The researchers at MIT’s Computer Science & Artificial Intelligence Laboratory (CSAIL) have developed a computer-aided knitting system which can automate the design and manufacture of knitted garments. Even non-expert knitters can use the system to create individual designs from customizable templates on a computer. The designs are then sent to a knitting machine, which fabricates the designs out of real wool.
Companies like Thingiverse and Shapeways has given customers a huge opportunity of creating garments according to their needs, such as cosplay accessories and replacement parts.
The templates are common shapes such as hats and gloves that are easy to recreate. Users can apply textures and patterns to the garment as well. Users also have the opportunity to create garments from scratch according to their preferences. There is a neural system which will create more panels from an already existing pattern.
A team of researchers at MIT’S CSAIL, headed by computer scientist Alexandre Kaspar, release two papers describing the software today. One is called InverseKnit, that will automatically create patterns from photos of knitted items. The second one is about new design software, CADKnit. CADKnit allows users to customize templates, adjust the size, shape and other decorative details.
The researchers believe that the system is something new and impressive, but it certainly lacks out some of the creativity achieved by knitting by hand. The researchers want to make it easy for new knitters to start creating items and the system to make manufacturing of garments more efficient and waste-free.
CADKnit and Inverseknit want to make designing as easy as 3D printing now. Kaspar says that it can also enable designers to spend less time learning how to create knitting patterns and save time in the manufacturing process.
Kniterate is a potential partner for CADKnit and InverseKnit. It is a company that creates digital knitting machines for beginners, marketplaces and small businesses. Kaspar has been connecting with Kniterate’s team about making knitwear customization more easily accessible.
How does it work?
InverseKnit was developed by creating a dataset of knitting patterns with matching images that were used to train a deep and neutral network to create machine knitting patterns. The system was able to produce accurate results for 94% of the time.
CADKnit combines 2D images with CAD and photo-editing software to create customized templates. It was tested with knitting beginners, who were able to create complex garments with patterns, prints and lace details.