Two-photon polymerization has always had a geometry problem. The process cures resin only at the focal point of a tightly focused laser, so it can resolve features far smaller than any desktop resin printer can. But it still builds parts as stacks of slices, and anything that curves, such as a lens, a mirror or a phase plate, comes out with a staircase on its surface. On October 6, Vienna-based UpNano announced 3D Adaptive Resolution (3DAR). The company says the technique removes that staircase by changing the shape of the printing voxel itself, in all three spatial dimensions, while a part is being printed.
UpNano's claimed numbers are aggressive. The company reports surface roughness down to Sq 3 nm and shape accuracy below 20 nm, and it says it can get there while keeping large layer distances, without the very thin layers that normally make smooth 2PP optics so slow. Trade outlet 3D Printing Industry also covered the launch and added context on the first customer and the competitive field.
The Staircase Problem, Explained
Any layer-based process approximates a smooth surface with a series of flat steps. On an FDM part that shows up as visible layer lines on a shallow dome. You can shrink the steps by using thinner layers, but print time grows with the number of layers, so you pay for smoothness in hours. Micro-optics have a far stricter version of the same tradeoff. A lens surface only works if its deviations are small compared with the wavelength of light, and that leaves no room for terracing.
In 2PP the "pixel" is a voxel, the small ellipsoidal volume around the laser focus where the resin absorbs two photons at once and hardens. Its size and position define the smallest detail the printer can lay down. UpNano's earlier Adaptive Resolution, a core feature of its NanoOne printers, let the machine switch between a fine voxel for delicate structures and a larger one for rapidly filling bulk volumes. Fine voxels and thin slicing give smooth surfaces and slow builds. Coarse settings give fast builds and stepped surfaces.
From Grayscale to Three Dimensions
UpNano presents 3DAR as the next step after grayscale lithography. That established technique modulates the voxel size to produce continuously varying height profiles, which lets it smooth out height steps. According to 3D Printing Industry, grayscale lithography is limited to 2.5D topographies. UpNano's release says 3DAR's grayscale voxel control is not limited to 2.5D surface topographies and can create complex, freeform 3D structures.
3DAR applies dynamic voxel control in X, Y and Z during printing. According to UpNano, this lets the printer keep large layer distances and still avoid staircasing, because the voxel is adjusted to match the target geometry instead of the surface being built from many thin slices. That is the source of the "From Grayscale to True 3D" headline. The company is claiming that freeform geometry, not just height-mapped surfaces, can now be rendered smoothly at a coarse layer pitch.
UpNano has not published the control scheme in the release. The details of how the voxel is modulated, and the materials the results were demonstrated in, are not part of the verified announcement, so the 3 nm and 20 nm figures should be read as the company's demonstrated best cases rather than guaranteed specs across every resin and geometry.
Speed Is the Real Headline
Smooth surfaces are only half the claim. The other half is throughput. UpNano says a spiral phase plate 125 µm in diameter can be fabricated in under ten seconds, and that an aspheric lens measuring 7 µm takes two to three seconds. A spiral phase plate is the kind of part that exposes staircasing badly. Its surface is a continuous helical ramp, so a stepped version is exactly the error the optic is supposed to avoid.
Times in seconds matter because of where these parts are meant to go. UpNano says micro-lenses, mirrors and other optical elements can be printed directly onto photonic chips, chip facets and fiber arrays, using automatic marker detection and alignment to position the structures. Printing in situ removes the need to manufacture optics separately and then align and assemble them, which is fiddly at micron scale. If each optic takes seconds, printing many of them across a fiber array starts to look like a production step and less like a lab demo.
Why Photonics, and Why Now
UpNano is aiming 3DAR at silicon photonics and co-packaged optics. Its release describes the technique as suited to micro-optics, photonics packaging and semiconductor applications. The commercial driver it cites is the growing role of fiber-to-chip and chip-to-chip links in AI data centers. Moving light between optical fiber and photonic chips efficiently depends on precise coupling in very confined spaces, and that is the class of part 3DAR targets.
There is already a buyer. Both sources report that the first 3DAR-capable printers were bought by an Asian company, which has expressed strong interest in acquiring more systems. Neither source names the customer.
The launch also lands in a market with an established player. 3D Printing Industry notes that German 2PP pioneer Nanoscribe reached the sale of its 400th system, citing growing industrial demand for its Quantum X printers in optics and photonics packaging. That shows that high-resolution 2PP is a real installed-base business and not a niche curiosity, and it is the field UpNano is competing in. Pitching 3DAR as a fix for both surface quality and speed on freeform optics is one way to stand out.
CEO Bernhard Küenburg is quoted in UpNano's release saying 3DAR lets the company "combine printing speed with the surface quality required particularly for optical and photonic applications."
What It Means for Makers
Nobody reading this is putting a 2PP system on a garage workbench, and 3DAR does nothing for a Bambu or a Prusa. The idea underneath it is still familiar to anyone who has fought layer lines. Variable layer height in FDM slicers trades print time against surface quality by changing slice thickness where a model curves. 3DAR goes after the same tradeoff at the nanoscale, and it does so by reshaping the unit of deposition instead of just changing how thinly the model is sliced.
The broader lesson is that "resolution" in additive manufacturing is not a single number. Layer height, spot size and how the process handles slopes all set the final surface, and the most interesting gains often come from controlling the deposition unit more cleverly rather than shrinking it. That applies to nozzles and resin pixels as much as to laser voxels.
For readers in photonics, packaging or research labs, the practical questions are the ones the announcement leaves open: which materials, how the results hold across geometries, and what the systems cost. The claimed combination of single-digit-nanometer roughness and print times measured in seconds is worth testing against real parts. If it holds up, printing optics directly onto chips and fibers could move from a specialist technique toward a routine packaging step.
