Screencasting To A Script

Following on from my previous blog post about Getting Back Into Screencasting I’ve put another one together. There’s been some really interesting discussion on Ian Ozsvalds Screencasting Handbook group about the methods people use for screencasting, and I’ve used a couple of ideas here.

As part of my experimentation, I made some changes, but a lot the process has stayed the same. I created a very simple screencast just going through the basics of the Software Sources app in Ubuntu. The result can be seen on the screencasts site. Here’s what I did differently. Feedback welcome.

Increasing screen resolution

The Ubuntu Live Installer Boot Screen was recorded at the resolution which the virtual machine ran at for the boot screen of Ubuntu – 640×480. This is pretty low, but the boot screen isn’t very complex, and the font sizes make it perfectly readable at that resolution.

I figured 640×480 was a bit cramped to show Software Sources and Update Manager so I went to 1024×768. On the screencasts site I have set the window size to 640×480 so it may be some text is unreadable in that window, but if you go ‘full screen’ or indeed download the video and maximise it, it should all be very clear.

Scripting (almost) everything

I decided this time to script the whole thing because I want to get some terminology spot on. In the past I have talks on the fly with no script and got a bit of a beating from people for getting a couple of facts wrong. Having the script online also means that those people who can’t keep up with my voice speed or accent can read the text and get the same information.

I did stray from the script once, but not too badly :)

Challenges

Having the VM on screen along with the script, a terminal running recordmydesktop and a stopwatch got a bit cramped, even at 3360×1050! Here’s a screenshot :) I had to maximise the text editor and reduce the font size down so I could get the whole text on screen at once. I tried normal font and this meant I had to wave the mouse over from the VM to the text editor to scroll which I think would be off-putting. Maybe a console based text editor would be better at conserving space.

I also found myself looking at the text editor whilst keeping my peripheral vision on the VM. So you may notice less mouse movement in this screencast as a result. Maybe a possible solution would be to have the script on a second computer with a separate mouse (and of course keyboard) so I could scroll the script whilst keeping the mouse on the VM focussed. Another option would be to print the script out.

At least partially as a result of using a script I re-recorded this screencast more times than the one I did a couple of days ago. I found myself stumbling over my own text, ad-libing and getting stuff wrong, or just realising that the script didnt work in places.

I rewrote the script a bit here and there and finally recorded the 5 min video after about the 10th take! This can be esepcially annoying if you’re making changes to a machine during the screencast. Luckily the VM software I use has a snapshot feature so I can roll back. But this is all time consuming for a 5 minute “record and go” screencast. It’s also quite frustrating having to start again and re-do the same thing over and over.

I think it worked out okay. It’s not a deeply technical screencast and doesn’t cover massive detail, but it’s only 5 minutes long so digestible to the youtube generation. Annoyingly I left the feature of recordmydesktop which outlines the recording area switched on. This is why the thumbnail you see on the site has a big black box on it. Messy.

Feedback welcome.

4 thoughts on “Screencasting To A Script”

  1. When I was thinking of doing screencasts, I realized it’s damn hard to record on your own. Thats probably why there’s a bunch of 2-person podcasts and radio shows.

    Being ever crafty, I’ve been looking at speech synthesis tools, mainly festival. Maybe take a look at http://hts.sp.nitech.ac.jp/, the results seem pretty good. If you look into details, they’re all based on recordings from 1999, and aren’t very high fidelity, which impairs things.

  2. Would some kind of autocue tool help? If it only showed one line at a time it’d take up less space so the font could be bigger and it would be easier to keep in the right place. It could have a global shortcut to go onto the next line. If you think it’s a good idea, let me know and I’ll knock something together.

    Keep up the good work, you’ve clearly got a lot more patience than I would have.

    Dave

Leave a Reply

Additional comments powered by BackType