Fairy Lights

Figure 1. Fairy Lights. (Image: Ochiai et al., 2016)

Figure 2. System setup. (Image: Ochiai et al.)

Figure 3. Rendering voxels (Image: Ochiai et al.)

Figure 4. Example applications of laser-based graphics technology. (Image: Ochiai et al.)

Fairy Lights is the name given to the images created by a method of rendering aerial and volumetric graphics using a focused femtosecond laser, and to the research project that introduced it. The aerial plasma, induced by laser, emits light without interaction with physical matter. The practical result of the ionization of air molecules is crackling, photon-emitting pockets of plasma (voxels) that can be arranged to create moving and interactive images. The technology was developed by Yoichi Ochiai, Kota Kumagai, Takayuki Hoshi, Jun Rekimoto, Satoshi Hasegawa and Yoshio Hayasaki, researchers at the University of Tsukuba, Utsunomiya University, the Nagoya Institute of Technology and the University of Tokyo, and first published in a paper called Fairy Lights in Femtoseconds: Aerial and Volumetric Graphics Rendered by Focused Femtosecond Laser Combined with Computational Holographic Fields. The paper was posted to arXiv on 22 June 2015, shown as a demonstration in the Emerging Technologies program of SIGGRAPH 2015 in Los Angeles (9-13 August 2015), and published in ACM Transactions on Graphics in February 2016.

According to Ochiai et al., the main contribution of the paper was "the production of an in-air SLM-based [Spatial Light Modulator] laser-plasma graphics that enables physical contact and interaction by ultra-short pulse duration laser." The points of light can be seen from any direction, and a finger that touches them feels an impulse from the shock wave the plasma produces. The authors tested two laser sources and reported spatiotemporal resolutions of 4,000 and 200,000 dots per second. The displayed images are very small (the paper reports workspaces from 1 mm³ to about 1 cm³), and the authors proposed uses such as spatial augmented reality on real objects, floating user interfaces with touch feedback and aerial volumetric images.

Reviewed 17 October 2026. Checked every claim against the arXiv paper full text, Crossref DOI metadata, the archived Nasdaq and PhysicsCentral articles, IEEE Spectrum, the Tsukuba project pages, Pixie Dust, S+T+ARTS, the US patent record and the bubble-display and Optical Trap Display sources. About review dates.

Background

Three-dimensional (3D) displays have attracted the attention of researchers and the general public for decades. The paper's introduction traces 3D displays back to Ivan Sutherland's head-mounted display of 1968, then to glasses-based stereoscopic methods (anaglyph, time-division and polarization) and glasses-free methods such as parallax barriers and lenticular lens arrays. The authors note that these planar methods must calculate precise images for multiple viewpoints and that users have to stay within a limited viewing angle. IEEE Spectrum's Evan Ackerman wrote that science fiction had promised three-dimensional midair displays "since at least the first Star Wars movie".

With volumetric displays, users can see the images produced from any angle. They can be divided into two categories according to the characteristics of the voxels. The first is voxels that emit light, and the second voxels that reflect light. Those in the first category can be LEDs, end points of optical fibers, or laser-induced plasma. The voxels in the second category that reflect projected light can take the form of a fog, water drops, or floating small particles. The paper further divides earlier systems by whether the position of the light-emitting or light-reflecting element is controlled, and places its own work in the "position-control and emission" category, alongside earlier laser-plasma displays.

In the study conducted by Ochiai and colleagues, they used laser-induced plasma, which has some advantages: 1) "it does not require physical matter arranged and suspended in air to emit light," 2) "it does not require wires and structures that possibly obstruct the line-of-sight because power is transmitted wirelessly", and 3) "the laser can be precisely controlled owing to the progress in optical technologies." The authors state that plasma generated by a femtosecond laser is safer than plasma generated by a nanosecond laser.

Previous studies pioneered the development of laser-plasma 3D displays using a nanosecond and femtosecond laser. As summarized in the paper, Kimura et al. (2006) demonstrated the basic concept with a nanosecond laser at a rendering speed of 100 dots per second, and Saito et al. (2008) reached 1,000 dots per second with a 100 fs laser; the voxels were bluish white because of plasma emission. Ochiai and colleagues described these studies as pioneering but incomplete, because they did not publish a detailed discussion of light emission, design space and scalability, and they intended to provide a complete discussion of laser-plasma graphics technology. They also noted that earlier in-air systems were harmful to the user's hand and that an in-water system rendered images in a transparent container filled with water.

How it works

There are two main methods to render graphics in air using a femtosecond laser. The first is holograms by spatial light modulation technology and the second is the scanning of a laser beam by a galvano mirror. The abstract of the paper states that the holograms and workspace of the system occupy a volume of up to 1 cm³, and its conclusion gives maximum values of 1 cm² for the hologram size and 5 cm³ for the workspace. Although these sizes are still too small for the practical applications proposed in the research paper, the authors describe the study as a first step towards designing laser-based aerial volumetric displays, and the sizes are scalable according to the optical devices and setup used.

Laser-induced emission

The paper describes three laser-induced effects that produce light spots. In fluorescence, a material absorbs one or more photons and emits a new photon; with multiphoton absorption the emitted light has a shorter wavelength, and a relatively low-intensity laser is enough. In cavitation, a laser focused in a liquid creates microbubbles that scatter the incident laser light, so the spot takes the color of the laser. The third effect is ionization, which works in air: above a laser intensity of about 10¹⁴ W/cm², tunnel ionization frees electrons from air molecules, and when an electron recombines with the atom a photon is emitted. This laser breakdown looks bluish white, and Fairy Lights relies on it. A visible dot also has a faint tail along the beam direction, caused by self-focusing from the optical Kerr effect; the authors note that it is practically invisible because the focal point is much brighter.

System setup

The basic setup to generate a simultaneous-multi-point volumetric display consists of a femtosecond laser source, an XYZ scanner (galvano scanner + varifocal lens), and a liquid crystal on silicon SLM (LCOS-SLM) displaying a CGH for simultaneously addressed voxels (Figure 2). The galvano mirror moves the emission point laterally (X and Y), and the varifocal lens moves its focus along the beam (Z). The computer-generated hologram (CGH) is designed with the optimal-rotation-angle (ORA) method, an iterative algorithm that produces an array of focal spots of uniform intensity, so a single laser pulse can light several voxels at once. The galvano mirror and the SLM each cover an area of about 10 × 10 mm²; in practice the SLM adds dots within a frame and the galvano mirror positions the rendered hologram. A host computer running Windows and programs written in C++ drives the SLM, galvano mirror and varifocal lens, captures user input at 20 Hz, and watches the workspace through a USB microscope.

To generate fairy lights, there is a need for a medium that can reflect or emit light at precise locations throughout a given volume. The researchers used air, turning it into plasma. Focusing a laser very precisely, it is possible to produce a small, isolated volume of plasma, thereby creating a voxel. These can be seen as the 3D equivalent of pixels and stand for "volume element" (Figure 3).

According to Doss (2015), "the location of the voxels is determined by using a computer-generated hologram, which is a method of digitally creating interference patterns that are printed on a mask or film. The mask or film is also called a spatial light modulator and its role is to alter the phase and location of the coherent light source illuminating it." In the actual system the hologram is not printed: it is shown on a liquid crystal SLM that the computer drives as an external display. The PhysicsCentral article concludes that using the femtosecond laser, it was possible "to create multiple plasma voxels with one pulse and create voxels at a faster rate, leading to a well defined, bright 3D holographic image safe enough to touch, and feel."

Specifications

The paper describes two systems. All of its experiments used System A.

Item	System A	System B
Laser source	Coherent; center wavelength 800 nm; 1 kHz repetition; pulse width adjustable from 30 to 100 fs; pulse energy up to 2 mJ (the abstract gives up to 7 mJ per pulse)	IMRA America FCPA μJewel DE1050; center wavelength 1045 nm; 200 kHz repetition; pulse width 269 fs; pulse energy up to 50 μJ
Spatial light modulator	LCOS-SLM, 768 × 768 pixels, 20 × 20 μm2 pixel size, 100 ms response time (described in the paper as a Hamamatsu PPM parallel-aligned nematic liquid crystal SLM)	None
Galvano scanner	Canon GH-315 driven by GB-501	Intelliscan 20i
Varifocal lens	Optotune EL-10-30	Optotune EL-10-30
Objective lens	F = 40 mm	F20
Total energy conversion rate	53%	80%
Workspace (aerial display application)	1 mm3	8 mm3
Maximum spatiotemporal resolution	4,000 points per second (with 4 simultaneously addressed voxels)	200,000 points per second

The authors state that these workspaces are smaller than those of earlier plasma displays but that the resolutions are 10 to 200 times higher, and that the image frame rate depends on the number of vertices in the image. IEEE Spectrum likewise described the workspace as about eight cubic millimeters and the spatiotemporal resolution as relatively high, at up to 200,000 voxels per second, with a frame rate that depends on how many voxels an image needs.

Experiments and safety

Previous successful attempts at creating aerial plasma used nanosecond laser pulses. These types of lasers had the effect of burning tissue since the energy associated with the lit voxels was high. Ochiai and colleagues used femtosecond (a quadrillionth of a second) laser pulses to generate short-lived, light-emitting plasma voxels that would not burn skin when touched; Doss wrote that they would still burn softer tissue like eyeballs, and the paper states that the plasma still poses dangers for the retina.

The paper's experiments, all run with System A, measured the following:

• Plasma threshold. With 30 fs pulses, plasma appeared from a pulse energy of 0.2 mJ, a calculated peak intensity of 36 PW/cm²; with 100 fs pulses, it appeared from 0.45 mJ (24 PW/cm²). Both are above the ionization threshold of more than 1 PW/cm² given in the paper, and the 30 fs setting needed less energy at the same average power.
• Other media. The required pulse energy differed greatly between air, water and a fluorescent solution; the authors give the order of required intensity as decreasing from air (PW/cm²) to water to fluorescent materials (MW/cm²).
• Simultaneous addressing. Using holograms on the SLM, one to four voxels could be lit at the same time at laser powers from 0.05 to 1.84 W; five or more were not visible, because the energy is split among the dots.
• Skin damage. Leather sheets, used as a substitute for human skin, were exposed to the 30 fs and 100 fs laser at 1 W for 50 to 6,000 ms. Below 2,000 ms (2,000 shots) only holes about 100 μm in diameter appeared, with no heat damage; longer exposures produced heat effects around the holes. A nanosecond laser used for comparison burned the leather within 100 ms.
• Sound. Measured 20 mm from the plasma, the maximum noise level was 77.2 dB SPL with 40 fs pulses, against a background of 55.7 dB SPL. Brighter plasma tended to be louder.

The study used two lasers, one with an adjustable pulse duration of 30-100 fs and one with a pulse duration of 269 fs. IEEE Spectrum summarized the safety finding as a pulse duration that minuscule not resulting in any appreciable skin damage "unless the laser is firing at that same spot at one shot per millisecond for a duration of 2,000 milliseconds."

The lasers are Class 4 sources, and the authors state that the system was designed and operated based on IEC 60825-1:2014. Users should not look directly into the laser beam. Because the plasma emits visible light in all directions, it can be seen from the side of the beam, but the authors recommend glasses with infrared filters until the safety of the display for the eyes is confirmed. Doss adds that, like lightning, the plasma also emits ultraviolet and infrared radiation that people cannot see. Citing an earlier study of porcine skin that found a 50 percent lesion threshold (ED50) of 21 mJ for a 44 fs, 810 nm laser with a 12 mm spot, the authors expect damage from their much lower pulse energies (2 mJ and 50 μJ) and spot sizes under 10 μm to be negligible.

Interaction and haptic feedback

Fairy Lights are interactive (Figure 1): a person can touch them, feel them, and control them. The haptic feedback is a result of the shockwave created when plasma interacts with material. The paper reports that contact between the plasma and a finger makes the light brighter, which the system can use as a cue that a touch has happened. For safety, the voxels are switched off within a single frame (17 ms, or 1/60 s) when a user touches them, well under the 2,000 ms exposure found to be harmful in the leather tests. The authors wrote that a detailed study of this plasma-generated haptic sensation was beyond the scope of the paper, but suggested uses such as "an aerial check box".

The paper's demonstrations include the SIGGRAPH logo, a cylinder, a "heart" that is broken by touch, a "fairy", "sprouts" coming out of seeds, and a light point that changes into a "jewel" when it touches a ring. A video released by the team also showed a box that checks itself and the word "LOVE" changing to "HATE" when touched.

Comparison with stage "holograms"

Floyd noted that earlier public demonstrations of so-called holograms, such as CNN's correspondent Jessica Yellin being projected into Wolf Blitzer's Situation Room during the 2008 election coverage and the projection of Tupac Shakur at the 2012 Coachella festival, were not true holograms. CNN filmed the correspondent from 35 angles and superimposed the composite image onto the footage, while Coachella used a technique known as Pepper's Ghost. Fairy Lights instead produces tangible, aerial, volumetric graphics.

According to Floyd, Fairy Lights overcome the limitations of quasi-holograms and Victorian-era illusions like Pepper's Ghost: because the ionized molecules themselves form the image, no medium such as glass, smoke or water vapor is needed to project onto. In spite of that, the image cannot be created in a complete vacuum. The same laser setup can also produce light in water and in fluorescent materials, through microbubbles and fluorescence rather than air plasma. Since the images created are 3D, they can be viewed from any angle. In this system the computer-generated hologram only shapes the laser beam; each point of the image is a light-emitting plasma spot in space rather than a projected hologram image.

Limitations

The first main limitation of this technology that needs to be addressed in future research is the system's scale-up, which is dependent on the optical devices used and the setup. Floyd reported the images as only about a cubic centimeter in volume, with 5 cubic centimeters to move around in. According to IEEE Spectrum and Floyd, a Japanese company, Aerial Burton, had a technology that projects much larger images through laser ionization in air, but these are too dangerous to touch and considerably louder.

According to Ochiai et al., three factors need to be developed in order to scale up their system: 1) increasing power of laser source; 2) shortening pulse width to increase peak power; 3) and increasing scanning speed. The development of these makes it possible to "have an amount of simultaneously addressed and scanned voxels within a single frame, keeping visible and touchable features." The paper explains that more laser power gives more simultaneous voxels but is limited by skin safety, unwanted ionization in the optical path and the optical components, while shorter pulses allow a higher repetition rate at the same peak power and lower pulse energy. The workspace size is limited mainly by the aperture of the objective lens, which is needed to reach PW/cm² intensities. The galvano mirror works at more than 1 kHz, while the SLM and varifocal lens work at less than 100 Hz.

The authors list further drawbacks: the SLM is not resistant to an intense laser, so the full laser power cannot be used; ionization can occur along the optical circuit and damage its components; and fast changes of the varifocal lens cause aberration.

Another limitation of the system is the safety issue. Plasma has high energy and can be harmful to humans. While tests on leather showed that the femtosecond laser reduces the damage to tissue, the plasma can still damage the retina. The Nasdaq article noted that the sensation of touching the image is plasma removing a tiny bit of tissue, and that the team recommends protective glasses for the eyes.

Floyd described a trade-off between large, dangerous displays and small, interactive ones, and noted that the color of the lights cannot be changed; Doss wrote that the plasma is a little noisy. The paper explains that the plasma color does not depend on the laser wavelength, which is why an invisible infrared laser can be used, and that full-color rendering would instead need multiple fluorescent materials or, in liquid, cavitation with lasers of different colors.

Applications

The authors wrote that if laser-induced plasma aerial images were made available, applications such as augmented reality (AR), aerial user interfaces and volumetric images could be produced (Figure 4), and that the technology could be applied to wearable materials and spatial user interactions. In the AR field, the authors argue that the technology has an advantage over conventional approaches in matching 3D spatial position, since the content is generated in a 3D space instead of on a 2D computer display. In their prototype, a microscope detects an object in the workspace so that the content can be overlaid on it and changed when the object and the plasma touch, an example of spatial augmented reality at the scale of the object.

Pixie Dust Technologies, the assignee of a related United States patent, lists Fairy Lights in Femtoseconds among its technologies and gives two use cases: a user interface in the air with tangible feedback, and an aerial display that can be viewed from any direction over 360 degrees. PhysicsCentral suggested examples such as a holographic keyboard projected by a future smartphone and three-dimensional vectors drawn and manipulated in a classroom.

Publication and recognition

Date	Item	Details
22 June 2015	arXiv preprint	arXiv:1506.06668. The project page states that the paper was submitted to ACM Transactions on Graphics after acceptance with major revision at ACM SIGGRAPH 2015.
9-13 August 2015	SIGGRAPH 2015, Los Angeles	Presented in Emerging Technologies (demonstration) and in the Posters program.
February 2016	Journal article	ACM Transactions on Graphics, vol. 35, no. 2, article 17, 14 pages.
2016	S+T+ARTS Prize	Honorary Mention for Yoichi Ochiai's Fairy Lights in Femtoseconds.
2016	SIGGRAPH Asia 2016 invited talk	Ochiai, "Aerial interaction with femtosecond laser induced plasma", in the workshop Virtual Reality meets Physical Reality: Modelling and Simulating Virtual Humans and Environments, Macau (doi:10.1145/2992138.2992153).
12 March 2019	US patent 10,228,653 B2	"System and method for rendering interactive aerial volumetric graphics and generating spatial audio using femtosecond lasers", filed 7 April 2016, inventors Ochiai, Hoshi, Rekimoto, Kumagai, Hasegawa and Hayasaki, assignee Pixie Dust Technologies. Besides the femtosecond laser, CGH, SLM and 3D scanner, the patent describes combining the plasma points with an ultrasonic phased array and using them to generate sound.

Later work

Yoshio Hayasaki's group at Utsunomiya University, whose members Kota Kumagai, Satoshi Hasegawa and Hayasaki co-authored Fairy Lights, went on to build a "volumetric bubble display", published in Optica in 2017. It uses focused femtosecond laser pulses, shaped by computer-generated holograms, to create microbubbles in a high-viscosity liquid (glycerin) sealed in a glass cell; the bubbles scatter light from an external source such as a halogen lamp or LED, so the image color can be changed by changing the light. Demonstrated images included a mermaid, dolphins and a 3D Stanford bunny, which can be viewed from all angles without 3D glasses or headsets.

In 2018, Daniel Smalley and colleagues at Brigham Young University reported the Optical Trap Display in Nature, a free-space volumetric display that traps a cellulose particle with light and scans it through the display volume while illuminating it with red, green and blue light, producing full-color images with ten-micrometer image points. Its reference list includes the Fairy Lights paper together with the earlier laser-plasma displays of Kimura et al. (2006) and Saito et al. (2008).

See also

• Volumetric display
• Voxel
• Holographic display
• Computer-generated holography
• Spatial light modulator
• Mid-air haptics
• Spatial augmented reality
• Jun Rekimoto