Three selections from the previous 20 ideas, to be turned into abstracts and (~hypothetically) submitted to a peer review journals:
1. To be submitted to Computational Aesthetics which I guess is actually a conference rather than a journal, but… It could also work for Eurographics:
Trend Montage: Automated Portraits of Popular Culture
This work is a study in our communal interests as a society. It is a two part program which first collects images of the most popular search terms for a given day as taken from Google’s public U.S. trend statistics, and then combines these collected images into a collage. Face detection, region of interest extraction, and other image processing techniques are used in order to achieve visually appealing compositions without the input of any human guidance. It is therefore a relatively objective way of portraying our collective thoughts. When viewed in a series, these collages literally visualize our popular concerns- the stories and characters we pay attention to, trends we follow, celebrities we idealize, foods we crave, commodities we fetishize, etc. On a higher level, they also reveal otherwise invisible patterns and cycles in what we care about as a community.
As an exhibit, this work will display the creation of these collages in real-time. It’s a very visually compelling process which takes somewhere between 20 to 30 minutes to finish, flashing face-detection and ROI extraction steps and positioning of images on the larger canvas as it happens in the program itself.
2. To be submitted to ICMC (International Computer Music Conference)
Enhanced Decorrelated Audio Upmixing Through Spatialization
Traditional chorusing/upmixing techniques tend to suffer from phase cancellation and/or unnatural delay-time predictability, both of which subtract from the illusion of multiple sources. This makes much it almost impossible to create a believable, dense, soundscape from relatively sparse input material. Many times however, stemming from either practical limitations or the desire to create very specialized sonic scenes, upmixing would be a quite invaluable tool were it not for these inherent difficulties. This project is an attempt to diminish these problems through the combined use of spatialization, moving delay lines, and filtering. When coupled properly, these techniques can result in a much more “realistic” or believable result which in turn suggest the possibility of many practical applications for music and sound design.
3. To be submitted to IHC. (They limit the Abstract to 100 words.) As far as journals, I think after I completed the project it would work for Artificial Intelligence or Computer graphics and image processing depending on how I slanted it.
Algorithmic Portraiture: Adaptive Online Beauty Crawler as a Model for Genetic Competition
This Project is an attempt to simulate and visually depict the process of genetic mutation through the use of machine-learning algorithms. Multiple agents or bots will be programmed to seek out a particular image representative of an “ideal” human form. Attempting to find this photo, they will travel through their environment, the entire Internet, seeking “attractive” close matches. During this process, these agents will “breed” with pictures they like, creating composites, or mixtures of the new image and all previously collected material (genes). It will be a work of art, a novel computer science application, and a unique tool for visualizing the process of organic reproduction.