I just had an ugly realization about abstract graphics: how will they handle tales? A tale is a sequence of causally linked substories, and it could be of any length. This creates a problem in that we can no longer view the abstract graphics display as a single substory; it must be able to handle multiple displays. Now, the obvious solution is to create a user-controlled sequence of single frames, but I fear that this will be tedious. We need something more compact. This implies that we must shrink the verb representations to fit more inside the display frame. It also means that we'll have to expand the size of the display frame. Should we perhaps dismiss completely any notion of a textual representation? I think so.
Getting back to the main theme, I am more and more thinking in terms of a comic strip, a sequence of panels, one panel for each verb. How then do we manage the presentation of the characters?
The crucial concept that I just can't get around is the notion that a verb must be reduced to an image. So here's a question: should the faces of the characters be included in the verb display, or should they be considered separate? I can specify a fixed spatial position for each of subject, verb, and direct object, and then turn the artist loose with that clear specification, or I can allocate all the screen space to the three in combination &emdash; but what is gained by giving the artist freedom to control the layout like this?
Then there's the question of variable layout depending on sentence type. For example, here are some of the basic variations:
Subject - verb - direct object
subject - verb - direct object - indirect object
subject - verb
subject tells about (subject - verb - direct object)
Thus, I must plan for anything between one and three faces. If I allocate fixed screen space, there will either be much wasted screen space, or insufficient space.
What if I allowed two levels of graphic detail? What if there were a composition sheet screen showing most of the tale in small images, and another set of fullscreen images? This seems a copout to me; if the small images do the job, then we won't need the big images, and if the big images are required, then all that clicking will drive the user crazy.
Since tales are variable length, what if we organize it as a literal strip? Suppose that each verb were allocated, say, 320h x 256v, so that the triggering substory panel in a thread is placed at the bottom center of the screen, with half of the preceeding and half of the following panels on either side of it? The user can slide the strip to the right or left with a single button or perhaps a scroll bar.
Yes, this might work.