Major: Computer Science
Project: EDUCE Digital Scroll Unwrapping
What is the best way to present the three-dimensional models of the Herculaneum scrolls? The scans first required cleaning up and resizing. These processes took lots of time. Once there we are given a reasonable amount of data to work with. The question being, “What is the best way to present this data?” VoreenVE provides a convenient interface to get various clear renderings of the data. However, when rendering animations the only options available are concerning the positioning of the model. The only way to create dynamic output is to use a ”Clock Processor” which runs independently of any animation processing. This causes the animations to be rendered asynchronously from the motion caused by the ”Clock Processor”. My solution is to create an ”Animation Processor” that acts similarly to the ”Clock Processor” but who’s timing is based off of the frame count in the animation rather than a clock. This allows the animations to be captured completely in sync with any dynamic motion. With this implementation in place I am able to make more complex animations that render seamlessly.
This experience with the Herculaneum scrolls has taught me a lot about computation and presentation. Having looked inside of the scrolls using digital rendering software I knew many of the different issues posed by these scrolls and have in the past months worked to overcome them. The largest of the issues, yet unresolved, is how to extract the layers from the scrolls in full. For now all the extraction has been done manually; tracing out each layer by hand. However, the process of rendering this data has shown to be a separate issue. For one, the total size of the volumes rendered must be able to fit in GPU memory. That means these 100GB scrolls must be quantized and re-sampled downwards to fit in 896MB. Given that the scrolls fit in memory there needed to be a proper tool for rendering animations. This meant learning to use a tool called VoreenVE. VoreenVE allows for various options in viewing digital volumes. The only problem was the lacking animation system which couldn’t perform operations in time with the recording. In order to fix this I added to Voreen’s source code to link the operations being performed to the frame count of the animation. Shortly thereafter, VoreenVE received an official update that centered around a much more in depth animation tool essentially obsoleting my changes. That being said, I opted to use my implementation as it’s simplistic nature is more fitting to my needs. The downside to that decision is that in not using the latest version of VoreenVE I am missing out on the other features added. Regardless of these downsides, I am able to produce pictures and relatively complex videos showcasing the scrolls with enough options to satisfy me aesthetically. Over all my time working with the Herculaneum scrolls and Voreen I have gained a lot more knowledge of how to render volumes to video on a high level and how to modify existing code on a lower level.