Makers have combined printed plastic and textiles for years, and the usual method is still to pause the print, lay in the thread by hand, and hope it stays put. A group at Delft University of Technology wants the printer to do that job. In a preprint dated October 10, 2026, Joram de Haas, Mehmet Ozdemir, Holly McQuillan and Zjenja Doubrovski describe FilaLoom, a mechanical tool-changing add-on that lets an ordinary desktop FDM printer alternate between extruding filament and depositing yarn. The prototype runs on a Bambu Lab A1. Its outputs include textile-like woven structures, soft yarn surface textures and parts with embedded cables that can drive motion. The paper is due at the ACM Symposium on Computational Fabrication (SCF '26), held December 6-8, 2026 in Tokyo.
The design is notable less for the idea of mixing yarn and plastic, which has been tried before, than for how little it asks of the printer. FilaLoom attaches and detaches without permanent modification. It also needs no extra electronics: the printer's own motion system does all of the work.
A Tool Changer Without Motors
Tool changers have become a recognizable category on the desktop, and the authors compare their approach with the systems on the Prusa XL and the Snapmaker U1. Those machines are designed around multiple heads from the start, with tool-changing hardware built into the gantry. FilaLoom goes the other way. It is a retrofit that reuses the motion a printer already has, so a single-nozzle bed-slinger can pick up a second tool, in this case a yarn depositor.
The bill of materials is short. The mechanism is made from printed PETG parts, a blunt-tip dispensing needle, a 1.5 x 10 x 15 mm compression spring and several 8 x 2 mm neodymium magnets. The needle carries the yarn and lays it onto the part. The preprint lists only these three standard parts, which says a lot about how much of the switching logic is handled by geometry and toolpath rather than by actuators.
When engaged, the depositor reaches 25 mm below the nozzle. That clearance matters because yarn has to be placed on top of or between printed features without the hot end dragging through work already done. The preprint gives a maximum yarn-embedding speed of 25 mm/s. The authors say yarn-placement accuracy suffers beyond that speed, while the pick-up and parking moves can run at up to 133 mm/s.
Needles, Yarn and Practical Limits
The team tested two 50 mm needle sizes: 15-gauge, with a 1.34 mm inner diameter, and 11-gauge, with a 2.5 mm inner diameter. Most of the samples in the paper used a 1.5 mm, 475 tex cotton yarn fed through the 15G needle. The researchers also used a much finer 0.6 mm, 100 tex polyester, paired with the 11G needle.
For readers who don't spend much time in textile catalogs, tex is a linear density unit, in grams per kilometer of yarn. A 475 tex cotton is a fairly heavy strand and 100 tex is considerably lighter. The authors say the needles are standard parts that can be swapped to suit the yarn diameter, and that the system handles yarn from thin lock thread to bulky knitting yarn. Stiff yarns such as metal wire or straw are not suitable, because the mechanism relies on yarn being pulled easily in any direction. Anyone building their own version should treat yarn choice and needle gauge as a single decision, not two.
The prototype is built for the A1, a bed-slinger. The authors say the concept can be adapted to other desktop FDM printers, CoreXY machines included. The preprint does not claim a drop-in kit for every printer, so expect to adapt the mounting geometry to your own toolhead and gantry.
What Comes Off the Bed
The interesting part is what yarn embedding makes possible. Covering FilaLoom for Hackaday, Ian Bos noted that the system places thread through a tool head, much like tool-changer printers, instead of relying on hand insertion. That lets it produce embedded structures larger than earlier examples of the technique. The demonstrations cover 2D and 3D textile structures, 3D knits and textile structures with embedded mechanisms.
Broadly, the outputs fall into three groups:
- Woven, textile-like structures, in which printed plastic and yarn interlace into a composite that behaves more like fabric than a rigid part.
- Soft yarn textures, in which yarn changes the surface feel of a print and gives tactile qualities that plain thermoplastic can't.
- Cable-driven parts, in which embedded yarn acts as a tendon that is pulled to bend or move a printed structure, with no separate cable routing step after printing.
The last group is the one most likely to catch engineering-minded makers' attention. Tendon-driven mechanisms such as grippers and articulated fingers are a standard maker project, and threading cables through printed channels afterward is usually the tedious part. Putting the tendon in during the print removes a step and opens the door to routings that would be hard or impossible to thread by hand.
What It Means for Makers
FilaLoom is a research prototype, not a product. The preprint is a conference paper and its DOI (10.1145/3828611.3845661) is still forthcoming. Some features of the design still make it unusually approachable for academic fabrication work.
First, the hardware is simple and familiar: PETG parts, a dispensing needle, a compression spring and some small magnets. Second, the printer stays stock. A system that attaches and detaches without permanent changes and without extra wiring is one many owners will be willing to try on a machine they also use for everyday printing. Third, the authors describe the add-on as DIY and easily implemented, aimed at consumer-grade printers.
The constraints are real. A 25 mm/s embedding ceiling means yarn-heavy prints will take a while, and the documented yarns are a specific cotton and a specific polyester rather than whatever happens to be in the craft drawer. Toolpath generation is also a hurdle: the authors generate the weaving G-code with a Grasshopper script and combine it with G-code from either the same script or a conventional slicer, so expect some manual setup.
Even so, the work points to a possibility that desktop machines have mostly ignored: soft, fibrous materials added during the print rather than afterward. For makers working on wearables, soft robotics or tactile objects, FilaLoom is worth a close look, and the full details should be available when the paper is presented in Tokyo in December.
