Recycling 3D filament is a great idea in theory, and we come across homemade filament extruders with some regularity, but they do have some major downsides when it comes to colored filaments. If you try to recycle printer waste of too many different colors, you’ll probably be left with a nondescript gray or brown filament. Researchers at Western University, however, have taken advantage of this pigment mixing to create colors not found in any commercial filament (open access paper).
They started by preparing samples of 3D printed waste in eight different colors and characterizing their spectral reflectance properties with a visible-light spectrometer. They fed this information into their SpecOptiBlend program (open source, available here), which optimizes the match between a blend of filaments and a target color. The program relies on the Kubelka-Munk theory for subtractive color mixing, which is usually used to calculate the effect of mixing paints, and minimizes the difference which the human eye perceives between two colors. Once the software calculated the optimal blend, the researchers mixed the correct blend of waste plastics and extruded it as a filament which generally had a remarkably close resemblance to the target color.
In its current form, this process probably won’t be coming to consumer 3D printers anytime soon. To mix differently-colored filaments correctly, the software needs accurate measurements of their optical properties first, which requires a spectrometer. To get around this, the researchers recommend that filament manufacturers freely publish the properties of their filaments, allowing consumers to mix their filaments into any color they desire.
This reminds us of another technique that treats filaments like paint to achieve remarkable color effects. We’ve also seen a number of filament extruders before, if you’d like to try replicating this.
I wonder if you could mix in the hot end, perhaps with some ultrasonic agitation to keep it moving through.
Kubelka-Munk theory works fine for a very thin layer of colorants mix (textile or printing), but perform poorly for relatively thick layers, like paints, not even talking about plastics. It will kind of work for completely opaque materials, but if plastic is even a tiny bit transparent, it will give bad results. For paints and plastics nobody use Kubelka-Munk theory to formulate recipe. Multiflux approach over theory of radiation is used in one or another form.
However, in the case of plastic waste reuse and if acceptable color difference is relatively large (say, ΔE2000 above 3 is acceptable, i.e. approx 80% of people will clearly see color difference between desired color and what we got), then, I think such tool could be used. Say, if you need abstractly green plastic for some standalone thing, and have some white, yellow and blue waste, then go for it. You will get some green. But don’t expect that you could create exact color you need, say, to print some nice addon for car dash that should look authentic, and nobody will notice color difference. :)
When I used to make PLA filament (I invented it) we would only use a maximum of 10% “regrind” – filament that was unsaleable due to colour changes, diameter changes during setup etc.
It all got used to make black. With more than 10% the quality drop was unacceptable by our admittedly high standards. Those are what differentiated us from overseas suppliers at the time. I hate to image the quality of recycled filament that as been heated up that many times – a miracle if it feeds without breaking.
We used to (and Imagin still do) sell “Virgin Black” for those in need of perfect quality.
Did you really invent PLA filament?
Also this explains why proportionally speaking, my black filament purchases have been awful in quality. Snapping mid print regularly, etc.
Yes, I did. I replicated the first RepRap back in the day. We got fed up with rolling CAPA by hand. As the things would replicate a lot, I looked for something vaguely environmentally benign that welded to itself, found PLA. Got Imagin Plastics of Auckland to extrude it on a knitting needle machine, developed a blend for printing with NatureWorks, and here we are. I deliberately did not patent it.
All I want is something for me to throw in my old prototypes and produce a reel of PETG to print with. I don’t care what colour is produced.