Primetime Hypermedia

Heavy Metal Umlaut: The Making of the Movie

by Jon Udell

When Wikipedia's page on the heavy metal umlaut made the rounds of the blogosphere recently, I decided to make a documentary screencast that would illustrate and narrate Wikipedia's editorial process. The screencast was well received, and a number of people wrote wondering how I made it. In this column I'll explain the process, then conclude with some general thoughts about the value of the screencast medium.

First, some background. In the fourth installment of this series, entitled "Movies of Software," I showed how to make a screencast using Windows Media Encoder. To create that meta-screencast demonstrating the use of Windows Media Encoder, I used the tool that's become my favorite for this work: Camtasia Studio.

Back in November, when I wrote that column, I didn't know what to call this medium I've been working in. So I posted a call for names on my blog, and a couple of days later, I collated the responses and settled on the term screencast. It looks like that name is going to stick.

Screencast Genres

The screencast medium isn't new, but its possibilities have been very much overlooked until now. As I've begun to explore the medium, I've been developing a taxonomy of genres:

Tutorial: A screencast that demonstrates how to use an application or service. The screencasts on my LibraryLookup page are examples of tutorials.

Tip: Examples include this 90-second short on Linky, a Mozilla/Firefox extension, and another shortie on Windows' hidden desktop search feature.

Demo/discussion: My first effort here was a demo of JotSpot, the "application Wiki," in which Joe Kraus drives the demo and I ask questions. A more recent example is this demo/discussion of Zend Platform.

Feature story: Looking back I can see that my first foray into this genre was probably Aunt Tillie's OS X Adventure, in which I respond to a pair of Eric Raymond essays [1, 2] on the horror of Linux's printer support. Eric had suggested that Mac programmers wouldn't make the same mistakes. But what I found on OS X, and documented in the screencast, was an almost equally bad experience.

Anatomy of a Screencast "Feature Story"

The Wikipedia screencast clearly belongs in the feature category: it tells the story of the life of a Wikipedia page, and traces several of its evolutionary motifs.

When I visited Wikipedia's heavy metal umlaut page and began stepping through its change log, it was clear that the sequence of revisions was intrinsically interesting. If you go to the beginning and step forward through the revisions, you can watch the history of the page unfold.

First I tried capturing the whole sequence with live narration, planning to edit that down to the final product. But each of the motifs I wanted to talk about--the typography and factual basis of the n-umlaut, the organization of the page, the episode of vandalism, and the Germanic theme--followed its own path through the history of the document. To trace each of these, I needed to break free from the linear sequence I'd captured.

One solution would have been to identify the revisions I wanted to talk about, capture each as a static image, import the images into Camtasia, and build a movie out of them. But that seemed like a lot of work. It's true that you can extend the duration of any single frame in order to accommodate narration, but adjusting a sequence of extended frames to a synchronized narration would be tricky.

Two-Pass Recording

The alternative I came up with was a two-step procedure. In step one I used Camtasia to record the sequence of revisions without narration, and then to produce the captured video as a QuickTime movie. In step two I played that QuickTime movie and used Camtasia to record the playback. This turned out to be a very powerful strategy. I could just let the QuickTime movie run, in which case you see the revisions morph page by page. I could stop the movie at any point, in order to synchronize the narration with a revision. And I could fast-forward (or fast-rewind) through the changes--with narration--to create a compressed view of the document's evolution, and to find particular revisions that I wanted to discuss.

The ability to scroll forward and backward along the document's timeline is, arguably, the coolest aspect of my Wikipedia screencast. On the first try, though, I ran into a problem. In the original capture, the mouse pointer was shown clicking on the Next revision link. But then in the secondary capture another mouse pointer appeared, as I used the mouse to draw attention to the page elements I was discussing. It was really confusing to see two mouse pointers at the same time, doing different things. So I redid the original capture, this time instructing Camtasia not to capture the mouse. Then the mouse pointer in the secondary capture made more sense. Ideally I'd have liked to alter its shape and size as well, to emphasize its role as a pointer rather than a control device. But although I tried a couple of custom pointer solutions for Windows XP, none worked at a sufficiently low level to fool the video driver.

The subject of this particular screencast lent itself particularly well to this treatment. But I think the two-pass technique will serve me well in other situations too. Synchronized narration, for example, becomes much easier this way. There are a couple of ways to add narration in Camtasia. You can record a second audio track underneath a video clip, or you can record one (or more) separate audio tracks that you import into Camtasia and combine with a clip. Either way involves a lot of mode-switching. When you record an original capture and then re-record and narrate a secondary capture, though, you can just let the "camera" run and do multiple takes, rewinding the video between them. Then in the final edit you mark the ins and outs for each take, review the alternatives, and delete all but the best take.

This method works nicely when you're both editor and narrator. It won't help much, though, when the sound track is a conversation that's been recorded live, as is typical of the demo/discussion genre. Still, even in those cases, it may be useful to re-record the original capture in order to add a layer of narration underneath--or between segments of--the demo/discussion.

Why This Matters

In the last week or so, I've seen a number of screencasts showing up around the Net. Nat Friedman used vnc2swf to make a couple of silent movies about Beagle. Mike Clark used Snapz Pro X to demonstrate CruiseControl. Raymond Kristiansen used Camtasia to showcase Socialtext. And Peter van Dijck discovered two things about Camtasia that I hadn't yet figured out for myself. First, that you can in fact use Camtasia to make a movie about Camtasia. Second, that Camtasia's zoom-and-pan feature -- which I hadn't gotten around to trying yet--is awesome. Peter's conclusion: "Camtasia: It's cool for screencasting, but [at $300] really expensive."

The screencast that really got me thinking, though, was something very different: the ACLU Pizza movie, a digital identity nightmare--realized as a Flash animation--that made the rounds of the blogosphere (again) while I was writing this column. The software it depicts is imaginary, but the underlying issues--privacy, identity--are very real. I'd seen this movie before, but this time around it struck me very differently. As more and more human experience is mediated through software, stories about how we use our software will become increasingly powerful tools of persuasion. Several folks who commented on my Wikipedia movie said things like: "Ah, now I can show people how this crazy volunteer encyclopedia actually makes sense." Many people who saw the ACLU Pizza movie said things like: "Yep, that's our future, like it or not." I've argued that technologists can't merely create software embodying an alternative vision for digital identity; they'll have to communicate that vision too.

Screencasting is a cool way to tell stories about software, and that's reason enough to care about it. But when the focus shifts from the software itself to our software-mediated social and political and economic lives, the true significance of the medium becomes clear.

Jon Udell is an author, information architect, software developer, and new media innovator.

Read more Primetime Hypermedia columns.

Return to the O'Reilly Network.