Preparing to launch

Having (mostly) debugged the loom, I’m starting work on my first project, the rendition of the SkySat-2 launch photo (a gift for the former CEO of Skybox). Here’s the original photo:

Since it was a bit dark, I asked my photographer friend Lieven if he could lighten it up a bit. He obliged with this version:

I’m not sure which version I’ll use yet. While the retouched image is definitely brighter, lightening up the clouds eliminated the contrast that makes the first photo so dramatic. I am leaning towards some kind of combination – using the lighter sky and brighter rocket in the retouched photo along with the darker clouds and dramatic blast of fire in the original photo. Gotta mess around in Photoshop a bit to get the final image.

The overall process for converting an image to jacquard draft is fairly straightforward. You reduce the image to a certain number of colors, then you assign each color a weave structure. You replace each pixel of color with a unit of the weave structure, and presto, you’re done!

Of course there are a few gotchas involved, especially when you are trying to reproduce a photo. In particular, when you are reducing to a handful of colors, you need to choose those colors carefully. Weaving is not like painting; you cannot smudge blobs of paint until you get the colors you want. Instead, you have a few weft colors and a few warp colors to work with, plus the combinations of those colors. So simply telling Photoshop to reduce the photo to 12 colors won’t work. You need to tell Photoshop what colors are available to use. Which, in turn, means you have to know what colors you can weave. This, of course, means sampling.

So I dyed about two pounds of yarn orange, blue, and brown, and spent several hours today making a weave blanket. This is an extensive set of samples containing all the weave structures you might use. Mine has 126 structures in it – the 84 possibilities for an 8-end satin in double weave using three wefts and two warps, plus another 42 possibilities created by swapping a white weft for the orange one. (Thankfully, I did not have to create the weave blanket myself; my weaving buddy Pat constructed the 84 structures and graciously sent me the files, so all I had to do was plug it into the TC-2 control software and weave. Thank you, Pat!)

Here’s a photo of the halfway-finished weave blanket:

Each row of the sample contains shaded 8-end satins using some combination of the two warps and three wefts. The satin structures transition gradually from weft-dominant (left) to warp-dominant (right). The illegible patches in each row are supposed to be numbers identifying each structure, but they were too small to read legibly, alas.

Having created this weave blanket, my next step is to convert the samples to a set of colors in Photoshop, so I can use the sample colors in the conversion. So I took some close-ups of the samples:

My plan is to use the Eyedropper tool in Photoshop, with a 31-pixel sampling radius, to “average out” the color in each sample. Of course it won’t look entirely like that in real life since there will be dots of color rather than a single color, but it will give a rough idea of how the colors will blend from far away.

Once I’ve got a set of swatches, I’ll reduce the image to weave-able colors using Photoshop. And then I’ll take the reduced image into ArahWeave and replace colors with weave structures. And then I’ll weave some more samples using the finished draft, before weaving the real deal. It will take awhile, but hopefully the results will be worth it!

All that said, I probably won’t get to this until next week at earliest. This coming weekend is expected to be relatively cool, so I’m going to launch into chocolatiering starting early Saturday morning, testing another set of new flavors. I’m really excited about the possibilities.

How excited? Almost as excited as Tigress in the photo below. Look at those giant pupils! This cat is ready to launch.