180 likes | 272 Views
Simple and Intuitive Visualization with the Web Automation and Translation Toolkit (WATT) Client. Minnesota Supercomputing Institute Undergraduate Internship Program Adam R. Momsen in collaboration with Professor David Yuen and the Virtual Laboratory for Earth and Planetary Materials (VLAB).
E N D
Simple and Intuitive Visualization with the Web Automation and Translation Toolkit (WATT) Client Minnesota Supercomputing Institute Undergraduate Internship Program Adam R. Momsen in collaboration with Professor David Yuen and the Virtual Laboratory for Earth and Planetary Materials (VLAB)
Project Goals • Create an online service for easy data visualization • Support fast development updates to limit service downtime • “Write once, install anywhere” using Java Webstart
WATT Client - Introductionhttp://webis.msi.umn.edu:8080/WattClientWebApp/start_page.jsf Consists of two parts: • Manipulator Window • Swing GUI Current start page upon WATT Client launch
How the Client Works Send Update Request Returns new “snapshot” Setup allows user to focus on data manipulation while server does the heavy lifting
Manipulator Window • Manipulator box based on GLEEM (“GL Extremely Easy-to-use Manipulator) by Kenneth B. Russell • Clicking corners scales, outside handles rotate, faces translate • JPEG “snapshot” of 3D rendered data mapped to plane within manipulator
GLEEM Classeshttp://alumni.media.mit.edu/~kbrussel/gleem/ Class list gives an idea of the complexity of graphical programming; in this case, just about every part of the manipulator has its own class! List only includes manipulator classes, transformers and viewers omitted • Manip • ManipManager • ManipMotionListener • ManipPart • ManipPartCube • ManipPartGroup • ManipPartLineSeg • ManipPartSquare • ManipPartTransform • ManipPartTriBased • and more!
Rotating the Manipulator • Source of earliest headaches: matching server and client camera and actor rotations - Server renders images using Visualization Toolkit (VTK), while client uses JOGL • Rotating object same as changing camera, but more computationally expensive • And always remember to agree on degrees or radians!
Scaling and Translating • Scaling - image and box are simply stretched and compressed as the user wills. • Translating - Camera panned freely along a plane parallel to initial JPEG orientation (gives illusion of moving box object)
JPEG Image To ease transmission, only a 2D image is displayed on the client, not a fully rendered 3D object. This made matching up the image and the manipulator difficult, and had to be addressed when rotating the client camera.
Java Swing Interface • Adding a Menu Bar will allow access to developmental and function information and commands to load data. • Input fields allow for manipulation of red cutting plane (visible in previous slide) • JButton allows for quick updates • Note: GUI functionality still in development
Maven 2 • Build structure and program (similar in idea to Ant, Java’s version of Make) - Sets up a generic project file structure that needn’t be rewritten for new projects - Standardizes project for collaborative development - Compiles, tests, builds, installs, deploys - Remote deployment allows new versions to be installed to service URL in a matter of seconds
Java Webstart • Central to “write once, install anywhere” motto • Service requires no software to be installed prior to use • User only has to accept, and Webstart will search for and download all the files the service needs to run
Challenges • Initial confusion on project purpose • Box perspective while translating • Image quality issues at high magnification • Matching rotations • Integrating GUI to render window • “No jogl.jar on classpath” error • Hit Points (detecting mouse clicks) • Switching to spherical handles • Point detection determines transformation type
Looking ahead • Embed service into webpage with AJAX - Allows for real-time image manipulation without need for constant page updates • Xith3D integration - JOGL overlay that may improve performance • Color map cutting plane • Low-resolution subset - Allows user to better predict final orientation of image
What I Learned • Graphical programming in OpenGL • Java Webstart • Maven large-scale project builds for collaborative efforts (restrictions on project source) • SVN versioning for group collaboration • Professional programming project experience
Acknowledgements • Professor Dave Yuen’s group: - Evan “Farmer” Bollig - Shuo “Mark” Wang - Martin “Frank Linux” Lyness - Sarah Wieber - Matthew Broten - Grace Schwab - Javier Roman - Paul “Pau Yensen” Jensen - Dan “Rao” Kigelman - Pedro Silveira - Cesar Da Silva
References • http://www.swiftless.cdadc.com/tutorials/opengl/opengltuts.html - OpenGL tutorials • http://www.opengl.org/code/category/C23/ - more OpenGL source • Davis, Tom, Jackie Neider, Dave Shreiner, and Mason Woo. “OpenGL Programming Guide”. Addison-Wesley, Boston. 1999. - “The Red Book” • http://xith.org/tutes/GettingStarted/html/deploying_xith3d_games_with.html - Xith3D documentation • http://www.flipcode.com/articles/article_frustumculling.shtml - Rendering and culling reference • http://csis.pace.edu/~bergin/sol/java/gui/ - Java Swing example • http://gorerle.com/vlab-wiki/index.php?title=User:Adamo - my VLAB Wiki page