DBJ: more than 2 colors per row 2

My first attempt to use a 3 color automatic separation was with img2track. There are 2 oddities for me using the program. One is that the default selector setting is for a single motif, perhaps on the assumption that the main use would primarily be for large-scale, nonrepetitive images. The other is that even if working with 2-row color changes with each color in each design row knitting for 2 rows, the first preselection row must be made from the color changer side. This is a necessity in Japanese color separations for 2 color work where the first color knits for only one row on moving from the right side to the left, but if 2 colors with the same selection are required at the start, one row of the 2 appears to be technically eliminated as a result. I had expected to use 3 shades of grey, a pleasant surprise: the program can actually import  3 colors other than grays. The test was using the 11 stitch repeat A pleasant surprise: the program can actually import a 3 color other than greys image. Using Gimp, images worked in RGB can be reduced to indexed 3 colors for this purpose. If there are rows where there is no color represented, then as explained later in the post, the indexing should be to 4 colors, not 3 for the planned for import to work properly. The design is automatically flipped vertically, so it will appear as intended on the knit side of the fabric. Yarn colors may be placed in the color changer matching the order in the assigned color numbers to match the placements in the original image. The program automatically adjusts for the vertical stretch, which changes the aspect ratio of the shapes. On my 930 I received prompts on which color to change to prior to doing so, eliminating any confusion. With no such prompts generally one can tell which color was knit last because it will appear on top of the previously used one on the left of the knit. With a stretch factor of 1 selected in img2track, the image height was reduced by half. To achieve a look closer to the intended shape,  the repeat needs to be rendered twice as long, or the stretch factor can be adjusted in the program itself to 2.0

In the past, I have preferred to elongate the design prior to importing with plans for download rather than to rely on memory for changing settings either in the download program or in the machine itself in future uses of the same design. The same yarns, tension, the total number of carriage passes, and settings were used showing the difference in aspect ratio between single color per row knitting and the img2track built-in color separation. The width of both swatches is essentially identical.
Images may be loaded into the program without the cable being connected to the machine. Error messages do appear with download attempts if the cable is not properly in place. The machine also needs to be powered on before the program is launched if a download is planned. I am working in Mac OS Mojave 10.14.6, have no present desire to upgrade to Catalina.
Executing fabrics that will knit each color in each row only once for every 2 passes on the main bed: back to that original repeat Here it is used as drawn, note vertical stretch set at 2, will be cut in half by the software, getting the image back to the original height l0 X 3 = 30 rows required for all colors to knit in turn; there will be 60 carriage passes to complete one single repeat. The 930 will provide prompts for the next color to be selected by pushing the matching number button on the color changer, avoiding any confusion in terms of what should be picked up next. img2track will also flip the design horizontally automatically so the image will appear as originally drawn on the knit side. Images are loaded as single motifs, so the change in the selector needs to be made manually for an all-over pattern.

In order to have each color in each row knit only once: cast on with preferred method and color. Set up the machine so that the yarn colors are placed in the color shown in the image presented by the program to match your design. End with the machine on the left-hand side for the first preselection row.
COL: set both carriages to slip in both directions and all its needles in B position, set lili buttons if not already in use, pick up color 1
knit one row to the right
COR: knit to left, color 1 will knit for a single row
**COL: change color, STOP! Push back any needles on the top bed back to B position, as you knit to the right the ribber only will be knitting, knit one row to the right
COR: preselected needles will knit in the color last picked up on the way back to the color changer, knitting only one row in that color on the main bed as you return to the left**
repeat the last ** 2 steps throughout. One can get into a rhythm. A cast-on comb, part of a garter bar, or any tool of adequate width can make it a quick process of pushing needles back to B when needed to
trust the software, not the selection expected by your eyes.
My swatch seemed to be growing in length at a faster rate than I remembered in the last exercise, here the results are shown side by side with the fabric executed previously   Obviously a success in terms of the single row for each color reducing elongation of the design shape.  While knitting occurs using the same yarns, at the same tensions, there is a clear difference in the length of each stitch on the main bed and their appearance. The reverse. Checking the ribber carriage I noticed on the left side it was set to knit only, not to slip: OOPS! N is king, so the ribber set as shown is knitting every other needle when moving to the right, but even with lili buttons in use, it knits on every needle when moving back to the left. Every other needle on the ribber will then be knitting for 2 rows as a result. The more knitting on the ribber for each pair of rows, the longer the stitches on the opposite bed. The backing is an interesting variation (half) birdseye. The elongated stitches on the main bed show more of the backing in between their shapes, it is referred to as bleedthrough. In some instances, the result can make the knit surface resemble weaving and its appearance far less familiar in a surprising, pleasant way. Beauty is in the eye of the beholder.
Remember always to double-check all settings at the start of any process in case something was missed or magically moved, and keep notes.

I was asked on Facebook whether the technique shown here is the same as the 3 color slip (skip) stitch patterns in the Stitchworld pattern book, my response: some of the stitch world patterns for 3 color slip are designed for or may be used on the double bed (often single bed as well). For instance, patterns 392 and 394. The original designs are rendered color separated. Numbers on the left suggest the order for color changes to achieve a look similar to that in the swatch photos. Assuming the pattern is not easily accessible because of its being built in your machine’s memory, it would have to be entered into a paint program manually to make it available for import and in turn for download. Note that the color-changing sequence may change over the course of knitting the image, so prompts or notes of some sort would be needed to keep it correct. Each color in each row knits twice after each color change. My goals in my blog posts so far have been to keep a constant color-changing sequence with each color in each row of any personal design knitting only once. I did it first with my own color separation, then, in turn, used the 3 color separation feature in img2track to achieve my desired result. Here are 2 images from the Stitchworld section in question The images could be replicated as given in a paint program, using only one color for the squares, but “should be reduced to black and white”. Attempting to import an indexed 2 color image drawn in a color other than BW may result in strange results. That said, if glitched knits are the goal, the above could work just fine. Curiously, here is the same process, using a different color, and a successful import. Checking again, I had forgotten to save the image after indexing it from RGB mode to 2 colors. and a test with another color 
Drop stitch lace periodically comes up, I have written several posts on the technique, now considered the possibility of producing it in 3 colors. The resulting fabric tends to be long, thin, and in need of blocking. There is no way to avoid the striped ground. Passap knitters often refer to this type of knitting as “summer fair isle”. I adjusted the repeat widthNote to self: if you are determined to use a punchcard carriage on your electronic machine remember there is no KCII to cancel end needle selection!
In order to knit this fabric, stitches must be cast on in whatever method you prefer, but prior to any patterning all stitches must be transferred down to the ribber, and the main bed needles are placed in the work position but are empty. Because of the single row knit for the first 2 design rows of color one, the start is a bit finicky. Alternative start follows lower in post
COL: KC II, main bed set to slip <– –>, ribber stays on N throughout, no lili buttons. Knit one row with color 1 (ribber only)
COR: as you knit to return to the left, the color 1 preselected needles will knit, while the ones corresponding to color 2 will preselect
COL: carefully drop any stitches on needles with color 1 on them without disturbing the new needle selection for color 2
**COL: change color, knit to right
COR: drop the stitches knit in the new color, make certain all needles are empty and in B position, knit back to left and needles will be preselected for the next color**
Repeat the ** to**
Again, the resulting fabric is narrow and long, it may take a bit of squinting to recognize the design.
If each color is not represented in each design row, the ribber only will knit with no selection on the top bed corresponding to those no pixel rows.  A single repeat results in about 3.5 by 9 inches of knitting, a far cry from what might have resulted in a single row per color design row dbj virtually shown here in a tiled format

The question then comes up re dropping only one of the 3 colors. Using the manual selection described above, the knitting on the top bed would need to be canceled on every color, every row except for the single row in the chosen drop stitch color. To my mind, that is too much to keep track of for any length of time. It would be easier achieved with specific self-drawn color separations.
There is a lot of testing that can go into developing any fabric in unfamiliar techniques that may or may not meet our expectations or our “like”, it all contributes to learning regardless of whether the tests evolve into projects.

My last post on drop stitch lace in from single color to two: revisiting the techniques on brother machines

2022: returning to the topic, exploring drop stitch lace in multiple colors 

Reviewing what happens within the program one more time, highlighting significant items to verify before beginning to knit Getting that first row to knit twice instead of a single time if that matters in your technique or is your preference:
advance design row to last one in the sequence, in this case, 30
in the 930 when the pattern is loaded, using the down arrow key gets to the last row in the downloaded pattern more quickly

COL: set machine to preselect pattern, the next 2 passes would need to not knit, so both carriages are set to slip in both directions
knit one row to the right
COR: number 30 design row flashes
knit one more free pass to the left, row 1 of color 1 will preselect as you knit
COL: number 1
design row flashes
check all settings,
main bed set to slip <– –>, ribber set to slip <– –> with lili buttons, and with an even number of needles in work for standard dbj pick up color 1 in yarn changer
the resulting knit design will be elongated
COL
: knit one row to the right.
As you knit the first row of color 1 design row 1 will knit, row 2 of color one design row 1 will preselect
COR: knit back to the left
as you knit back to left row 2 of color 1 design row 1 will knit, row 1 of color 2 design row 1 will preselect
COL: pick up color 2,
continue in pattern changing colors every 2 rows

In designing your own, small repeats can easily be rendered even in a simple paint program. Larger, more complex ones, may best be worked using layers and masking in a variety of programs ie. Photoshop or Gimp. It is possible to combine 2 color dbj and 3 color dbj in the same piece simply by using different color separations for each segment, but the look of the fabric in terms of the length of each stitch on the knit side may be quite different when moving from one segment to the next. Of course, the backing will change as well. Though the aspect ratio of the design changes in terms of height when one knits 2 rows for each color in each design row, it remains the easiest separation method. Adding the hand selection as described above so that the main bed knits in one direction only helps reduce some of the extra height, making the original image more recognizable. There is no option within img2track to perform the action automatically in terms of adding the necessary blank rows to replace some of the colored ones in the separation.

Passap knitters are not left out of this idea. Passap preselects pushers. On the E 6000 Tech 179 emulates the built-in KRC option in Japanese model machines. My guess was that technique 180 for 2 colors, perhaps 197 with both arrow keys on the back may work for 3 colors. TBD in my next spurt of knitting activity on it. That said, the console’s built-in designs may be used cross-brand. It is possible in Passap to layer repeats in order to add the third color to the mix. Here is one sample reworked for use on Brother and downloaded into img2track. If black and white repeats are already in your library, one may easily recycle them adding a third color. Here I did so with a repeat intended for a very different topic in future posts. The image was altered and tiled in Gimp for a repeat alignment test and is also shown imported into img2track for possible knitting. What about using images published in punchcard machine references? It is probably best to start with smaller repeats. That said, this is from a Deco pattern book.  Deco punchcards were 40 stitches wide, could be joined together in length as can those for the Japanese models The image of the separation on the above right has not been proofed for accuracy. If it were, the next step would be to elongate it X 2 for color changes every 2 rows. One method is to elongate the original in a paint or photo processing program. The width is fixed (40), the height is scaled X 2. The resulting BW indexed image may be imported, using a 1.0 stretch factor, it remains unchanged. When I tried to elongate the unstretched image in img2track by 2.0 my first try failed. It turned out the reason was I had saved the import without first indexing it to 2 colors. With that corrected, the result matched the one from scaling X 2 in height in the paint program
For years now I have been doing color separations which at first could be extremely slow, one pixel at a time. With increasing familiarity with Gimp and Mac Numbers over time, I have been able to decrease the speed to achieve them immensely. So here, with img2track I now have a program that can work with and separate multiple colors at a time (up to 6 in its pull-down menu option). Here I returned to my first separation in 3 colors for the now-familiar repeat A, elongated it in the paint program B and imported A into img2track choosing 2.0 stretch and 3 colors C. I can totally live with the fact the colors are not the same. The color changer can be set up with my chosen colors in any order I choose

In my first tests with charting a repeat missing any color in some rows, I had scaled the original image taken from a spreadsheet to the wrong size in Gimp, so operator error resulted in crazy results in img2track. Here the image is scaled properly for each color represented for a single row in height, and also scaled again for double-height for possible knitting in Gimp. The Gimp scaling failed to be accurate for me (second image from left) until I indexed the original to 4 colors as well instead of 3. The no color rows as we view them actually serve as a fourth color in the separations. Importing the proper size png into img2track for separation of 4 colors per row now gives results that make sense: note the daunting estimate for the total number of carriage passes for a single repeat height
If the ribber has knit on every needle by its return to the color changer and the machine is set to slip both ways with no needle selection on the main bed, the “no color” can be executed as an empty yarn holder in the color changer combined with no yarn in the feeder.  The rows involved should simply not knit on the top bed, with no dropping of any of its stitches since no needles will have been selected thus coming forward with the yarn in the hooks and traveling behind the latches and in turn, slipping off the needles as a carriage with no yarn pushes the needles with now empty hooks back to B position.

In testing concepts, I prefer to work initially with small repeats. Punchcard books can be an excellent source, but the Stitchworld books have the advantage of actually showing a single repeat for each design, so an 8 stitch repeat, for example, would be shown as such and it would not be up to the person using it shown in a 24 stitch card one to isolate it. I randomly chose pattern 394. I realize that if used from a built-in library of patterns in any machine model the prompts may be provided by the machine, but here I am looking at simply duplicating the pattern for import into img2track.

After duplicating the repeat and associated numbering using Numbers, these were some of my results. A pleasing surprise was that colors in the 2 color import need not be only black and white. That said, the pale green failed, the red did not, and the results from importing both the red/white and the black/white were identical. Another future time saver for me to keep in mind. 

If colors were assigned to each pair rows representing each color used, the white squares were read as an added color on import, and the only way to have those few blocks for the 4th color to show up in the visual representation in an imported repeat was to assign a 5 color separation, resulting in a scrambled pattern. Removing the few squares in color 4 will produce inaccurate separations for actual knitting of the given pattern whether imported to be rendered for 3 colors or 4. The black and white on the right is the “correct” match, leaving any prompts for use of shifting color changers to be tracked somehow by the knitter. Punchcard knitters may have the easiest knitting variable color sequences since cards may be visually marked up with colored pencils matching needed change locations and taking into account your eyes are a number of rows above the row being read by the card reader. This number depends on the machine’s brand and model.

Some spreadsheets that may help with tracking row color changes or other regular actions on paper: 2018/04/tracking-rows-1.pdf

I have been asked about the position of the slide lever being fixed in my dbj illustrations. I have found the alternate positions can be wild cards, would rather make adjustments in carriage settings if needed rather than let the factory settings do some of the work for me. If the settings are accidentally in the wrong place as multiple pieces are knit, or in knitting ribber bands for garment pieces, the gauge is changed if the alternate setting is not adjusted and kept constant, and that may not be noticed until the piece of knitting in question is completed, needing a restart. With 15-20 machines active in each studio class and lab session, out of habit, I tried to reduce as many variables as possible.

A video resource on using img2knit for 3 colors per row knitting with blank color rows in the design by Tanya Cunningham, and a Ravelry thread on the topic

 

DBJ: more than 2 colors per row 1

I began to machine knit in the glory days of using the tool and developing techniques. There were internationally attended seminars, hosts of wonderful teachers, and a community of potential peers brought together physically in pre-internet and online forum days. Punchcard machines were the standard. As more electronic models became available it was not uncommon to find the rep for DAK at one end of the market spaces at seminars, and Cochenille at the opposite end, promoting early interface and design programs. My choice at that point was the Cochenille products. I attended design school beginning 1989, the computer lab at the time was all Amiga, and I never really became interested in DAK or for that matter, Garment Styler. I was up and running soon with Bitknitter for the Passap, had the option of the Forma on the Passap or the Knit leader on Brother for less traditional shapes, and the magic formula worked for any of the math I needed in days of using graph paper or that other paper that has lots of lines in it for keeping notes. The calcuknit and styler at the time worked on established default curves, the same that could be achieved by charting math or even simply connecting lines between dots on graph paper, drawing lines in between the dots, and following the drawn lines guesstimating increases and decreases along the edge of the drawing. I have never had any experience with DAK, was never really interested in owning it.  😉
A bit of early low tech: when aspect ratio mattered pages of acetate sheets with vertical lines spaced at different intervals could be lined up atop of each other and placed on a design to approximate knit gauge, and the result could then be charted out and planned for a punchcard repeat or intarsia. Some acetates were sold gridded to approximate stitch and row default proportions, there were also published paper charting aids for both standard and bulky gauges. Some examples:

Laying acetate or even plastic wrap on the computer screen itself and outlining designs and processing them with overlays, was a version of “poor man’s digitizing”. The technological advancement of black and white security cameras used in combination with RGB filters made the first digitizing of non-repetitive images possible.
I had never really planned on teaching. As it turned out I interviewed a few weeks before completing a textile program, had several finished pieces with me from my undergrad work, and was amazingly hired instantly as the machine knitting instructor for apparel and textiles. I found myself faced with a Brother punchcard lab and a summer of crash learning on the specific models followed. In order to teach effectively, I learned to analyze and break down stitch types and techniques in order to help others understand them. Summers would often include studying trends and knitting lots of swatches. Many of those are still in a couple of bins I held onto when I downsized a few years ago, giving up my studio space. It is that analysis that keeps me interested still in working out how-tos. Although I use electronics, I still like to break things down when I can so punchcard knitters are not left out.
Getting to the topic of multiple colors and nonrepetitive images in the “early days”: punching holes in cards is time-consuming and without the aid of electronics or specific software, color separations for dbj and assorted double bed fabrics need to be worked out manually before punching, and accuracy is important. Consequently, some knitters found compromises to achieve the look of multiple colors per row. Julie Schafler Dale’s Julie-Artisans Gallery on Madison Avenue became the central plug-in point for the wearable art movement, eventually representing artists, some of whom are no longer knitting, and some may still be found presently online
Susanna Lewis, Norma MinkowitzJean Williams Cacicedo, Linda Mendelson,
Nicky Hitz Edson, and not caring about floats, often working single bed with some linings and over-dyeing additional colors  Janet Lipkin .
The non-repetitive image was worked out in length by dividing the image lengthwise into panels often 40 stitches wide for use with early Passaps and continuing on to the desired height.  Widthwise, the design was split into the maximum width of 40 stitches with added seaming allowance and consecutive panels were joined to compose the large image. The kimono was the most common shape because of its simplicity in components. As for getting the look of more colors per row: multiple thin strands in color families were used in each yarn feeder. For example, one can knit with  4 strands of red shades, 4 yellow sequentially lined up, used in pairs in the same feeder. Keeping the red constant, for the second strand, the work may begin with either the lightest or darkest yellow, knit X rows, switch yellow to the second shade for X rows, switch to each of the next hues, and in theory, the final set of rows should look like a different color than the one used in that feeder at the start of the process. the ratio of red could be played with as well, expanding the range. Knitting pieces in black and white particularly allows for limited edition knitting. The third color may be applied on portions of the white, providing a look-alike for 3 colors per row. With the range of space-dyed yarns or custom dyeing blanks and then unraveling them for larger pieces to control pooling one can also get the look of more colors per row. Not to be forgotten are hand embroidery and fabric applique.

The question remains as to how to knit more than 2 colors per row on home knitting machines, and color separations to make the fabrics possible. The more colors per row, the more passes to complete any design motif. The final result may be thick and unyielding unless thin yarns are used. More rows are often knit in the back than on the front of the fabric, resulting in further elongation of the shapes involved. The easiest default separation is to knit 2 rows of each color for 2 rows sequentially, allowing paired passes to and from the color changer. This results in an elongated design, even with playing with standard backing variations to reduce vertical stretch such as in birdseye. In a post years ago I began to look at 3 colors per row and some potential issues with separations for dbj using it the initial method:

Each row will have to knit twice to get 2 and from the color changer for color changes every 2 rows, for less punching, each row can be represented once, but the result will need to be elongated X 2

The graph below shows the motif in repeat, the next column the color separation, with expanded rows, and in the third, the black indicates the knit stitches (black squares on mylar, punched holes in card) prior to elongation. 

The repeat was not workable when using the standard method used to avoid elongation, where each color for each design row knits only once. There are segments where the new color would not align properly on stitches knit on the previous row, and to get the design to work I had to amend it.
Fast forward 5 years. I now have access to both ayab and img2track, no longer use Photoshop and Excel, still use GIMP and Numbers as my way to color separate and produce charts for my blog. There are active forums both on Ravelry and Facebook with general machine knitting information, or specific to both interface options and methods for using them.

Sarah Spencer @ heartofpluto developed code that reduces the number of rows knit making the common impossibles possible. The code is freeware, has been implemented in Ayab and available there as a menu option. Adrienne Hunter’s posts on Ravelry explore the concept involved.
To begin, I was interested in using part of the concept on my own while keeping the color sequence constant ie 1,2,3,1,2,3,1,2,3 and so on throughout the piece and always begin tests with simple shapes that allow me to note what is happening to stitches, changes in scale as the result of the techniques used and started with designs where each color is represented at least once in each row. The color separation used: The ribber was set for birdseye at the start of the swatch pictured. At its top, N in one direction and slip in the other were used, with the ribber set for knitting every needle when the main bed slips them all on blank punchcard or pixel rows. The tension remained constant throughout. The KC preselection row can happen from either side with this particular color separation depending on whether one starts the design with a blank row or one containing pixels. Since 2 rows knit in each “color”, I prefer to start the first preselection row on the right. Both birdseye and N settings on the ribber produce a single row stripe on the back of the piece for every 2 passes of the carriages and completed  2-row color change sequence. The N setting contributes to elongated stitches on both faces, leading me to prefer the lili birdseye one. 

In theory, because of the way the birdseye rows set up a base for the next row of knitting,  that makes the stitches in any color workable in pattern colors. I created a new repeat for that motif working in Numbers following that up, I planned for the 60 rows needed to complete the repeat, hid 30 rows,  using this menu

copied and pasted the black and white repeat on the above right, then unhid the 30 hidden rows (see previous posts using command key selections), having a new color separation repeat, created the final 2 color BMP in Gimp (whew!)Since my separation is planned on 2-row sequences, I began with COR and my preselection row was from right to left. I initially had issues with the lili buttons behaving erratically, once the ribber carriage was oiled again that seemed to solve the problem for me. I did not arrange colors in the yarn changer to match the illustration in any way, but that could certainly be done to match a chart. With COL pick up the first color in the sequences, there will be no needles selected on the top bed (no black squares on the first row of repeat). Remember to have both carriages set to slip in both directions and the rule for even the number of needles in work on the ribber. As the carriages move to the right, every other needle will knit on the ribber, the first row with black squares will be preselected. With COR on the pass back to the left, the pattern squares will knit, a blank row will be preselected, and the ribber will complete a single row of the background color by knitting on the alternate needles from its previous pass. As a result, one row of backing and one design row for each color (including no color in some instances) are completed with every 2 passes of the carriages.
I had a better edge and needle selection on the ribber with an extra needle in work on its left as compared to the main bed, adjusting after the cast-on row, maintaining the ribber needles in pairs, and evaluating the swatches after the fact would probably set up an extra needle on the right side as well.

This swatch and the ones following it use yarns in 3 different weights and different fiber contents. The blue is acrylic and the thinnest of the three. Some bleed-through of the colors traveling behind the blue shapes, in particular, can be seen behind it. The magenta is cotton, the yellow wool. Better choices of both the fiber content and colors of the yarns used would improve the overall look. If one plans color changes to match the chart, a  paint program can help visualize some of the possible results

What of shapes with fewer than 3 colors in some rows? Again, there is some bleed-through, and even with this technique there is some elongation of each of the shapes, but less so than with other techniques. The backing of all these fabrics produces single lines in each color used Even fairly small repeats can take time to color separate in this manner. The technique, however, is the only method available to punchcard knitters. Electronic machines with download cables and varying software open a very different world in terms of ease and range in repeat size possibilities.
Here the repeat is adjusted from 11 stitches in width to 12, making it suitable for punchcard machines, and widening the center shape by one stitch may also reduce lengthwise stretch and help it appear rounder after knitting