My original conception was to attempt to effectively render a story in a very "expert" way from the story that a student writes. This was a seriously imposing task because of the sheer size of the space of stories students could come up with. But rather than fret about the challenges of properly rendering the story in some wonderfully cinematic way from students' texts, why not treat our limited understanding and rendering capacity as a "feature". Rather than offering students the promise of well rendered movies based on their stories, why not quietly encourage them to test the capacity of our system and see what they can get it to render. In this way, students will be iteratively refining their stories to explore the system rather than refining their stories to simply make their stories better. In that way, we need to have a tight, rapid feedback loop that will encourage iterative writing and revising. Movie rendering needs to happen quickly. We also need to provide them with hints as to how they can get results so that they don't just get tired of playing with the system. More so, there should be mechanisms for students to record, bookmark and SHARE their stories+movies with their classmates. In this way, we can foster semi-collaborative exploration of the story space. Students could see what other students are writing and the effects of their writing on the resulting visual representation.
I am now much more excited (aka less intimidated) about this project, because it removes some of the burden of having comprehensive understanding and rendering technologies (which are probably currently impractical, even in the restricted domain). Rather than try to take on too much, lets leverage the strengths of the problem and existing technological shortcomings to our advantage.
