They've figured out how to print a 3D object all at once

February 2, 2019 9:37 AM

Parallel 3D printing. Instead of the usual way of layer by layer.

Said they were inspired by Computer Tomography. Resolution of 0.3mm and massive printing speedup. Basic technique: "We used a digital video projector to output our computed intensity-modulated projections, which we time-sequenced to the rotation rate of the uncured photopolymer precursor material. "


Additive manufacturing promises enormous geometrical freedom and the potential to combine materials for complex functions. The speed, geometry, and surface quality limitations of additive processes are linked to the reliance on material layering. We demonstrated concurrent printing of all points within a three-dimensional object by illuminating a rotating volume of photosensitive material with a dynamically evolving light pattern. We print features as small as 0.3 mm in engineering acrylate polymers, as well as printing soft structures with exceptionally smooth surfaces into a gelatin methacrylate hydrogel. Our process enables us to construct components that encase other pre-existing solid objects, allowing for multi-material fabrication. We developed models to describe speed and spatial resolution capabilities. We also demonstrated printing times of 30–120 s for diverse centimeter-scale objects.

posted by aleph (29 comments total)

Cool! Extra cool because the technology is a DLP projector, a turntable, and some math.

Reminds me of this “top down” printing method pioneered by some Sci-Arc students a few years ago.
posted by q*ben at 9:44 AM on February 2, 2019 [3 favorites]

They have some videos in the supplementary material page.
posted by dhruva at 9:44 AM on February 2, 2019 [6 favorites]

Thanks dhruva, should have linked them.
posted by aleph at 9:45 AM on February 2, 2019

One step closer to Star Trek replicators.
posted by duffell at 10:26 AM on February 2, 2019

I like that one demo example was a philosopher icon and the other was a stabby thing.
posted by polymodus at 11:39 AM on February 2, 2019

This is extremely cool and also a kinda "d'oh!" thing. Why didn't anyone think of this before?
posted by sjswitzer at 12:06 PM on February 2, 2019

They mentioned they were inspired (in a reverse kinda way) by Computer Tomography. So until it came along and the techniques were widely known...
posted by aleph at 12:10 PM on February 2, 2019

odinsdream: I imagine it's a trade-off between how fast they can rotate the platform (with the object) and how much power they have to pour in to solidify the object as fast as necessary (to keep it stable, develop fully, etc). With how fast they can rotate => how fast they can make the whole object
posted by aleph at 12:17 PM on February 2, 2019

But CT has been around for ages, and with modern GPUs it's well within the reach of an average laptop computer to do. So someone had a great "aha!" moment to put all that together. This wonderful because of--not in spite of--the fact that once somebody finally thought of it, it was so stunningly "obvious."
posted by sjswitzer at 12:17 PM on February 2, 2019

Not disagreeing. Does seem like a "aha" moment and hard to say when those will come.
posted by aleph at 12:23 PM on February 2, 2019

If only the matrix "fluid" could be semi-solid.

How does the object remain stable within the liquid?

The supplementary videos are kind of annoying to view because you have to download them, but one of them (Movie S5) shows that they're using a very viscous resin.
posted by teraflop at 12:30 PM on February 2, 2019

That actual answer to the question of "why didn't anyone think of it before?" is that people did think of it before. They just didn't follow through or didn't have the resources around them to make it a reality. Inventing something new or making a new discovery is like committing a crime. You need motive, means, opportunity, and a lot of luck.
posted by runcibleshaw at 12:31 PM on February 2, 2019 [5 favorites]

From the paper:

"We used highly viscous (up to ~90,000 cP) or solid (thermally gelled) precursor materials to minimize relative motion between the printed object and the precursor."

The viscosity of thick honey is about 20,000 cP (from here; one centipoise is one millipascal-second) so the feedstock is very, very viscous indeed.
posted by Major Clanger at 12:37 PM on February 2, 2019

About this time last year I was futzing around with a DLP projector and photosensitive gels in a way that was both similar to this and very much not - we were doing it at very tiny scale by focusing the projector output through a microscope, but only creating 2D patterns with some thickness, which of course skips all of the neat tomography stuff they did to make it work with the turntable. Very cool!
posted by btfreek at 12:40 PM on February 2, 2019

Oh, and yes, as someone who has separately studied both CT scanning and 3D printing, I read this with the mother of all forehead slaps - as I suspect did many others.

It's so obvious when someone else does it, which is usually the mark of inspired genius.
posted by Major Clanger at 12:41 PM on February 2, 2019 [6 favorites]

The accuracy and level of detail is really impressive, given how cloudy the gel looks. There's more detail about the optimisation they did on page 8 of the supplement.
posted by lucidium at 1:03 PM on February 2, 2019 [1 favorite]

The image projected onto the gel is fascinating...it looks like the 'dancing lady' optical illusion.
posted by sexyrobot at 1:07 PM on February 2, 2019

Looking at page 8 of the supplement linked above, it occurs to me another reason this has not been done before is that it involves a boatload of hairy math and some exacting lab skills as well. So, yeah, as someone pointed out above, being able to even realize the idea is a pretty crucial aspect. Well done.
posted by sjswitzer at 1:12 PM on February 2, 2019 [2 favorites]

Metafilter: a boatload of hairy math
posted by dhruva at 1:29 PM on February 2, 2019 [1 favorite]

It's much more like a holodeck than a replicator to me. For a long time the idea of pumping energy into a tiny space in 3d has been a thing. Using sound to blow up kidney stones, targeting tumors with radiation, rotation plus projection to create 3d images in a defined space.

It's really cool that somebody is going and doing it for this.
posted by zengargoyle at 2:19 PM on February 2, 2019

The same process could use a camera that rotated around the substrate, right?
posted by grumpybear69 at 2:25 PM on February 2, 2019

I kinda get how this works, but -- how does it work? Reconstructing a 3d shape from tomographic data is computationally intensive and not particularly "local", while in this case a given bit of the resin only knows the integral of its illumination over time. Does this work for any shape, or are there some "inexpressible" shapes? Does the fact that the resin can apparently perform the computational step tell us anything about how to improve tomographic reconstruction software?
posted by the antecedent of that pronoun at 2:27 PM on February 2, 2019

or, like the "matrix time" special effect was done, could you project with a large number of cameras all at once?
posted by the antecedent of that pronoun at 2:28 PM on February 2, 2019 [2 favorites]

the antecedent of that pronoun, I've only glossed the article. In one dimension is looks sorta like this: Sinc function. You'll notice that the wave is mostly flat until that little bit where it spikes up. In two dimensions it looks sorta like: gnuplot demo script: surface1.dem (look for the 'Sinc' section, but same thing... the wave is mostly little but in that center there's a big spike. They are doing this in three dimensions from two projectors. So in 3d space there's a whole bunch of low level waves going through the matrix and doing nothing but at that 3d surface bit where they've calculated things just so, there's a spike in the energy. That spike is the thing that causes the matrix to become solid.

You could do this 'matrix' style, but you'd still have to build up the layers from the inside to the outside. They are roatating the matrix to bring new surface deposition points into the field of view of the two projectors that they are using.

The cool thing (If I gloss correctly) is that instead of using say two precision aimed lasers of 0.3mm resolution and solidifying the matrix a point at a time, they are projecting a 2d image through two projectors to do the whole field of view of those projectors all at once. Then they rely on the rotation of the medium to build up the layers.

Matrix style would just take more projectors placed around the matrix so that the whole workable area was within the field of view of the projectors so they could build each 0.3mm surface layer in one go. But it's sorta like a lathe, It's easier to just use two projectors and rotate the matrix.
posted by zengargoyle at 3:24 PM on February 2, 2019

It's like this: INORI (Prayer) but instead of projecting onto a face, they use two projectors and it's the sum-total of the light energy as it passes through the matrix that causes the matrix to solidify.
posted by zengargoyle at 3:31 PM on February 2, 2019

Parallel 3D printing.

Damn. I just threw out my parallel printer cable.
posted by w0mbat at 4:32 PM on February 2, 2019 [6 favorites]

If you wanted to avoid sloshing the medium around, you could build a mechanism to revolve the projectors around the vessel instead of rotating the vessel. Then the next step would be adding the capability to move the projectors independently — it might not bring any technical advantage, but it would look super high tech and awesome (especially when you immerse a person into the goo to print articulated space combat armor around them).
posted by thedward at 6:53 PM on February 2, 2019 [2 favorites]

Finally a good use for those three spinning ring things.
posted by Mitheral at 4:46 AM on February 4, 2019

Needs more Utah Teapot.

Although, I suppose being a new technology, it should have a new standard object. I demand that "The Thinker" be included in any further demonstrations of CAL.
posted by ArgentCorvid at 1:26 PM on February 4, 2019