Revision as of 13:09, 19 July 2008 by Sj
Michael Lew writes:
Basically I can foresee that the following developments might be useful, in order to develop a more powerful video activity for the OLPC than the current "Record" tool. Ultimately the idea would be that kids should be able to use the XO as a low-quality camera (Fisher Price camera / web cam / super8mm camera / phone camera), to shoot things around them, capture footage from outside sources (just as the web, USB, CF), share footage within the local mesh with local users, and then edit a film that could then be posted online for example, or viewed locally. It would be nice to draw inspiration from all the research done at MITML on video editing tools, especially collaborative ones, to come up with interface & functionality ideas. 1. FILMING For me, the first step is already to enhance the "Record" application, so that the XO can be turned into a handheld video camera. Hopefully in the next generation, the camera can be tilted, to make framing and composition easier. For now, the XO can be used with an external viewfinder : http://flickr.com/photos/curiouslee/2137058574/ http://flickr.com/photos/optilevers/2175612089/ Although there is of course a slight parallax such as the one that existed in cameras before they invented Reflex / SLR. But there are all kinds of settings we could try to play around with, to allow the videographer to manually adjust the camera settings, such as : shutter speed, white balance, brightness, digital zoom & other digital effects. For this we need to use low-level routines to talk directly to the image sensor (OmniVision7670). I will need to get in touch with the hardware guys, or people who worked on the drivers and operating system. 2. RECORDING This step involves playing with the codec, so that the image quality is good, while it does not take too much space on the local drive (or whichever external drive used, such as an online drive or an external CF card or other USB device). Since the laptop does not have a MPEG hardware chip built-in, the MPEG or MJPEG compression needs to be done in software and it's time-consuming. An idea would be to let people capture footage uncompressed, and then compress footage later, either as a background operation, or as an operation triggered by the user. There are various compression algorithms, all based on JPEG, but also motion estimation for MPEG. 3. INTERFACE Once we have a decent image quality, we need to come up with original interfaces that correspond to the activities that we want children to undertake. A simple goal is to implement a simplified version of iMovie, but that guides children into making short films with a structure. A good idea would be to first have the kid plan the film, by using a simple storyboard software. The child could choose between different types of shots (establishing shot, two-shots, close-up, etc.) and then sequence them on a storyboard. Then the child would shoot or capture shots, cut them, and match them to the shots planned in the storyboard. The overall length of the film should be limited to 3 minutes or so, to force the child into making editing decisions and to keep the film concise and dense. When the child creates a new project, there could be pre-made storyboards, as examples, that can be used. Interface proposals should be made on the wiki. Ideas for collaborative video applications should be made too, based on the many projects done in the IC/MF group over the years. So while we all brainstorm over interface ideas and application (activity) ideas, I will start by working on 1. + 2.