Re: mouse movement learning/synthesis
« Sent to: Taharok on: January 26, 2013, 10:08:38 pm »

Hey, thanks for replying.  Hah I guess this must have been a surprise receiving a message from someone almost a year after you made the thread.. I hadn't even noticed the date. I'm very much interested in having some kind of software that could learn/interpolate/synthesize mouse movement, as it would be a a core component of project that I'm working on.

Here's a quick little back story: I recently spent a couple of years in graduate school for painting(printmaking), and pretty much used my computer/photoshop the whole time in the program as my primary drawing and painting medium. I think working with digital tools helped develop and refine my aesthetic in a way that traditional mediums couldn't. My interests since school have shifted towards machine learning  and generative art, and being somewhat bored with simply drawing/playing around inside photoshop, I've become interested with the idea of having my computer emulating how I interact with photoshop.. making image-manipulating "decisions" in a manner that shares a compatibility with my personal aesthetic. This is sort of a backwards-way of 'typical' generative art production, .com/search?q=generative%20art where these sorts of 'generative art' images are very self-evidently 'about' process and generative-ness FIRST, and the aesthetic that ensues is essentially a secondary by-product of generative-ness (usually). My procedure is the other way around: images yielded from the generative process would simply look like  I made them 'on my own.'

I realize it's a bit ambitious to get from 'a' (a blank canvas) to 'b' (an image that looks like something i would have made), so I'm starting with much simpler goals first, compartmentalizing what's essential to making the overall process work. I could very well just go about this task by avoiding the use of the mouse all together, and have photoshop perform actions based on non-mouse activities..  which would be fine, except my style of producing images is very much about 'drawing' .. thus the need for the synthesized mouse component. I'm presently making data recordings of many different types of mouse activities I would typically perform when drawing inside of photoshop, including different kinds of  'random' mark-making, manipulating lines/shapes already present, etc, creating a huge  library of information concerning my drawing activities inside of photoshop. At some point, synthesized mouse data (based on these prerecorded mouse-movement learning sessions) would be needed in an autonomous computer drawing environment, using specified mouse motion at appropriate times when the computer is carrying on 'emulating me' inside of photoshop.

So, I've got my work cut out for me, but at least its a nice goal to have.. something to keep me busy in the mean time. The real hurdle is the synthesized mouse component. Sorry if this is a little scatterbrained, this is really the first time I've spoken with anyone about what I'm hoping to do, so at least its nice to kind of think things through in writing.