or a tuck loop in the actual knitting. Swatches in this post have begun with the first needle in work on the ribber and a forward, to be knit, stitch selection on the first needle on the top bed, with an even number in work on both beds. The end needle selection is canceled unless stated otherwise. The cam setting used for these tests 
After the return pass to the left, the tuck loops are evident on top of the needles preselected forward to knitting position on the next pass to the right. Two carriage passes fill one row of knitting, here ending COL. 
On the ribber, as seen in the thumbnails, one may choose slip or tuck in one direction alternating with an all knit row in the other, or tuck or slip in both directions in an alternating pattern EON_EOR, on an even number of needles with the addition of lili buttons. The appearance on the ribber of the formation of the loops echoes that seen on the main bed changing needle placement unless the carriage is set to tuck in only one direction. 
A test can quickly be made to observe the stitch formation with the main bed set to knit every row. I chose the left tucking lever down to R when the carriages were on the left, up to P when the carriages reached the right. The tuck stitches with no further action line up over the knit stitches in the previous row 
Experiments outside the interlock family: the two-by-two repeat is supplied in most packets of cards that come with the purchase of a punchcard machine. At one point in time, Kate Armitage published a book containing 104 variations of knitting using the card, both single bed and combined with ribber use. There was also the equivalent for card #3, the point being variations are limited only by time and imagination. 
If the intent is to have solid vertical columns of color, those areas as in column marked A, need to be adjusted for using alternating colors as well. Using the repeat on the left of the chart 
Half Milano stitch formation on the left, full on the right 
In executing the fabric as a hand technique, the main bed is still set to slip in both directions, the ribber to knit every row
Begin with simple shapes, examining the quality of increases and decreases, whether single or multiple, eyelet formation. Picking up from the row below before the next pass with the contrast color eliminates eyelets
picking up from row below at any point during knitting 
Simple increases or decreases are made by moving stitches laterally in either or both directions. Increases may be made by moving contrast color stitches laterally, followed by the choice as to whether to fill in the empty needle or allow it to create an eyelet. 
Fully fashioned increases or decreases are made by moving a stitch or a group of them to the adjacent needle/s to the left or the right and then taking the double stitches back to the original position, leaving a single empty needle for the planned eyelet formation. There should not be multiple needles with no stitches on them unless the goal is to expose a stripe of ground 
typically, in these exercises, there should be single empty needles after transfers, making certain proper needle selection for the pattern group is maintained 
Combining eyelets with lateral increases
When transferring stitches, watch for any loops getting caught on gate pegs, as seen on the left below, increases and decreases may be pre-formed on more than single stitches
At the top of the single medallion, the stitches were transferred to the ribber prior to knitting with the same color once there was no needle preselection for it on the top bed.
The first preselection row after the chosen cast on is from right to left with end needle selection canceled and the knit carriage already set to slip in both directions, with all required needles on the top bed in the B position.
To eliminate the extra white rows, the main bed stays set to slip every row, the ribber settings alternate. It is set to slip for two rows immediately after knitting with the red yarn, then will be reset and knits for 4 consecutive rows. Transfers to create eyelets are made on selected needles on each of those two rows, always toward the carriage, even as the transfers themselves change directions as the angles of the shape decrease toward its center on the top half of the design. After the first transfer and the carriages travel to the right, a long float will be evident, will “disappear” on the return to the left. 
Patterning selection will reappear as the carriages return to the left. 
As with single bed lace, the first pass after transfers creates loops on empty needles, which here need to be kept in upper work, D position after transfers. For non Brother knitters, Brother positions are A, B, D, E, skipping C. 
Knitting over the loops on the next pass on that bed completes the stitch. This design is knit as continuous, the striping at the bottom is wrong because the red was not picked up after the first 2 rows knit in pattern with white, most sequences for the remaining fabric are 4 passes with white in the feeder, followed by 2 in the red. All eyelets here are reduced in size by picking up from the row below, all transfers for decreases are made laterally, the border is set to a width of 4 stitches, the pivot point for the repeat has been narrowed 
the differences at the edges of the shapes. 
Many of the same principles may be applied to designs using tuck stitch settings, where the striping will appear vertically rather than horizontally
Here the fold is created by 2 stitches tucking for 2 consecutive, then knitting on the same needles for 2 rows on regularly spaced pairs of needles on either bed. Most knitting is on a single bed. A lacey series of eyelets begin to appear, and in some random racking at the top of the swatch, the possibility of developing a secondary pattern due to the combination of racking and tucking begins to show. The middle image is of the fabric slightly stretched. Passap 
Brother: the ribber can do the stocking stitch background, every needle in work, carriage set to knit. The setup is the same as the Passap diagram. A repeat with 2 black rows of squares followed by 2 white can be programmed on the top bed. On every needle selected rows, pairs of needles will knit, on the white, no selection rows the same pairs of needles will tuck for 2 rows. 
Moving away from vertical ribs becomes significantly easier if one has a G carriage. The alternative option is to create geometric folds that require transferring between beds. Any of these fabrics are best knit in a yarn that has memory and can spring back. Yarns such as acrylic can be permanently flattened by pressing, resulting in loss of texture. A quick experiment: black cells represent knit stitches, blue purl ones 
The needle setups: after casting on, transfer for a stitch configuration based in this case, of blocks that are 5 stitches wide. A single needle on the opposite bed is used on each outside edge of all needles in work. 
The repeat, 10 stitches by 40 rows. 
More on 
Pleated one color “shadow lace” in 
The extra needle selection prior to the next all knit row helps track the direction of the moves, stitches are moved three at a time, there are no cable crossings
The next step for me was to explore cable crossings on elongated stitches working double-bed. A basic pattern on any programmable machine for playing with elongated stitches on one bed while knitting every stitch on the other is to program pairs of blank rows followed by solid punched or black pixel rows. The yellow line in this chart illustrates the row on which cabling might occur. 
Programming the width of the needle bed allows for only the stitches forming vertical columns in chosen locations to be put into work, allowing one to place groups that will involve crossings anywhere on the chosen pattern width. A base is knit in the ground color, which slips for 2 rows on the main bed, creating the elongated stitches that will be cabled. I had no problem with 2X2 cables, 
Cabling, returned to in a later post, with adjustments, making things work.
Continuing with shapes on striped grounds, this is the result of a self-drawn pattern 
The approach is different than in the blog post on 
Punchcards, in theory, may be used as given and set to double length, while for use in electronics drawing the pattern single height and using the double-length setting is also an option. Starting sides and fixing errors have always been more confusing for me when using the double-length feature, I prefer to punch holes or program pixels as I intend to knit them. The isolated reduced repeat for use in the electronic is charted, with an initial one-pixel error in 2 consecutive rows, marked with red cells. In transcribing any design, it is worth checking repeats multiple times after eyeballs and brains have had a rest. This was my start:
The design process using Numbers before exporting the repeat to Gimp for reduction to B/W png: in this approach, the repeat is drawn double height to start with. The red cells represent stitches that will be transferred down to the ribber before knitting the next row in the pattern color from left to right. 
The first test is of an isolated motif. The yellow arrow points to the pixel error, the cyan to the positions where some needles in the full repeat were “accidentally” placed in A position, not B, resulting in pattern stitches not being formed. 
Two other options for charting the fabric in numbers: A. draw the repeat as given
Any simple Fair Isle repeat may also be used. The numbering in the charts matches what is normally seen on the left edge of the tables
The world of possibilities grows even further for single color shadow lace, when, examining the same design, one recognizes that the pile knit card, with the blank rows filled in in pattern, is the same as the fair isle version of the repeat, rendered double long 
Some authors have suggested plaiting as an alternative to creating shapes with true brioche, which can be complex. To my mind, plaiting falls in the beauty being in the eye of the beholder category, I prefer far crisper color distinction in my knits. This sample from the previous shadow lace post uses thick and thin yarns 
its accompanying test swatch 
the file for multiple repeats after color reverse 
There is an interesting scale and depth of fold comparison between this version and the first using the repeat, achieved by tightening the tension as much as possible, and possibly by reducing the size of the eyelets. 
Take care if copying and pasting single columns to alter a repeat width that the whole column is indeed copied and that if using the pencil tool flood fill is not used unintentionally. The original intent was also to correct the elongated slip stitch segments on the edge of the programmed vertical designs marked in blue, but the paste with errors in red accomplished creating the same issue 
The first segments were knit using striper backing, with the ribber knitting every stitch, every row, in both colors. When a slip stitch is used with needles out of work on the main bed, end needle selection should be canceled. In A it was not. The result is that end needles alongside the out-of-work column knit with each color in each row. In B, end needle selection was canceled, and one can now see the elongated slipped stitches that result from areas that should have been marked with the contrasting color 
As long as the number of stitches on the ribber is even, lili buttons may be used, affecting the scale of the pattern in both height and width. In A, they were used with the ribber set to slip in both directions, in B, set to tuck in both directions. C marks the return to the N/N setting, with needle transfers to mark a possible pleat. 
The initial pleat idea charted out for single stitch folds, stitches transferred to ribber in the R columns, to the top bed in the T columns 
The result is a fairly soft pleat, the choice below was to retain end needle selection. 
Various ribbed pleat configurations are explored in 
Paired transfers in the planning stages: because the repeat is small and has a single center pivot point, it is rendered once more, adding columns
Here the transfers planned to opposite beds are marked on a 48 stitch repeat with red cells.
The resulting fabric relaxed on the left, lightly steamed on the right 
Note: the color positions in the design have been reversed from those in the first swatch. If “floats” are noted at any time in the spaces where needles are out of work on the ribber, look for dropped stitches.
Needles in locations where only the backing is to be shown are transferred down to the ribber. Leaving the eyelets, they were transferred back up to the main bed when brought into work to reverse or change the shape. Addition and subtraction of stitches take place before the next pass with the alternate color. Here movement is random, to get some sense of the effect, it could be made deliberate by following a chart or color separating and automating the pattern, with its starting side on the right. 
What of having shapes appearing in each of the 2 colors on a striped ground? Eliminating some of the guesswork I used the repeat from a previous single-bed blog post on 
The resulting sample, the yarn is thin, might have benefited from tighter tension and more contrast. 
In terms of positioning the carriage, a wire that is akin to that found on Passap strippers is on its underneath. In positioning the carriage on the beds, check visually that it is indeed lying between the gate pegs of both beds prior to attempting to travel with it to the opposite side 
For my test I used EON needles on the ribber, planned alternating selection for each new transfer. This could be done by selecting dashes and blank spots on needle tape ie. dash in the above photo, blank spaces below 
My first attempt at creating shapes includes a band at the bottom where the EON transfers as above were made, but every row. Simply bringing needles into work on the opposite bed creates an eyelet. They can be eliminated by sharing stitch “bumps” on the opposite bed, but for the moment they are a design feature. The texture created appears in the areas involved on both sides of the knit 
It is possible to transfer single needles at sides of shapes ie 
or whole rows, but the change lever needs to be set to position accordingly.
The enclosed punchcards: 
Transfers of stitch groups, whether by hand or using the accessories are made on rows where no needle preselection occurs on the main bed 
This series is a proof of concept for my approach to developing electronic cues
The original repeats were modified to include 2 blank rows between segments that allow for transfers between beds not hampered by needle preselection on the top bed. The motifs are color reversed, but not the blank rows between them. 
The knit carriage is set to select needles KC I or II, end needle selection does not matter. All needles on the top bed knit every stitch, every row, whether or not those design rows contain black pixels. No cam buttons are pushed in. Blank areas between black ones indicate the number of needles that actually need to pick up loops on the ribber to create shapes, filling in spaces between selected needles until an all-blank row is reached for making transfers. The chart on the far right illustrates a shape where the easiest method becomes one where stitches on the ribber are manually transferred to the top bed in order to reverse the shape and maintain every row preselection. The selected needle corresponding to the black square marked with the top of the red arrows is pushed back, the ribber stitch below is transferred onto it, the needle with the couples stitches is brought to E position, moving across the bed in proper locations prior to knitting the next row. 
In this repeat, the side vertical panels of ribbed stitches are added. The knit stitches on each side of them roll nicely to the purl side, creating what in some fabrics can actually be planned as an edging. 
My takeaway is to test the accessory with some patience, sort out the sweet spot for the ribber needles in relation to main bed ones in terms of handling transfers and yarn thickness, use colors that allow for easy recognition of proper stitch formation, keep good notes, and “go for it”.
In the first sample, equal thickness yarns were used, the colored yarn was a rayon slub with no stretch and slippery nature. 
Cast on for EN or EON rib. Transfer all the main bed stitches down to the ribber. Extra stitches can be cast on and transferred in addition to the planned width of the repeats to create a border on either side of the designs. During patterning there will be stitches in work on both beds at intervals, so the pitch needs to be set to H while knitting. When the top of the piece is reached, transfer all ribber stitches to the main bed and bind off.
If any stitches are pushed all the way back or in mixed alignment during transfers, 
A closer look 
The same repeat may be used to produce a tucked version. In many punchcard machines, a card is supplied with a 2X2 check. With the main color, in a suitable yarn, cast on for 1X1 rib. Set the knit carriage to tuck and the ribber carriage toknit. Knit 2 rows with the contrast color, followed by 2 rows with the main color, repeating for the desired length of the rib. Knit the last row in the main color with both carriages set to knit. Transfer the ribber stitches to the main bed to continue knitting single bed.
with my color changer in this threading sequence throughout
where normally each color in each design row knits twice. Because selection occurs for pairs of rows, the first preselection row is from right to left. To decrease the backing rows, the ribber is set for birdseye. I prefer to have an end needle on each end on the ribber, keeping in mind that the total number of needles in use there needs to be even. The machine provides reminders as to which color should be knitting. My samples are knit using KCI on the top bed. Because the preselection happens twice, it is easy enough to knit in pattern from left to right, 
when the carriages have reached the right side, simply use a ribber comb to push all needles back to B. 
Speeding things up with color separation, beginning with the method that will have each color, each design row knitting twice. The repeat is 10 rows high, so it is expanded X6 to 10 by 60 rows. In the final result, the second row for each color in the separation is in turn erased. The red was added to make all 3 colors visible while working the separation, avoiding confusion with the white ground. The knittable result as usual is in a black and white png 
The img2track settings are for now for 2 color knitting, the prompts for the color changes are lost.
The ribber can also be set to knit every row, resulting in elongation on the knit side, while creating an interesting striper backing
Comparing this version to the birdseye backed one for repeat height 
Comparisons: HoP, pushing back needles to B, and color separation results. In the latter, the design is likely elongated in part due to a change in the distribution of thinner yarns to larger design areas with no tension adjustments 
The Ayab lettering as opposed to numbers move from right to left. The manual states that the color separation order is: white C, grey B, black A with their sequence = C (3), B (2), A(1). If the prompts for changing colors as given are followed it provides a very valuable in tracking them, but if out of habit one knits in the usual 1,2,3 sequence, the color placement occurs in an unexpected order and may result in errors. The on-screen letter prompt corresponding to the anticipated color change sometimes occurs with the knit carriage on the right, sometimes as it approaches the changer, and the size of the font was hard for me to see since the screen was not close enough for easy visibility.
If the png is copied from the post it is likely to appear in RGB mode and it will require conversion to 3 color bitmapped. Its grey color map entry is seen below 
The small file makes for a quick test of proper color selection for each of the three colors used 
Having some idea of stitch counts for each color in the design in the first few rows can help identify proper, planned color placement errors 
My first swatch using the heart of Pluto separation and a greyscale motif 
My tested color change sequence is #1, #2, #3 colors throughout, I disregarded the prompts for color changes at the bottom of the Ayab screen. Some things to ponder: in pieces that require color changes, starting with waste knitting in the same colors can help assess the best tension, whether each color will be picked up properly, and if the colors work well together. Looking at these 3 small tests, it appears that a choice should be made when casting on about using color 1 or 2 for the preselection and cast on rows. If the setting to slip is forgotten for the first move to the left, the color in the feeder will knit every stitch rather than a pattern selection. Always check settings when on the right, making certain lili buttons are set as well. This pattern does not contain 3 colors on every row. In addition to that, when working DBJ with other color separations one is likely used to seeing knit bed needle selections on every row. That is not true here, is a function of the technique, not a patterning error. On rows that have colors missing, when that color is in use, the main bed slips, the ribber works every other needle, first in one direction, then the other, adding to the row count on the purl side of the knit. 
The mess at the bottom was due to the green yarn getting caught on the needle bed and not knitting the necessary stitches on the ribber, so dropped stitches were formed 
The assumption is that if the C, B, A rotation and prompts are to be followed, the middle color 2 can stay in place, and the placement of 1 and 3 can be exchanged.
