In teaching a class online this summer, I've needed to gain control over making video lectures. This turns out to be a rather complex process; I had several problems or difficulties to overcome.
The basic idea is the narrated PowerPoint lecture, but with instructor presence at the beginning and at the end, so students can see me for ten or twenty seconds before or after the slide show itself.
I should mention that I don't actually use PowerPoint; that product is powerful, giving many attractive visual flourishes (visual themes and transitions), but it's ill-suited for mathematics. Nor do I use Beamer, a technology that creates slide-show PDF documents using LaTeX mathematical typesetting; it is powerful, but a bit laborious to author documents with, and somewhat rigid visually (you don't have much control over fonts, etc). Instead, I used an HTML/JavaScript slide deck template, with MathJax for the mathematical notation. (I saw similar slide deck templates online, but authored my own simple slide deck. It's very simple visually, but easy to author and control.)
The product I use to record my narrated slide show lectures is Camtasia, which does a terrific job with screen-capture recordings, and gives the user good control on how to display webcam video (for instructor presence) with the screen capture: you can show yourself as a video inset, or you can alternate with the screen capture (my approach). There's a bit of a learning curve, but once you catch on, it's easy and graceful to use.
But there have been some puzzles to solve, to obtain good results.
One little puzzle: I noticed that the audio on a Camtasia screen capture was only in the left channel, when using a Shure M58 microphone through a Focusrite Scarlett Solo USB audio interface. Apparently, the interface is meant to be used with a DAW (digital audio workstation), not a product such as Camtasia. The solution is to mix the audio to mono, which can be done in Camtasia in the properties of the audio track. (Separate the audio from the video track first.)
I should digress to mention why I am using the Shure M58 microphone, rather than more common microphones for this kind of work. The M58 is a stage microphone, suitable for singers. It produces good quality audio, it is rugged, and it is not expensive; well under 100 dollars. There is a drawback in that it requires a USB audio interface; the popular Focusrite Scarlett Solo costs about 100 dollars and works well. So the total investment in this equipment was about 250 dollars (including a microphone stand and a cable), rather more expensive than more common solutions such as the Blue Snowball microphone (which I tried). But the big advantage of the Shure M58 is that it is a dynamic microphone, not a condenser microphone. So the M58 is less sensitive, because as a dynamic microphone, it is driven by the sound of the voice; condenser microphones are amplified and very sensitive. The M58 also has a cardioid pattern, meaning it is sensitive to sounds at the end but not from the sides. All of this means that the microphone does a very good job of isolating my voice. Ambient sounds like the mouse rubbing on the table, or clicks of keys on the computer keyboard, or sounds from outside, are not picked up by the mike. At any rate, I can hear the difference between the M58 and other solutions, such as the Blue Snowball, or (especially) the kind of inexpensive headset mike often used for telephony.
Another puzzle: I wanted to use a cursor in my slide show. But the ordinary cursor (when using Chrome) is small, and also will turn into an I-bar when it's over text. I thought I would make the cursor big by using the accessibility features of the Mac OS, but when I exported a video, the cursor was small again. I learned that Camtasia only records cursor positions, not the pointer itself. However, Camtasia has cursor effects that can be applied. One of these is to put a colored circle over the cursor, whose size, color and transparency can be chosen. The resulting effect is pleasant, although I wish Camtasia did provide a big pointer for the cursor.
But the biggest problem to solve has been how to record a good video, free of errors and with a good flow and fluency of narration. Of course, Camtasia lets the user edit out mistakes and pauses. But it is frustrating to delete a recording and start again, over and over, because I can't stay on track (going off on an unplanned tangent, or forgetting to say something important). An unscripted lecture can yield an informal and fluent performance, but it takes too long to get a good result. The solution is to write a careful script, and read from it. Unfortunately, this proved impossible to do when I also needed to advance slides and transitions, and use my cursor to point at things. But Camtasia allows the user to do multiple screencasts or webcam recordings, and combine them. So I was able to record the audio from reading my script, and then do a screen capture without audio, just recording slide transitions and the cursor while listening to the recorded narration. It proved very easy to edit these together properly (synching the slides and cursor with the narration). This should allow me to put together good narrated slide shows relatively quickly. It seems to be impossible to avoid having to start a recording from scratch, but mainly because errors crop up in the script or the slides; these are not hard to fix.
An unscripted lecture has another problem: it is very time consuming to edit machine-generated closed-captions; with a domain-specific subject such as mathematics, the 'mechanical' closed-captions our learning management system generates become a miserable hash, and it is necessary to rewrite most of the captions. (I noticed the mechanical closed-captions rendered "triangular number" as "trained killer number".) But if recording begins with a verbatim prepared script, editing closed-captions is much easier; just copy-and-paste the language. My sense is that it cuts in half the amount of time needed for this task. By the way, the need or value of closed-captions is underscored by the experience of preparing these myself: I discovered that it's not always clear what I said, when trying to correct captions for an unscripted lecture. I am positive that my students will find captions useful, even if they don't need this accommodation (due to limited hearing).
I cannot say that the videos I've been preparing are of high quality, but I think I've made substantial progress towards that goal. I have a somewhat complicated system that might enable me to prepare videos of an acceptable quality in a more efficient manner. Of course, videos of genuinely high quality require professional production, using equipment and expertise beyond my reach. But I do recommend Camtasia as a valuable tool for anyone who needs to produce screen-capture videos. (I wish my university supplied us with that software, but the academic price is not exorbitant; less than $200.)