I’ve been hard at work designing Everett’s stole. It’s proving to be a seriously difficult problem. Which is not a bad thing – I love complicated puzzles – but I also need to get this one done!

I finished weaving the samples a short while ago. Here’s what they looked like on the loom:

The structure is double weave. The black warp is on top and is doing all the interesting stuff; the white warp is on the bottom, and its main function is to stabilize whatever wefts aren’t being used on the top layer.

And here’s what that sample piece looked like when done:

This 15-inch sample took 8-9 hours to weave, because the density of the weft threads is insane. There are 253 picks per inch! (By way of comparison, a medium-weight handwoven fabric is usually about 20-30 picks per inch, so – scientists take note! – this cloth has a thread count that is a full order of magnitude higher.) The sample is also woven with four different colors of weft, which means weaving with four shuttles. That’s much slower than weaving with a single shuttle. Even though I’m an extraordinarily fast weaver, I can only weave about two inches of this cloth per hour. Which means that the five feet of fabric I’ll need for the stole should take about 30 hours to weave.

(Just for fun, I once timed myself weaving a warp of the same width, but in a medium-weight weaving yarn (10/2 cotton, about 3x the weight of sewing thread). I can weave 2.6 yards of that medium-weight plain weave fabric per hour, so in the time needed to weave five feet of fabric for the stole, I could crank out two hundred thirty-four feet(!) of the simpler fabric.)

But wait! It gets better. One of my discoveries from this sample is that four weft colors is not enough; I’m going to need five. That will mean (a) increasing the thread density to 300 picks (weft threads) per inch, and (b) using five shuttles, which will further slow my weaving speed. So I expect progress to slow to about 1.5 inches per hour, or 40 hours to weave five feet of cloth. (During which I could weave 104 yards of medium-weight fabric – enough to get you the entire length of a football field and into the end zone for a touchdown!)

But that’s the easy part. I don’t mind spending a lot of time weaving; I enjoy the process, and it’s fun to watch the cloth develop. The design challenges are what’s got me banging my head against the wall.

Let me explain. (It’s complicated, so bear with me.)

Jacquard weaving is basically “paint by numbers,” only with weave structures. (Weave structures are the pattern in which the threads interlace to form cloth.) To design my jacquard pieces, I start with an image, and I reduce it to a limited color palette. Then I move the image into Arahweave, my jacquard design software, and I assign each color a weave structure. Then Arahweave takes the image and replaces each pixel of a color with a unit of its assigned weave structure. It’s a little like doing counted cross stitch – each color/symbol indicates a different structure.

The image I feed Arahweave doesn’t have to look like the design drawing. Here’s the design sketch for Bipolar Prison:

Here’s the image I put into the weaving software:

And here’s the finished piece:

The purpose of the middle image – the one I fed into the weaving software – is just to specify which weave structure appears in what area. I deliberately chose colors that were easy to tell apart.

If you want to reproduce an image precisely, though, there are a few more steps. Step 1 is to weave samples of every weave structure you’re thinking of using, with every color you plan to use it with. The number of samples depends on the complexity of the image. For this stole, I wove about 240 samples, most of which are shown at the start of this post.

Step 2 is to photograph the samples, being careful to get an accurate representation of the color.

Step 3 is to take the photos into Photoshop, average the image of each sample into a single color, and create a palette consisting of the sample colors.

In Step 4, you tell Photoshop to “translate” your design sketch into an image that contains only the colors available in your palette of samples.

In theory, this should be easy. And some projects actually are that simple. My piece Goodbye, Ma! was quite straightforward. Here’s the design sketch:

Here’s an enlarged view of the image with reduced palette:

You can see how the pixels blend into something very like the original, even though the palette is very limited.

Unfortunately, Photoshop’s color substitutions don’t always work for jacquard weaving. When substituting, it will tend to choose colors that are close in value (darkness) rather than colors of the same hue (color family, e.g. red, green, blue, etc.). So if it’s looking for a color that is “similar to” forest green, it may well pick brick red rather than medium green, because the brick red is about the same darkness as the original color. This more or less works for photos because the pixels are tiny, so the eye blends everything together. Weaving happens at a much larger scale, so the colors don’t average as much. A dot of brick red in a sea of medium green is pretty obvious.

It’s actually worse than that. In weaving, your colors are limited to mixtures of warp colors and weft colors. So the medium green is is actually dots of bright green weft mixed with dots of black warp, which averages out to medium green. The brick red is actually dots of bright red weft mixed with dots of black warp, averaging to bright red. But since the dots are relatively large, the eye picks out the bright red dots in the sea of green, and says, “Hey, what’s up with that?”

I had the same problem with Bipolar Prison, which contained a lot of similar colors. To keep Photoshop from getting confused, I had to separate the design sketch into four parts: the face, the red bar, the blue bar, and the background. I reduced each component to the right set of colors for that component, then reassembled them into a single file.

Everett’s Stole has the same issues, with an additional complication: It contains a lot of colors. Too many to do with a single set of wefts, unless I want to weave with 11 shuttles. Even I am not that much of a masochist – and it would create other problems.

So to get the many colors I need, I plan to use different weft colors in different sections.

Since it’s been awhile since I last mentioned it, let me describe the project.

My friend Everett is in seminary, and commissioned me to weave him something special for his eventual ordination as a Unitarian minister. At his request, I sketched out a lake at sunset, using a design that could be divided in half, each half running down one side of the stole. When worn, it will look like this:

My plan is to weave both halves at once, separated by an inch or two of cloth in the center, and with an inch or two of seam allowance at either selvage. This ensures that the halves will match perfectly.

Here is the sketch for the two halves put together. The thin red line down the middle shows where the cloth will be divided in half to make the stole. Just before weaving, I’ll move the two halves apart in Photoshop and add enough cloth between them to create the seam allowances.

 

This design contains a ton of colors – more than I could get by mixing four or five wefts + black. So my plan is to use different weft colors in each section of the stole. Here’s my initial plan, showing the colors I planned to use in each section:

This approach breaks the stole into five sections, each with a different set of weft colors:

1. The bottom of the lake: Silver, indigo, forest green, teal
2. The middle of the lake: Indigo, teal, medium green, forest green
3. The line between the sunset and the lake: Gold, orange, lavender, forest green
4. The sunset just above the lake: Gold, orange, lavender, indigo
5. The night sky: Silver, indigo (I added navy blue and deep purple later)

Unfortunately, breaking up an image into sections creates problems wherever the sections meet. Let’s suppose that you are using orange in one section but not the next. If orange appears in the image right up to the boundary between sections, there will be an abrupt and highly visible line where the orange disappears. Similarly, the orange needs to appear gradually, or there will be a harsh line at the bottom of the section.

So in addition to dealing with Photoshop’s conversion problems, I need to figure out how to phase colors in and out at the section boundaries. Unless I discover a handy Photoshop trick, I’m pretty sure this means a lot of tedious hand-painting.

Nonetheless, I’m enjoying this project. I’m enjoying it precisely because it is difficult. I love projects that are so hard that I don’t know if they’re possible. I love projects that require me to invent new methods and stretch the boundaries of what I already know. I love a good challenge, and this one is one I can really sink my teeth into.

I have a deadline on this project, so I’m probably going to focus mostly on it for a few weeks. Minus the time spent on tomato planting, of course. That project has also gotten a little (um, well, actually a lot) out of hand, but this blog post is already insanely long, so I’ll write about it on a different day.

I will, however, leave you with a photo of the two most gorgeously wonderful cats in the world (I bet you can guess who that is!) I’m pretty sure they were hunting birds. Or trying to.

 

I’ve been benchmarking my weaving speed on Lady Ada. The change has been pretty darn impressive.

When I started out, I had to relearn how to weave on a treadle loom. I made a ton of mistakes, and the weaving went correspondingly slowly – about 20 minutes to weave an 10″ wide, 8-inch long sample in 2/2 twill (and the result still looked pretty crappy):

 

20 minutes for an 8-inch long sample at 30 picks per inch = about 720 picks per hour.

Then my friends Sandi and Kaye came over and tuned up Lady Ada (my 8-shaft Baby Wolf loom) for me. And I switched her from a friction brake to live-weight tension brake. (Live-weight tension provides perfectly even tension on the warp, whereas tension must be manually adjusted on the friction brake, potentially leading to all sorts of tension variations while weaving.)

Here’s the sample I wove yesterday morning (the photo is a little wonky because the sample is still on the loom):

See how much more even the weaving is?

But that’s not the amazing part. The amazing part is that an 8″ long, 10″ wide sample now takes me just over 6 minutes to weave. That works out to 2,250 picks per hour, or 1.6 seconds/pick! (For non-weavers, a pick is a single throw of the shuttle, laying down a single weft thread.) That is a new personal record – my previous best was 1.8 seconds/pick, on my wedding dress. (Admittedly, that was on a 24″ warp, which would naturally take longer.)

My guess is that I can still increase that speed somewhat – there’s some fine-tuning I still need to do. And of course it’s a very simple weave on a very narrow warp. But the idea of being able to weave 75 inches (just over 2 yards, just under 2 meters) of cloth per hour is simply amazing. I could weave off 6 yards of samples (24 samples) in just three hours! Talk about instant gratification.

By way of comparison, I’m also weaving samples for Everett’s stole on Amazing Grace:

Weaving with four shuttles is much slower – about 700 picks per hour. At 230 picks per inch, that’s…3 inches an hour. I could weave 25 yards of samples on Lady Ada in the time it would take to weave a single yard on Grace! Each row in this diagram represents about 20 minutes of weaving (200 picks).

Of course, speed is not everything – Grace’s patterning capabilities far outstrip Lady Ada’s, and she is working on a much finer thread count as well. Her work will be far more spectacular than the simple weaves I’m doing on Ada. But the weaving I’m doing on Ada gives much more instant gratification!

So far I have woven four of the six sample sets for Everett’s stole. After I finish all six (next week I think, assuming our heat wave doesn’t continue), I’ll be done with samples and can get on with the design. The design will be a real technical challenge – I expect it will involve considerable head-scratching and image manipulation. I’m hoping I can pull it off – if I can, the results will be spectacular. (And if not, I’ll modify the design to something do-able. But I’m pretty sure I can figure this out.)

More on that in a future blog post!

I’ve now woven a few warps on Lady Ada, and am starting to get the hang of working on a treadle loom again. I haven’t owned a treadle loom since I was a wee beginner – I bought one in October 2006 when I started weaving, but within seven months I had moved to a computer-driven dobby loom. I’ve been weaving on various computer-driven looms ever since. So despite having almost eleven years of weaving experience, I’m basically a complete novice to weaving on treadle looms.

What’s different between a treadle loom and a computerized loom like my AVL 40-shaft loom or Amazing Grace?

Quite a bit. They both require skill and attention to produce good work, but the needed skills are different. On a treadle loom like Lady Ada (who, for new readers, is an 8-shaft Baby Wolf), everything is mechanical. Before you throw a single pick, you have to choose a treadle, press it, and hope that you tied the correct shafts to that treadle, so it raises the correct set of threads. Then you throw, beat, release the treadle. And then you have to find the next treadle in the sequence, press it, and so on.

This probably sounds like a stupidly simple process, but there are all sorts of mistakes that can be made, and I have been merrily making all of them. In the course of a single afternoon, I managed to make at least eight different kinds of mistakes, most of them repeatedly. Who knew there were so many ways to screw up weaving a simple 4-shaft twill? But because I am an extremely talented and determined woman, I managed to tie the treadles incorrectly, press the wrong treadle, lose track of where I was in the treadling sequence, accidentally treadle the sequence backwards, and a few other errors which I won’t mention because they are simply too embarrassing. All in the course of a single afternoon! It took me nearly 45 minutes to weave a single eight-inch sample of 2/2 twill, and I’m not going to tell you how many times I had to start over or unpick things.

Of course, that was only the start. I also rapidly discovered that advancing the warp on Lady Ada can be an exciting journey into adventure. How was I to know that if you press too hard on the friction brake release, the warp flings itself off the warp beam with the speed and fury of a striking snake? Or that when you advance the warp, you have to be super totally careful to get the tension just the way it was before, if you don’t want weird lines in your cloth?

Here’s the kind of cloth I was producing that first afternoon. The right side is plain weave, the left side is 3/1 twill.

Obviously I was having major problems keeping the beat even – the yellow striations are where the yellow weft beat in more densely, and the black stripes are where the weft was less dense.

Since then I’ve gotten a bit better, and Lady Ada has gotten a complete tune-up thanks to my friends Sandi and Kaye. Here’s a more recent sample:

There’s still some variation, but not nearly as bad.

The challenge of weaving on a treadle loom is that you are doing all the work. It’s up to you to choose the right treadle, to tie the treadle to the shafts correctly, to advance the cloth, to set the tension, and a host of other small details that most computer-driven looms handle for you. So if you’ve spent virtually your entire weaving life on a high-tech loom, as I have, you probably haven’t learned how to do all those mechanical things. You haven’t needed to!

Until, of course, you sit down at a lovely little treadle loom and try to weave something…

I’m not saying that a treadle-loom weaver is more skilled than a high-tech-loom weaver, by the way. The skills are different. Amazing Grace, my TC-2 jacquard loom, is about as high-tech as you can get in a hand-operated loom. When I’m weaving, I just push a button and she automatically lifts all the correct threads; after I’ve thrown the pick and beaten in the weft, she automatically advances to the next pick in the sequence. If the phone rings or I need to run off to rescue something from the cats, I don’t need to worry, because Grace will remember exactly where I was in the sequence when I got up. Grace maintains a precise warp tension for me automatically, using embedded sensors to keep the tension exactly identical over every inch of weaving. Finally, she auto-advances every fraction of an inch, keeping the fell line in exactly the same place, so it’s easier to keep the beat even. Had I woven that black-and-yellow sample on Grace, I wouldn’t have had any of the unsightly striations because Grace handles all those adjustments for me.

But working on Grace requires a different set of skills. For example, you need to know how to maintain and troubleshoot her. Treadle looms, once you’ve got them working, are mechanically simple. If you press a treadle, and the wrong shafts lift, it’s because you tied it up wrong. On Grace, or on her predecessors, if the wrong threads lift, it could be any of a long list of possible problems, each with a different solution.

You also need to be able to design things to weave on her. It took me years to figure out how to design effectively for multishaft, computer-driven looms. Using Grace to her full potential requires a much deeper understanding of weave structures than my previous looms did – plus I had to upgrade to more powerful weaving software and design techniques to do it.

Grace (the jacquard loom) also requires making more adjustments as you’re designing and weaving. I am weaving about 200 color samples on Grace for my current project, and then I’ll weave an aspect ratio sample that tells me how much I need to stretch or squash my design to compensate for shrinkage, etc. All that will feed into an incredibly complicated project that is far beyond anything I would attempt with a treadle loom.

I love Grace because she can produce incredibly complex designs, and requires serious thinking (I love a challenge). I love Ada because she is simple yet beautiful, much quicker to set up and use, and exercises a totally different set of skills. (Plus, if the power goes out, she’s the same wonderful loom, whereas Grace turns into a 1000-pound rock.)

All that said, having now woven about 15 yards of warp on Lady Ada, I’m back to working on Grace for awhile. Last week I dyed the last of the yarns I’ll need for Everett’s stole, and this morning I started weaving the rest of the samples I’ll need to complete the design. I’m weaving a total of 200 sample swatches – here are the first 30:

These are for the bottom of the lake. I’m using four wefts: metallic silver, deep indigo blue, teal, and a medium green. This batch of samples shows shadings with all possible pairs of those weft colors on top, plus the four possibilities with one weft on top.

After I’m done weaving all 200 swatches, I’ll photograph each one, color-average it into a single shade, and do some more design voodoo to convert my initial sketch into a full draft.

If you’ve forgotten what the stole project looked like, here’s the sketch again:

The thin red line running down the center (if you click for the larger image) is where the stole will be divided when draped around the neck and shoulders.

The fabric for the stole will need to be about 60″ long, and the samples are weaving up at 230 picks per inch. So the stole will require 13,800 picks (weft threads), or about 20 hours of weaving. I better get going!

First, the grand news: I AM ACTUALLY WEAVING!!! I have finished debugging both warp and loom and have started weaving the samples for Everett’s stole. Hooray!! Six months of threading and debugging are over, and I can finally weave on my beloved loom.

Here are the samples I’ve woven so far. Not that impressive looking, but still quite informative:

This set of samples is testing out different combinations of warp and weft. There are four weft (crosswise) yarns being used: gold, orange, pink, and green. In the top set of samples (above the yellow line) I am experimenting not only with each individual color but also combinations of the colors; if you click through to the larger image, you’ll find that the row above the bright orange is a mix of bright orange and green, and the salmon-y shades in the same row are mixes of orange and pink.

I’m going to do a LOT of these samples. You’re seeing 24 samples (eight per row times three rows); I will probably make 100-200 of them. Then I’ll photograph all the samples and average out the color of each sample in Photoshop to a single color. After I have a large enough palette, I’ll reduce the image to just the set of sample colors, and use that image to generate the file for the loom.

Tedious and time-consuming, but well worth it in the end.

And…I AM WEAVING!! For the first time in six months. What a joy, and what a relief!