zden
commented
10 years ago Of course, 'save script' operation would store parameters and how many steps these parameters were used. It would be sort of simple 'timeline script'. But I was not planning to store how many lines were drawn in each step. That would work automatic I guess. I would offer ability to set the same random seed for each frame so when applied on several video-footage images the same sequence of random pixels would be 'fired' to the algo so images would not "shake" much. Please look at this video which was created with this filter ( https://www.youtube.com/watch?v=z7kajB5pM_s ). If you feel like collaborating on psychological studies game or something less serious, we can discuss that. How did you mean it with line count? In replay script mode, script whould hold the ending step. User could choose also less steps. But not generally counting number of used lines/rectangles/triangles/circles... But I plan also to use this filter in sort of 'vector video format' so then I'll save exact coordinates for each geometric element. But that is somehow a different story.
stewie
suggesting "optional" b/c I realize that accessing the DOM to update the counter element is a relatively costly (slow) operation.
Counter var would increment continually, but element displaying "lines drawn: ____" would remain hidden (and empty) until user click button requesting count display. Another click would hide the counter element, and updating its innerHTML would be suspended.
The "future" blurb on the page states that you are interested in developing a "playback" feature for the app. With this request, I'm pointing out that playback begs (in addition to source image plus drawing parameters) a record of "how long the drawing operations progressed" in terms of number of lines drawn. Working with various images, I've noticed a "diminishing returns" effect ~~ earlier lines drawn wind up being overdrawn. Once the drawing details are fairly well filled in, changes resulting from continued drawing are barely noticeable. (In fact, without zooming, it becomes difficult to gauge whether drawing is still underway, or has stalled/completed.)
I can envision several "neato" scenarios which would be possible if controlled playback could be achieved. (um, semi-controlled. A degree of randomness exists in the generative drawing.) A game / contest might have players competing to identify "what IS it? What's portrayed in the image?" -- timed trials and high scores... or collaborative real-time chatting and guessing while the image progresses. Also, given a base set of images, it might be interesting to study (age-based and other demographic factors) which protoimages are more easily (quickly) recognizeable.
As a related request/suggestion: The "Save Image" command could generate an output filename based on concatenated "sourceimagename_" + counter + "lines.png"