String art is deceptively complex. It seems pretty simple: it is just thread wrapped around posts (usually nails) that makes a pattern, such as text, a shape, or an image. But the math behind making a string art image is extremely advanced, because the accuracy of the image comes from the density of overlapping strings and their angles. A talented artist can do that by hand, filling in areas as they go along. But building a machine to produce string art is a massive challenge, as demonstrated by Paul MH.
In this video, Paul walks viewers through his journey to develop a string art machine from scratch. That came with numerous hurdles to clear and the mechanism to insert nails is a perfect example. The machine itself is Cartesian, with linear X and Y axes very similar to a CNC router. It operates under the control of a RAMPS 1.4 board, which is common for machines of this style. But grabbing nails and sliding them into the picture's base (foam board) was not a trivial task.
Paul initially developed a mechanism similar to what you'd see on a nail gun, with a hopper on the moving machine head that fed nails to an end effector. But that was unreliable and also required the machine to carry around the full weight of all the nails, which could number in the hundreds. That weight made the machine slow, so Paul devised another remote mechanism to feed nails for the end effector to pick up. That was still unreliable, so it required a webcam with a neural network that visually checks to see if a nail is present.
This machine also required a way to feed thread, which was actually the easiest part of the project. The thread comes off the spool through a PTFE tube out of a "nozzle" on the end effector. After attaching the thread to the first nail, the tension is enough to keep unspooling the thread as necessary.
The software was Paul's most impressive feat. Using the dark magic that is mathematics, it takes a photo as input and calculates the thread paths that would create the best visual approximation. It is a little bit like dithering a monochrome picture, but far more complicated. The software then generates g-code for the machine to follow so that it can wrap thread around the nails in the proper order while avoiding collisions.
We're only scratching the surface of the work that went into this machine, so be sure to watch the full video to see how Paul made this all come together.