I’m taking an interesting course this semester, called Computer Vision. It’s a great chance to work on some zany interdisciplinary stuff and hang out with grad students. :hr:
Our final project for this course is a music video, and I’d like your input. I’m hoping to build something in Flash that can do the following:
[LIST=1]
[]Grab a video off of YouTube
[]Analyze any frame of said video
[*]make persistent changes to the video
[/LIST]
In detail: Using as3youtubelib, I’d like to get a video of the user’s choosing from Youtube and load it up. Then I would like to be able to grab visual data from a frame of that video, by drawing the Video object onto a BitmapData object. I think there could be some sandbox issues there. And then the big, final step is, I’d like to generate a video file of a common format (not necessarily FLV, but that would be nice). Each of these steps are a separate problem, and the plans for my project depend on which problems have a solution.
So I ask you:
First, is it possible to draw a Video object to a BitmapData object, assuming that the loaded FLV in the Video is in the same sandbox as the running Flash instance?
And second, is there a preexisting utility I can run on a server, that can create a video file of a common format from a sequence of images?
Oh, and check out the course website that I linked to. There’re some really great ideas on there.
**EDIT: ** I should mention my alternative plan for the project, in case it’s too difficult to generate a video file. Instead, I would store user data (including the source URL and changes made to the video during the user’s session) in a local SharedObject, which would be retrieved the next time the user visits the site. This would make it difficult to share the edited video with others, but at least it could be playable.