The research, presented at the European Conference on Computer Vision 2026 (ECCV 2026) on September 10, 2026, addresses a persistent gap between what commercial AI animation tools can do and what creative professionals actually need. Current systems can transform a neutral motion, such as an ordinary walk, into a stylized one, such as a joyful or angry walk. That binary transformation is useful, but it is coarse. Most existing tools offer only fixed versions of a given emotion or expression, with no mechanism to increase or reduce how strongly the style comes through. The result is an all-or-nothing choice that forces animators to either accept the single stylized output or fall back on laborious manual editing, undermining the very efficiency that AI generation is supposed to provide.

The root of the problem lies in how these AI systems are trained. Motion-generation models typically learn from paired examples: a neutral motion and a fully stylized version of the same action, both captured through motion-capture technology. To teach a model what a half-stylized or quarter-stylized walk looks like, one would need to record actors performing the same movement at many intermediate levels of expression. That would demand multiple long, expensive motion-capture sessions for every action and every style, an approach that is costly and impractical for real-world production pipelines where time and budget are always under pressure. The Science Tokyo team, led by Specially Appointed Assistant Professor Chen-Chieh Liao and Professor Hideki Koike of the School of Computing, set out to eliminate that dependency entirely.

Their solution is elegant in its economy. The Motion Style Slider learns from only two endpoints: a neutral motion and a stylized version of the same action, for example neutral walking and happy walking. From this minimal pair, the model generates a continuous family of motions controlled by a user-defined scalar value called the style intensity, denoted alpha. Values such as 0, 0.5, 1.0, and 1.5 correspond to low, medium, and high style intensities, and the slider can be set anywhere along that continuum. Crucially, the range extends beyond 1.0, meaning the system can extrapolate into stronger, more exaggerated reactions that were never demonstrated in the input data. As Liao explains, this type of control matches the way artists and directors naturally ask for a motion to be a little more or a little less expressive, translating subjective creative direction into a precise numerical dial.

Under the hood, the framework builds on diffusion-based motion generation, the same family of generative techniques that has transformed image and video synthesis. Rather than relying on fixed style labels, the system learns a direction in a learned motion-style embedding space along which motion changes between the neutral and stylized endpoints. In practical terms, the model discovers the abstract axis that separates an ordinary walk from a happy walk, and then lets users travel along that axis by any desired amount. Information representing this style-change direction is combined with information about the motion content and the user-specified slider value before being fed into a pretrained motion-generation model. This architecture separates what the character is doing from how expressively they are doing it, allowing the two factors to be manipulated independently without retraining the underlying generator.

Evaluating a system that promises smooth, continuous style control requires more than a single qualitative demo. The team tested their method on multiple benchmark motion datasets and created an additional dataset specifically designed to assess performance on styles beyond those seen during training, a rigorous test of generalization. They compared their approach against established techniques, including the Multi-condition Motion Latent Diffusion Model and DeepMotionEditing. According to Liao, the proposed strategy achieved favorable results in the consistent control of style intensity, smooth transitions between intensity levels, and extrapolation to highly expressive motion. In other words, the slider does not just produce plausible intermediate motions; it produces them in a predictable order, so that increasing the value reliably increases the expressive intensity rather than jumping unpredictably between different performances.

Consistency and smoothness matter enormously in production. An animator who sets the slider to 0.5 needs to trust that the result sits meaningfully between the neutral and fully stylized endpoints, and that nudging the value to 0.6 will produce a correspondingly subtle change rather than a jarring shift in the character’s demeanor. Without that monotonic relationship between the slider and the perceived intensity, the control interface would be meaningless no matter how technically sophisticated the underlying model. The endpoint-supervised design directly encodes this expectation into the learning process, which is what distinguishes the Motion Style Slider from prior style-transfer approaches that treat stylization as a single, indivisible transformation.

To validate that the outputs actually read as intended to human eyes, the researchers conducted a user study using a Likert-scale questionnaire with 11 university participants. The participants rated the differences in style intensity among motions generated by the Motion Style Slider as clearly distinguishable, while the naturalness of the motions remained comparable across intensity levels. This second finding is just as important as the first. A system that achieves fine-grained intensity control at the cost of robotic or unnatural movement at intermediate settings would be of little use to animators, since fixing unnatural motion would reintroduce exactly the manual labor the tool is meant to eliminate. The study suggests the framework preserves motion quality across the entire slider range.

The practical implications extend across the entertainment industry. In video game development, where characters must perform countless actions under tight production schedules, the ability to derive a full spectrum of expressive variants from just two captured motions could dramatically reduce motion-capture budgets and iteration time. In film and virtual production, directors could fine-tune a character’s emotional delivery in real time during review sessions, asking the system for slightly more restraint or slightly more energy and seeing the adjustment immediately. Beyond games and film, the same technology could benefit virtual influencers, social VR platforms, embodied agents, and any application where digital humans need performances that feel tailored rather than templated.

There is also a broader conceptual significance to the work. The Motion Style Slider demonstrates that meaningful continuous control can be extracted from sparse, endpoint-only supervision, challenging the assumption that fine-grained generative control requires densely labeled training data. By learning a direction in an embedding space rather than memorizing discrete style categories, the approach points toward generative systems that behave less like black-box oracles and more like creative instruments with responsive, understandable controls. For a field often criticized for producing outputs that artists cannot steer, that shift in philosophy may prove as influential as the specific technical achievement. As AI-generated animation becomes an ever-larger part of games, films, and virtual content, tools like this one suggest a future in which the expressive nuances of digital characters obey the creative vision of the people directing them, one slider position at a time.

Subject of Research: Continuous AI-based control of stylized human motion intensity for character animation

Article Title: Motion Style Slider: novel framework enabling fine control of nuances in character motion

Article References: Motion Style Slider: novel framework enabling fine control of nuances in character motion. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: Motion Style Slider, character animation, motion generation, diffusion models, style transfer, motion capture, ECCV 2026, Science Tokyo, Cygames, computer graphics, AI animation, style intensity control

The dataset is described as 4D because it records three-dimensional information over time, allowing researchers to study not only what humans and dogs look like during an interaction, but how their bodies move through space second by second. InterPet4D comprises 6.83 million synchronized frames drawn from 161 recording sessions involving 23 human participants and 13 dogs representing 11 breeds. Twelve synchronized third-person cameras captured the interactions from multiple viewpoints, while an egocentric camera recorded the scene from the human participant’s own perspective. Aligned audio and text captions accompany the visual data, and the recordings were processed to estimate human body and hand motion alongside the dog’s 3D pose.

Collecting such data is harder than it might sound. Close-range interactions between a person and a dog frequently cause one participant to occlude the other, blocking the camera’s view and complicating detailed motion reconstruction. Preserving the timing and context of each exchange, which is essential if a machine is to learn the causal link between a cue and a response, requires precise synchronization across many sensors. By combining multi-view video, egocentric video, audio and 3D motion in a single curated resource, the team aimed to overcome both problems at once. “By capturing human–dog interactions, we wanted to create a dataset that reflects their multimodal nature and supports more systematic research,” says Yichen Peng, Specially Appointed Assistant Professor in the Department of Computer Science at Science Tokyo, who led the work with Professor Hideki Koike.

The interactions in InterPet4D are organized into four categories: petting, commanding, calling, and free-form activities such as fetch, tug-of-war and chase. This structure reflects the reality of human–dog communication, in which physical touch, hand gestures, verbal commands and distance calls each elicit different changes in the dog’s posture, position and movement. Having labeled categories spanning both contact-based and distance-based interaction gives machine learning models a way to learn how different types of human cues map onto different canine responses, something no previous public dataset has offered at this scale.

The dataset served as the foundation for a second contribution: InterPetMoGen, abbreviated IPMG, an AI framework that generates plausible dog motion conditioned on human body and hand movements and accompanying audio cues. The system first converts complex motion sequences into compact motion tokens, discrete representations that neural networks can process efficiently. Generation then proceeds with an autoregressive transformer, which produces the dog’s movement sequentially while attending to the context of the ongoing interaction, so that each predicted frame is consistent with both the human cue and the motion generated so far.

Two further components shape the quality of the output. A PetVAE module, a variational autoencoder, learns a compact latent representation of dog movements, effectively compressing the enormous variety of canine motion into a smooth space from which realistic sequences can be sampled. Meanwhile, modality-aware attention allows the model to weigh how strongly each input type, body motion, hand motion or audio, should influence the generated response at each moment. A spoken command might dominate during a calling interaction, whereas hand trajectories matter more during petting, and the attention mechanism lets the network learn those differences from data rather than through hand-crafted rules.

The researchers evaluated the generated motion using established metrics. Fréchet Inception Distance, or FID, measures how closely the distribution of generated motion resembles real motion, with lower values indicating greater realism. Retrieval Precision assesses whether generated dog movements actually correspond to the human cues given to the model, and diversity scores capture the variety of movements produced. IPMG achieved a kinetic FID of 11.21, compared with 21.22 for a Seq2Seq-Transformer baseline, a 47.2 percent reduction. It also improved hand-motion alignment from 0.41 to 0.63 and body-motion alignment from 0.38 to 0.59, while increasing motion diversity from 5.01 to 5.93.

Human judgment reinforced the quantitative results. In a user study, 12 participants rated the naturalness and appropriateness of the generated dog responses on a seven-point scale. The full IPMG model received scores of 6.58 for naturality, 6.55 for responsiveness and 6.63 for overall quality, compared with 4.04, 3.67 and 3.67 respectively for the Seq2Seq-Transformer. “The results suggest that modeling different modalities and broader interaction context can help AI-based frameworks predict and generate dog movements that are both more realistic and consistent with human cues,” Peng says. In other words, an AI that listens to the voice, watches the hands and tracks the body produces dogs that behave like dogs, rather than generating generic four-legged animation.

The implications extend well beyond the laboratory. A standardized foundation for human–pet interaction research could support behavioral analysis, helping scientists quantify how dogs respond to different handlers, commands and environments. In animation and games, the technology points toward virtual animals that react believably to player gestures and speech instead of following scripted paths. In robotics, plausible dog-motion models could inform socially aware machines that interpret and respond to animal behavior, a capability relevant to service animals, veterinary settings and human–animal interaction studies. The dataset’s multimodal design, pairing video, audio, text and 3D pose, makes it a reusable benchmark for any system that must understand two very different bodies acting in concert.

The authors are candid about the current limits. The present work focuses exclusively on dogs, does not model the physical forces involved in contact interactions such as petting or tug-of-war, and generates fixed-length clips of ten seconds. Longer sequences, richer physical simulation and extension to other species are outlined as future directions. Even so, the arrival of InterPet4D marks a shift in how machines can learn from the oldest partnership between humans and animals: for the first time, the everyday language of gestures, commands and wagging tails has been recorded in enough detail, and in enough quantity, for artificial intelligence to begin speaking it back. The findings were presented at the 19th European Conference on Computer Vision (ECCV) 2026, one of the leading international conferences in computer vision, held on September 9, 2026.

Subject of Research: Multimodal 4D datasets of human–dog interactions for AI-based pet motion generation

Article Title: InterPet4D: capturing human–dog interactions to generate AI-based pet motion generation

Article References: InterPet4D: capturing human–dog interactions to generate AI-based pet motion generation. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: InterPet4D, human–dog interaction, motion generation, artificial intelligence, computer vision, 3D pose estimation, multimodal dataset, autoregressive transformer, variational autoencoder, animal behavior, ECCV 2026, Science Tokyo

Now, a research team at Institute of Science Tokyo, working with a colleague at Kyushu University, has developed a nanoscale platform that converts hydrostatic pressure directly into fluorescence, offering a way to read pressure optically in aqueous and biological environments that were previously out of reach. The platform is built from pyrene-modified polyionic complex vesicles, nicknamed Pyr-PICsomes, and its most striking feature is that the pressure response can be programmed by tuning the stiffness of the vesicle membrane. The study, led by Assistant Professor Hayato L. Mizuno and Associate Professor Yasutaka Anraku of Institute of Science Tokyo together with Professor Gaku Fukuhara of Kyushu University, was made available online on August 15, 2026, and published in Volume 9, Issue 36 of the journal ACS Applied Nano Materials on September 11, 2026.

The building blocks of the platform are PICsomes, a class of polymer-based vesicles that self-assemble spontaneously in water from oppositely charged polymers. Because they form through electrostatic attraction rather than hydrophobic self-association, PICsomes are inherently water-compatible, which immediately addresses one of the major shortcomings of earlier molecular pressure probes. In this work, the researchers chemically crosslinked the vesicle membranes using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide, commonly abbreviated as EDC. By varying the concentration of this crosslinking agent, the team could dial in the mechanical stiffness of the membrane while keeping the vesicle size constant at approximately 100 nanometers in diameter, a scale small enough to disperse readily in complex fluids.

Into these membranes the researchers incorporated pyrene, a flat, aromatic fluorescent molecule that serves as the actual pressure-sensitive element. The sensing principle rests on a well-understood photophysical behavior of pyrene. When pyrene molecules sit far apart from one another, they fluoresce mainly as isolated units, called monomers, emitting light at around 380 nanometers, in the near-ultraviolet part of the spectrum. When hydrostatic pressure compresses the vesicle membrane, neighboring pyrene molecules are pushed closer together, allowing temporary pairs known as excimers to form. These excimers emit at a distinctly longer wavelength, around 500 nanometers, in the visible blue-green region. Because the two emissions are spectrally separated, the ratio of excimer to monomer fluorescence provides a direct, ratiometric readout of the surrounding pressure.

What elevates the work beyond a simple demonstration is the discovery that membrane stiffness acts as a control knob for this response. The softest vesicles, with a membrane stiffness of 7.3 piconewtons per nanometer, showed the largest change in the excimer-to-monomer fluorescence ratio across a pressure range of 0.1 to 50 megapascals. Their sensitivity reached 0.28 per megapascal, whereas the stiffest vesicles, at 39 piconewtons per nanometer, managed only 0.02 per megapascal. In other words, a softer membrane transmits the compressive force of the surrounding fluid more effectively to the embedded pyrene molecules, increasing the likelihood of excimer formation and amplifying the optical signal. The amount of pyrene loaded into the membrane mattered as well: more pyrene molecules led to greater excimer formation and correspondingly higher pressure-detection sensitivity.

While probing how pyrene influenced the excited-state dynamics of the vesicles, the team identified a second, independent sensing mode based on fluorescence lifetime, meaning the average time a fluorescent molecule remains in its excited state before emitting a photon. Remarkably, this mode favors the opposite end of the stiffness spectrum. The stiffer vesicles showed stronger pressure-dependent changes in fluorescence lifetime, with the stiffest vesicles exhibiting a lifetime sensitivity of −0.09 nanoseconds per megapascal across the 0.1 to 50 megapascal range. This behavior makes the rigid vesicles attractive candidates for fluorescence lifetime imaging microscopy, or FLIM, a technique widely used in biological research because lifetime measurements are largely independent of probe concentration and illumination intensity.

The practical implication is that there is no single optimal design; instead, the membrane should be tailored to the intended measurement method. Soft vesicles excel when pressure is read from fluorescence intensity ratios, while stiff vesicles are better suited to lifetime-based imaging. As Mizuno explained, an important aspect of the platform is that it does not rely on a single sensing mechanism. By changing the membrane stiffness, the researchers can access different fluorescence readouts, providing flexibility in how pressure is measured. This tunability, Mizuno noted, allows the sensing mode to be tailored to different environments and measurement methods, opening possibilities for studying pressure in complex aqueous systems.

The team also tested how the platform would hold up outside the controlled conditions of the laboratory, and the results point toward genuinely demanding applications. The covalent crosslinking of the membrane helps the vesicles maintain their structural integrity in saline conditions, such as seawater or physiological salt concentrations. Meanwhile, the hydrophilic outer layer of the vesicles helps protect the membrane environment in biological fluids, shielding the embedded pyrene probes from direct interference. On the basis of these properties, the authors propose that Pyr-PICsomes could eventually be used to investigate localized pressure in tissues, cell cultures, ex vivo samples, and even deep-sea organisms, environments where conventional sensors cannot operate and where optical readout through a microscope or imaging system is the only practical option.

Beyond the immediate sensing application, the study establishes a broader materials-design strategy: membrane mechanics can be used to program optical response behavior. Rather than treating the mechanical and photophysical properties of a nanomaterial as separate concerns, the researchers demonstrated that deliberately engineering one, in this case the crosslink density of a polymer vesicle membrane, directly controls the other. This coupling turns a simple self-assembled structure into a programmable optical device whose behavior is written into its physical architecture before it ever encounters a stimulus.

The resulting platform is versatile, water-soluble, biocompatible, robust, and tunable, a combination that few pressure-sensing approaches can claim simultaneously. It lays the foundation for smart materials capable of detecting hydrostatic pressure in complex and inaccessible environments, with potential uses spanning oceanography, cell biology, and biomedical research. The work was published in ACS Applied Nano Materials, and the authors declare no competing interests. As pressure-dependent phenomena continue to attract attention across disciplines, from the mechanics of living cells to the physics of the deep sea, tools like Pyr-PICsomes suggest that the answers may come not from harder instruments, but from softer, smarter materials designed to glow under pressure.

Subject of Research: Programmable pyrene-modified polyionic complex nanovesicles for fluorescence-based hydrostatic pressure sensing

Article Title: Pyr-PICsomes: novel programmable nanovesicles for hydrostatic pressure sensing

Article References: Pyr-PICsomes: novel programmable nanovesicles for hydrostatic pressure sensing. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: nanovesicles, hydrostatic pressure sensing, PICsomes, pyrene, fluorescence, excimer, membrane stiffness, FLIM, polyionic complex vesicles, ACS Applied Nano Materials, Science Tokyo, smart materials

The logic behind the research begins with a simple economic reality. Large geostationary communications satellites can deliver enormous capacity, but they are expensive to build and costly to launch. Small satellites, by contrast, are cheap, light, and increasingly capable, which makes it practical to fly them in constellations: hundreds or thousands of spacecraft working in concert to blanket the globe with coverage. Such constellations promise affordable high-speed connectivity for remote regions, ships at sea, aircraft in transit, and communities that terrestrial fiber infrastructure has never reached. Yet small satellites face a fundamental constraint. The volume and weight allowances on a launch vehicle are brutally limited, and conventional radio hardware, particularly the large antennas needed for high-gain directional communication, simply does not fit inside a compact spacecraft bus.

The solution pursued by the Science Tokyo team, led by Professor Hiraku Sakamoto of the Department of Mechanical Engineering, Associate Professor Atsushi Shirane of the Laboratory for Future Interdisciplinary Research of Science and Technology, and Professor Kenichi Okada of the Department of Electrical and Electronic Engineering, draws on the ancient art of origami. Foldable membrane antennas are thin, lightweight structures that can be stowed compactly during launch and then unfolded in orbit into a large radiating aperture. The concept is elegant on paper: the satellite gains a big antenna without carrying a big antenna through the atmosphere. But elegance collides with physics the moment the membrane deploys. Thin flexible structures do not spring into a perfectly planar shape. They warp, bend, and ripple, and those small geometric deviations have outsized consequences for a phased-array system.

A phased-array transceiver works by coordinating many individual radiating elements so that their signals combine constructively in one direction and destructively in all others, steering a focused beam electronically without any moving parts. The technique depends on exquisite synchronization: the relative timing and phase of each element’s signal must correspond precisely to the physical positions of the elements. When a membrane antenna deploys with even small non-planar deformations, the actual spacing between elements no longer matches the assumed geometry. The beam pattern degrades, sidelobes grow, and the carefully engineered directional performance collapses. This deformation problem has been the central barrier preventing foldable membrane arrays from graduating from laboratory curiosities to operational satellite hardware.

The Science Tokyo team’s breakthrough was to build a transmitter that measures and compensates for its own deformation. Their Ka-band, 16-element active phased-array transmitter was mounted across two separate boards on a deployable membrane and carried aboard the RAISE-4 small satellite, which launched on December 14, 2025 as part of the HELIOS-R project within JAXA’s Innovative Satellite Technology Demonstration-4 program. After the membrane unfolded in orbit, the system applied a calibration method previously validated in ground testing to estimate the actual bend between its two boards. With that estimate in hand, the transceiver electronically adjusted the signal timing of each antenna element, restoring the coherent coordination that the warped geometry would otherwise have destroyed.

The in-orbit experiments, which began on March 13, 2026, delivered unambiguous results. Directional beam control operated correctly both independently on each transmitter board and simultaneously across both, and the calibration scheme successfully compensated for the membrane’s non-planar shape after deployment. In practical terms, the satellite demonstrated that a folded, flexible, imperfectly deployed antenna could still deliver the precise electronic beam steering that modern satellite communications demand. The team also examined how transmitter temperature affected communication performance, gathering flight data that will inform the design of future systems expected to endure the thermal swings of the space environment, where surfaces cycle between intense sunlight and frigid eclipse every orbit.

The significance of the demonstration extends well beyond a single successful experiment. By showing that a high-performance phased-array transceiver can be reliably packaged inside a small satellite, the work establishes a foundation for scaling up to larger deployable arrays. Larger apertures mean higher gain, stronger links, and greater data throughput, which are exactly the qualities next-generation satellite networks require. The architecture validated on RAISE-4 suggests a path by which constellations of modest, inexpensive spacecraft could collectively deliver the kind of capacity and coverage traditionally reserved for their giant, costly counterparts, bringing genuinely global broadband coverage a step closer to reality and promising better connectivity for underserved regions that have waited decades for it.

Sakamoto and his colleagues see the work as a catalyst for the coming generation of wireless infrastructure. “We expect our results to accelerate the adoption of satellite constellation-based wireless communication systems in the 6G era, helping realize a society in which anyone can connect anytime and anywhere,” Sakamoto said. The statement captures the broader vision driving the field: a future in which the boundary between terrestrial and space communications dissolves, and connectivity follows people rather than the other way around. Direct-to-device services, resilient disaster communications, and ubiquitous broadband all depend on the kind of compact, deployable, electronically steered antenna technology that this demonstration has now proven in the harshest test environment available: space itself.

The project also carried a human dimension that the researchers regard as central to its legacy. More than 20 students from Science Tokyo participated in the development of the space-demonstration system, gaining hands-on experience in every stage of designing, building, and operating flight hardware. “The project therefore contributes not only to technological advancement but also to the education and training of future professionals in the increasingly important fields of space technology and wireless communications,” Sakamoto concluded. As the space economy expands and the demand for satellite communications expertise intensifies, that pipeline of trained engineers may prove as consequential as the technology itself.

From origami-inspired membranes to self-calibrating electronics, the RAISE-4 experiment weaves together mechanical engineering, radio-frequency design, and orbital operations into a single working system. The research, published in the IEEE International MTT Symposia under the title “On-Orbit Demonstration of a Deployable Ka-Band 16-Element Active Phased-Array Transmitter,” marks the moment deployable phased-array technology moved from promise to proof. If the vision holds, the satellites that carry humanity’s 6G traffic may well unfold in orbit like paper sculptures, quietly bending to the realities of space while keeping their beams locked on the users below.

Subject of Research: In-orbit demonstration of a deployable Ka-band phased-array transceiver with non-planar deformation correction for small satellite communications

Article Title: Wireless communications breakthrough: a deployable phased-array transceiver for small satellites

Article References: Wireless communications breakthrough: a deployable phased-array transceiver for small satellites. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: small satellites, phased-array transceiver, deployable membrane antenna, origami antenna, 6G, Beyond 5G, satellite constellations, RAISE-4, JAXA, Ka-band, beam steering, Science Tokyo

The research team, led by doctoral student Toranosuke Horikawa of the School of Engineering and Associate Professor Masatoshi Kondo of the Laboratory for Zero-Carbon Energy, has been developing what they call an alternative approach to resource extraction based on liquid tin. Their earlier work established the core principle: when desalination brine is brought into direct contact with hot liquid tin, water evaporates and can be collected as distilled water, while dissolved elements such as magnesium, sodium, calcium, and potassium are absorbed into the liquid metal. In effect, the molten tin acts as both a heat-transfer medium and a chemical sponge, separating the valuable components of brine in a single integrated operation.

But the earlier experiments left a critical problem unresolved. Alongside the useful metals, chlorine and sulfur from the brine were also incorporated into the liquid tin. These nonmetallic contaminants react with magnesium to form compounds such as magnesium sulfate, which would sharply limit the industrial usefulness of any recovered material. Magnesium is a strategically important element, used in automobiles, aerospace components, batteries, cement, and refractory materials, and conventional recovery methods typically demand large quantities of chemicals or electricity while generating secondary waste. A product contaminated with sulfur and chlorine would undermine the entire premise of the process.

The team’s solution was to add a vacuum-degassing step before magnesium recovery. In the experiments, approximately 8 grams of tin was melted and held at temperatures between 573 and 673 kelvin, roughly 300 to 400 degrees Celsius. Artificial brine, prepared with heavy water to enable isotope tracing during gas analysis, was supplied onto the liquid surface at a rate of 0.1 milliliters per minute. The water evaporated and was collected separately as distilled product, while magnesium, sodium, chlorine, sulfur, and other brine-derived components were taken up into the liquid metal below.

The brine-loaded tin was then heated under a high vacuum of about 10 to the power of minus 3 pascals, ramping from 573 or 673 kelvin up to 973 kelvin, approximately 700 degrees Celsius, at a controlled rate of 2.5 kelvin per minute, and held at the peak temperature for roughly ten minutes. Gas analysis revealed that the unwanted chlorine- and sulfur-containing species were released as volatile gases, including hydrogen chloride, deuterium chloride, molecular chlorine, sulfur dioxide, and hydrogen sulfide. By measuring the change in the mass of the liquid tin before and after treatment, the researchers estimated that about 59 to 67 percent of the brine-derived gaseous components had been removed from the metal.

With the contaminants largely stripped away, the next stage exploited a simple thermodynamic principle: the solubility of dissolved elements in liquid tin falls as the temperature drops. After degassing, the tin was slowly cooled while the vacuum was maintained, and as the solubility declined, dissolved elements precipitated out of the metal in a process the researchers describe as recovery by cooling, or solubility-driven precipitation. Microscopic observations showed that magnesium-rich and sodium-rich regions formed at different locations within the solidified tin, indicating that the elements could be spatially separated from one another during the cooling process itself.

The quantitative results were striking. When 20 milliliters of brine was processed at 673 kelvin, the magnesium concentration in the recovered precipitate reached up to about 280 times that of the original brine, and the ratio of magnesium to sodium increased by roughly a factor of 5,300. The amount of magnesium recovered was estimated to be equivalent to more than 99 percent of the magnesium that had been absorbed into the liquid tin. Electron microscopy and electron backscatter diffraction analysis further confirmed that the major crystalline phase in the magnesium-rich precipitate was magnesium oxide, MgO, a chemically simple and industrially relevant compound rather than a complex contaminated salt.

One of the most significant advantages of the approach is what it avoids. Because the process separates and concentrates magnesium without relying on large additions of alkaline chemicals, it may substantially reduce chemical consumption and the secondary waste streams associated with conventional precipitation methods. The thermal energy required to keep the tin molten could, in principle, be supplied by concentrated solar thermal energy or by high-temperature industrial waste heat, opening a pathway to low-carbon operation. A conceptual design developed by the team estimated that, under idealized conditions, the electricity required for the vacuum-degassing step could be as low as about 1.7 kilowatt-hours per kilogram of magnesium recovered.

To illustrate the potential scale, the researchers modeled a solar-driven module designed for a sunny, water-stressed region such as Egypt. In this scenario, a solar concentrator with a diameter of 15 meters was estimated to have the potential to produce about 970 kilograms of freshwater and recover approximately 2.8 kilograms of magnesium per day. While these figures come from an idealized design study rather than an operating plant, they sketch the outline of a technology that could turn the economics of desalination on its head, converting the largest liability of the industry into a co-product revenue stream while easing the environmental burden of brine discharge.

The researchers describe the broader concept as mining without digging: obtaining valuable resources not from newly excavated ores, but from water streams that are currently treated as waste. The vision is compelling, but the path from laboratory demonstration to practical deployment still requires substantial engineering. The next step, according to the team, is to develop a continuous system in which liquid tin circulates through the full cycle of freshwater production, vacuum degassing, magnesium precipitation and recovery, and tin reuse. Long-term stability of the molten metal, compatibility with structural materials at operating temperatures, purification of the recovered magnesium product, overall energy consumption, and economic feasibility will all need careful evaluation. If those hurdles can be cleared, the humble brine pipe at the back of a desalination plant may one day rank among the world’s more sustainable magnesium mines.

Subject of Research: Magnesium recovery from seawater desalination brine using liquid tin direct-contact distillation and vacuum degassing

Article Title: Mining without digging: liquid metal tin turns desalination brine into a magnesium resource

Article References: Mining without digging: liquid metal tin turns desalination brine into a magnesium resource. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: desalination, brine, magnesium recovery, liquid tin, vacuum degassing, magnesium oxide, water scarcity, solubility-driven precipitation, solar thermal energy, resource recovery, Science Tokyo, Distalination journal

Classical LDPC codes are the quiet workhorses of modern digital life. They protect data in 5G wireless links, satellite broadcasts, Wi-Fi signals, and solid-state storage drives. Their power comes from a mature and well-understood design theory. Engineers choose how many parity checks connect to each bit, retain just the right amount of randomness in the connection pattern, and carefully avoid short cycles in the graph that describes the code. When these ingredients are balanced correctly, the resulting codes achieve two prized properties simultaneously: a large minimum distance, which guarantees that many errors must accumulate before information is corrupted, and a threshold phenomenon, in which the decoding failure rate drops sharply once the noise falls below a predictable level. This threshold behavior is what makes LDPC codes approach the theoretical limits of communication.

Quantum LDPC codes face an additional constraint that has no classical analogue. Because quantum mechanics forbids the direct measurement of the encoded information, two different families of error checks, one for bit-flip errors and one for phase-flip errors, must be made orthogonal to each other so that they do not interfere when both are measured. In previous constructions, imposing this orthogonality requirement across the entire design process tended to destroy the very features that make classical LDPC codes work so well. Short loops crept back into the structure, connection degrees became difficult to control, and designers were forced to choose between a good minimum distance and a good threshold, rarely achieving both at once.

Kasai’s insight was to ask whether the orthogonality constraint truly needs to govern every part of the design. His answer was no. In the construction described in his paper, titled Breaking the Orthogonality Barrier in Quantum LDPC Codes, he built the code from affine permutation matrices and applied the orthogonality condition only to the selected, or active, rows that are actually used for error correction. The complementary, or latent, parent rows remain free of the quantum constraint and retain their randomness. This separation means that connection degrees, randomness, and the avoidance of short loops can all be engineered according to classical LDPC principles, while the quantum-mechanical requirement is satisfied exactly where it matters.

The quantum constraint is essential, but it does not have to govern every part of the design, Kasai explains. By applying it only where error correction uses it, the classical LDPC design freedom needed to pursue both a large minimum distance and threshold behavior can be preserved. The resulting code is a (3,12)-regular construction with girth 8, meaning the shortest loops in its graphical structure contain eight edges, and it is written in standard quantum notation as [[9216,4612,d]] with d at most 48. That notation means the code protects 4,612 logical qubits using 9,216 physical qubits, a ratio of nearly one logical qubit for every two physical qubits. For comparison, the surface codes used in many current hardware proposals typically demand hundreds or thousands of physical qubits per logical qubit, so a high-rate code of this kind could dramatically reduce the hardware overhead of fault-tolerant quantum computing.

The evidence for a large minimum distance is particularly strong. Weight-48 logical operators can be constructed explicitly, and the X-type and Z-type distances associated with the latent structure of the code are both exactly 48. Detailed searches and simulations at low error rates found no logical errors of lower weight anywhere in the code. Although a rigorous global lower bound on the minimum distance remains an open mathematical question, these results together provide compelling evidence that the overall minimum distance sits near 48. In addition, the deliberate elimination of 4-cycles and 6-cycles reduces the trapping sets, those troublesome small structures in the code graph that can stall belief-propagation decoding and cause otherwise correctable errors to slip through.

Decoding performance was evaluated using belief propagation with low-complexity post-processing. The code exhibits a clear decoding waterfall that sits close to the density-evolution prediction for the corresponding classical, non-orthogonal random (3,12)-regular LDPC ensemble. That classical benchmark, which describes the best a belief-propagation decoder could do on an idealized infinite version of the code, is approximately 0.05702, and Kasai emphasizes that this figure is a reference point rather than a measured threshold of the quantum code itself. With post-processing, the quantum code achieved a frame error rate of 10 to the power of minus 8 at 4 percent depolarizing noise, corresponding to roughly one decoding failure per 100 million trials. The significance, Kasai notes, is not merely that one high-rate code performs well, but that threshold behavior, minimum distance, short loops, and hard-to-decode error patterns can now all be considered within the same design framework that classical coding theory spent decades perfecting.

Classical LDPC coding has spent decades learning how to predict decoding limits from connection degrees and to design codes that approach those limits, Kasai explains. Seeing the quantum code’s waterfall approach the same classical benchmark suggests that this predictive methodology can now guide quantum LDPC design as well. He cautions, however, that the present results rest on theoretical analysis and numerical experiments, and that performance on a specific quantum processor must be evaluated separately before any hardware claims can be made.

The wider community has not waited for that hardware evaluation. Kasai released a preprint on arXiv on January 13, 2026, and while the paper was still under review, researchers at Harvard University, MIT, and QuEra Computing adapted the construction to reconfigurable neutral-atom quantum computers. Their paper, posted on April 17, designs codes and error-detection procedures around atom rearrangement using acousto-optic deflectors, and refers to the resulting family of codes as Kasai codes. QuEra highlighted the development in official posts, including one titled Kasai Code Breakthrough in Quantum Error Correction. Kasai’s paper was accepted by Quantum on August 3, 2026, and between February and July he presented the work in eleven invited talks at universities, companies, and international workshops, including QID2026 at the Korea Institute for Advanced Study in Seoul.

Independent teams have continued to extend the framework. A group including Tsinghua University’s Center for Quantum Information posted a family of Cornucopia codes on August 3, 2026, emphasizing regular structures suited to simultaneous atom movements, and Willers Yang and colleagues posted GALA codes on August 7, 2026, offering a general framework that recovers the earlier rate-1/2 Kasai constructions as special cases while incorporating atom movements and logical operations directly into the code design. Further invited presentations are scheduled at LG Electronics, Kyoto University’s Yukawa Institute for Theoretical Physics, a superconducting quantum computing conference in Italy, CWI in Amsterdam, RIKEN’s Center for Quantum Computing, and the Joint Mathematics Meetings in Chicago in January 2027. As Kasai concludes, independent teams adapting the same design principles to hardware layouts and more general code families suggests that these codes are becoming a common foundation for international work on scalable quantum error correction, a development that could bring fault-tolerant quantum computers significantly closer to practical reality.

Subject of Research: Quantum low-density parity-check error-correcting codes designed using classical LDPC design theory

Article Title: Bringing classical LDPC code design theory to quantum computers

Article References: Bringing classical LDPC code design theory to quantum computers. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: quantum error correction, LDPC codes, Kenta Kasai, Science Tokyo, minimum distance, belief propagation, neutral-atom quantum computing, encoding rate, density evolution, Quantum journal, trapping sets, fault tolerance