October 9, 2024 HPC Creates Art of HPC Posters Scientific Visualization Share this page: Twitter Facebook LinkedIn Email By SC24 Communications With scientific datasets growing in size and complexity, organizing and visualizing them is becoming increasingly challenging. This is especially true when multiple variables in the output need to be studied and represented. With so much data being collected and processed, traditional visualization methods often fail to capture its full scope, and critical details are obscured, limiting scientists’ ability to derive accurate insights. We talked with researchers Francesca Samsel and Greg Abram from the Texas Advanced Computing Center (TACC), researchers behind the Sculpting Vis project, which is leveraging novel artistic methods to tackle this challenge. We learned that overcoming traditional visualization limitations requires both advanced technology and a creative, artistic approach. The Sculpting Viz Collaborative The project started when researcher Daniel Keefe, University of Minnesota realized that traditional scientific visualization methods are powerful but also limiting in certain ways. Scientists often work with immense datasets involving multiple variables that evolve over time. Keefe was frustrated as the tools available to visualize this complexity were often too rigid or abstract. Seeking a solution, he approached Samsel, a trained artist, to collaborate on expanding the visual language for complex data. “In art, we focus on components used to construct images – line, shape, form, color, and texture – all of which are essential for encoding data. That’s where we started,” explained Samsel, a research scientist at TACC who spent decades as a practicing (and teaching) artist before transitioning into the world of scientific visualization. Soon, the team grew as she recruited her husband, Greg Abram, a PhD computer scientist with more than 40 years of experience in large-scale data visualization. His technical expertise provided Samsel with invaluable direct input on her ideas. Together, they worked to bridge the gap between artistic vision and data complexity. A multivariate dataset visualization from the Sculpting Visualizations project, using varied glyphs and colors to represent different data variables, combining art and HPC for deeper scientific insights. “In art, we focus on components used to construct images – line, shape, form, color, and texture – all of which are essential for encoding data. That’s where we started.” “I am sort of the technical end of the project,” explained Abram. “Francesca brings the vision, and I focus on the guts of it, the data and how we can represent highly multivariate data in single images using advanced visualization techniques.” Together, they say their goal was not only to make data visualizations more accurate, but to humanize the process, making it more intuitive and accessible for a broader audience, including scientists and non-specialists alike. Overcoming the Challenges of Visualizing Multivariate Data One of the core challenges tackled by Sculpting Viz is the difficulty of representing multiple variables in the same space. In scientific visualization, multivariate volumetric data often lead to overlaps that obscure critical details. “It’s very difficult to represent multiple scalar values in the same space without obscuring them,” explained Abram. “You can look at endless cross-sections of data, but you won’t get a three-dimensional understanding of how the data is configured in real space.” “People have been working on color for a long time, but this was the first step toward applying a broader visual language to scientific data in a practical way.” The Artifact-Based Rendering (ABR) Library created by Francesca Samsel, featuring a collection of glyphs, lines, colormaps, and textures. This visual library expands the traditional scientific visualization toolkit, enabling more detailed and intuitive data representation. This “language” became the foundation for a new approach as Samsel got to work developing a rich “glyph vocabulary” that brought new details to the forefront, helping to co-locate multiple variables in a single visualization. This marked a significant step in expanding the scientific visual vocabulary beyond color. “People have been working on color for a long time, but this was the first step toward applying a broader visual language to scientific data in a practical way,” she said. An example of this can be seen in the couple’s work with climate data from Los Alamos National Laboratory. Researchers studying the Ronne-Filchner Ice Shelf in Antarctica needed to understand how different water masses, characterized by temperature and salinity, interacted beneath the ice. Scientists once examined slices of data, but new techniques allowed the Sculpting Viz team to reveal patterns and relationships in a single visualization. “By expanding the visual vocabulary beyond traditional methods, we’re allowing scientists to explore their data in ways they never could before,” said Samsel. “It’s not just about seeing the data. It’s about understanding the relationships and patterns that were previously hidden. That’s where the real breakthroughs happen.” Expanding the Data Canvas While introducing a new glyph vocabulary represented a significant step forward, the team did not stop there. The goal was to push the boundaries of scientific visualization even further, incorporating more artistic elements and expanding the data “canvas” to include texture, lines, and immersive environments like augmented reality (AR) and virtual reality (VR). These innovations allowed scientists to interact with their data in new, dynamic ways that traditional visualizations could not offer. “We started with glyphs, but we quickly realized that the world of visualization had much more potential,” explained Samsel. “In the art world, the possibilities for visual encoding, whether through lines, textures, or even movement, are endless. We saw no reason why the same couldn’t be applied to scientific data.” Perhaps one of the most groundbreaking advancements in Sculpting Viz has been the incorporation of immersive technologies like AR and VR, which allow scientists to explore their data in three-dimensional environments. As Abram described: “Instead of just viewing data on a flat screen, augmented reality lets you immerse yourself in the data. You can stick your head into an ocean model and interact with it in real time, touching specific points to trigger simulations or explore various layers.” A demonstration at the Texas Advanced Computing Center, where Francesca Samsel and Claire Fitch guide Michael Dell as he explores a 3D, multivariate visualization of the Gulf of Mexico using virtual reality and the Artifact-Based Rendering system. “In the art world, the possibilities for visual encoding, whether through lines, textures, or even movement, are endless. We saw no reason why the same couldn’t be applied to scientific data.” This immersive approach is transforming how scientists engage with their data. “When you can touch the data and see it evolve in response to your interactions, it’s a game-changer,” Samsel noted. “The idea is to bring the data into the world with us rather than keeping it confined to a flat screen.” Color has also become a key aspect to experiment with after the team realized the standard palettes used in scientific visualization limited the amount of detail scientists could see. Many commonly used color maps, such as the rainbow palette, introduced visual distortions that misrepresented the data. “The standard color map that’s been used for decades doesn’t take full advantage of the color spectrum, which means you lose a lot of detail and clarity,” explained Samsel. The team set out to develop richer, more nuanced color schemes that not only represented scalar values more accurately but also aligned with how the human brain interprets color. “Standard color maps can introduce visual artifacts – distortions that don’t represent real data but rather tricks of our visual system. By using more advanced color schemes, you eliminate those distortions and help scientists see the data as it truly is,” explained Abram about their results. Applications and Future Direction Despite being in its “alpha” phase, the Sculpting Viz Collaborative has already demonstrated its potential across a range of scientific fields, from climate modeling to neuroscience and even cosmology. In one example, they worked with ocean scientists exploring biofuel farming in the Gulf of Mexico. By representing the density of phosphates, nitrates, and other nutrients with glyphs, they provided a clear, three-dimensional view of where these resources were concentrated, helping researchers identify optimal locations for farming. In another example, they worked with neuroscientists tracking synapses and processes moving throughout the brain. ” “Each glyph represented a voxel, and the direction, color, texture, and cross-section of each glyph encode specific information about that voxel,” explained Samsel. “There are thousands of these glyphs in a single visualization, allowing scientists to see complex interactions in real time.” A multivariate visualization of brain activity, utilizing the Artifact-Based Rendering system. This detailed visualization uses custom glyphs and color mappings to represent various neural processes, providing scientists with an intuitive and immersive way to explore complex brain data. Looking ahead, the team’s focus is on refining and expanding their tools. As Abram explained: “While we’re still in the starting stages and this is not available for wide use, we’ve worked closely with an array of scientists and projects, and they’re all very supportive of the potential.” The pair explains that one of their primary goals is to compartmentalize the tool, allowing it to be used as a plugin for existing visualization systems. They want to enable scientists to engage these powerful, higher-definition visualizations with tools they already use. “We’re just scratching the surface of what’s possible,” said Abram. “As data continues to grow in complexity, the tools we’re developing will help scientists see deeper into their datasets, unlocking insights that could lead to major breakthroughs across countless fields.” Join Us in Atlanta Collaboration and continuous learning are key to realizing supercomputing’s full potential. SC24 offers an opportunity to expand your knowledge and enrich your experiences within the HPC community. Attendees engage with technical presentations, papers, workshops, tutorials, posters, and Birds of a Feather (BOF) sessions – all designed to showcase the latest innovations and practical applications in AI and HPC. The conference offers a unique platform where experts from leading manufacturers, research organizations, industry, and academia come together to share insights and advancements that are driving the future. You can also see Samsel and Abram’s work in the Art of HPC exhibition.Join us for a week of innovation at SC24 in Atlanta, November 17-22, 2024, where you can discover the future of quantum, supercomputing, and more. Registration is open!