Bringing visual effects software to scientists

04.18.17 -

New website simplifies process of importing and rendering astrophysical data in 'Houdini.'

Rendering a 3D sun while it makes a coronal mass ejection used to be a task only doable by skilled scientific visualization programmers, like the National Center for Supercomputing Application's (NCSA) own Kalina Borkiewicz and AJ Christensen. Now, scientists can learn to make their own visualizations, thanks to a how-to website the two made with collaborator Dr. Jill Naiman at the Harvard-Smithsonian Center for Astrophysics. was established in response to a major breakthrough NCSA at the University of Illinois at Urbana Champaign implemented in the visual effects industry's staple program, Houdini. Previously, scientific data had to undergo complex processing to become editable in the program. Now, a Python tool developed by NCSA Research Scientist and Illinois Professor Matthew Turk takes care of the conversion, with a few commands and scripts Borkiewicz and Christensen lay out on the website, and in a recent paper.

"A large percentage of the time we spent on our last planetarium show was devoted to writing software and plugins for data handling," Borkiewicz said. "Each new data set from each new researcher in each new field required a different translator, and now this one tool reads many different kinds of data. We created the tool for ourselves, but are opening it up to people in the physical science community so they can make high-quality visualizations in fewer steps, without having to write new, custom code."

Borkiewicz said the website will also help astronomers communicate their science better. Rather than potentially waiting in a queue with one of the few visualization labs across the country, they can get the ball rolling themselves.

"This is one of the most streamlined ways we've seen to do high-quality graphics of scientific data, so much so that students and artists can use it," Naiman said.

The paper was accepted to Publications of the Astronomical Society of the Pacific on November 21 and was published on Tuesday. It walks novice visualization programmers through an outline of how they can make a production quality render from their data, check its accuracy and combine multiple observational and computational datasets to convey a scientific concept.

"The paper is meant mostly as an example, because the way you use the program is going to change from project to project. The website expands the user’s tool set with detailed tutorials that cover other workflows we and our colleagues have had experience with," Christensen said.

"Scientists are already distributing their findings through things like databases, publications and press releases. This tool and website helps them make their research even more intuitive through imagery and interactivity, and in a format that can be shared with students and artists," Christensen added.

The team is already planning updates to ytini, including a method to import more efficient, higher-resolution data into Houdini. As opposed to a simple grid with a data value at each cell, Adaptive Mesh Refinement (AMR) data has grids within grids, allowing scientists to have more resolution in the more interesting parts of the simulation.

"This allows you to be able to see the big picture of a whole galaxy, and within the same simulation, zoom in to see individual stars or star clusters," Naiman said. "Rather than having a single high-resolution grid, which would waste memory on empty space, the AMR data format is very efficient and only has resolution where it’s needed."