How to Make Tutorial Videos and Screencasts

Occasionally, people write to ask me how I made these movies. It’s not hard. It’s more of a craft than a talent. You can learn a craft through practice, rather than needing inborn talent. My purpose in writing this article is to help people to do this well, mostly for selfish reasons: there’s too much crap video on the web these days, and I want it to stop. :)

The Importance of Audio

It’s natural to focus on the video aspects. Video capture equipment costs more than audio equipment of equal perceived quality, it takes more computer power to work with video than audio, and the software needed to work with video is usually more expensive and more complicated than equally powerful audio software. In addition, most people are visually oriented. Just look at the language we use: we call these “videos,” not “audio/visual programs.” The term “multimedia” has thankfully fallen into disuse, even though it’s more accurate.

When it comes to making non-fiction videos like mine, audio is at least half the battle. If you think I’m exaggerating, try starting one of these videos, turn off the monitor, and just listen. I think you could get maybe three-fourths the value of most of the videos this way. Flipping it around, what if the audio were so bad as to be unlistenable? Would better visuals fix it? No. Audio is more than half the value in videos like the ones I have here. We must pay close attention to it.

Capturing Good Audio

The only way to get great audio quality is by using a dedicated microphone, positioned near your mouth. The farther the mic is from your mouth, the more noise you will get in the sound track; conversely, the closer it is, the better your ability to drown out room noise with the power of your voice. This rules out microphones built into the computer, or the cheap desk-mounted ones that sometimes come with sound cards.

A headset can work well, and indeed I did use one for my first several videos. I’ve been moving away from them over time, though. Every time you touch the cord, it transmits the vibrations up to the headset and then down the mic boom, resulting in a thudding or scraping sound on the soundtrack. Even on the simple screencasts, this happens a lot. If you’re like me, you make gestures while talking even when there’s no one in the room with you, so even if you’re careful when typing and moving the mouse, you’re still in danger of touching the cord. I’ve tried tying the cord up to something like a nearby bookshelf so it stays away from my arms, but the cord usually isn’t long enough to do what you want here. To remove all danger of hitting it, you have to take a circuitous path to also get it out of your line of sight to the computer screen, for screencasts, or the camera, for traditional videos. Then you end up adding extension cords and other such hackery. At some point, you seriously start questioning the “simplicity” of a headset.

I’m not the first to discover these problems. The professionals all know this, which is why they use dedicated microphones.

The simplest thing is a microphone on a boom of some sort. If you go this route, you need to either get a boom long enough that you can mount it off of your work surface, or get a shock mount for the mic. Otherwise, every time you hit your working surface, the vibrations travel along it, up the boom and into the mic, just as with the headset. This is a serious problem with screencasts, because even your typing and mouse movements can be enough to cause audio problems this way. A shock mount is an expensive way to fix this problem. If you have a book shelf or something like that near your work table, you can get by without the shock mount if your boom is long enough to reach from that to put the mic near your mouth. It doesn’t matter how you decouple the mic from the work surface, but ya gotta do it.

There’s a downside to using a fixed mic like this: you have to train yourself not to move your head around a lot as you talk. You might have to go back to the headset idea if you have to move your head around a lot. You then have to train yourself not to hit the cord, but you might find that training easier to take.

Another option is a lavalier mic. This clips onto your clothing, so it always travels with you. Careful routing of the cord will ensure that you don’t yank on the mic, giving the same sound glitches as described above. I’ve never bothered with these, because their main value is when the mic’d-up person has to be on camera. I don’t want to be on camera. :)

Capturing Quality Video

First, your captured video should be as close to pristine as possible. The better the source quality, the better the quality of your final deliered video. For screencasts, this means setting the capture program to use as little compression as possible; ideally, no compression. You only need 5-15 frames per second for a screencast, so there’s not much incentive to skimp on quality here, given today’s huge hard drives. For traditional camera-captured video, you probably can’t get away with uncompressed video, especially with HD. Still, use the lowest compression settings you can stand. The quality you get down the line will pay for the hard drive space it takes.

Second, be willing to edit the video. You will get a more natural presentation if you do a straight no-cuts, no-second-takes capture, but don’t take that to extremes. Even professionals let slip the occasional um or ah, or spend a little time fumbling for something. Cutting this useless stuff out can really tighten up a tutorial. You don’t need a really expensive video editing package to do this, though the better the software, the faster the edit goes.

Third, add lights. This can mean opening blinds and curtains, pressing table lamps into service as secondary illumination, setting up reflectors to fill in harsh shadows, even buying dedicated video lighting kits. The brighter the scene, the less noise in the captured video. Video noise is one of the biggest enemies of compression. Besides, you don’t want deep shadows in this sort of video. We’re trying for clarity here, not drama.

Fourth, remove all extraneous elements from the scene. Don’t just shoot your messy work table as-is. Remove everything that doesn’t help tell the story, and clean all the surfaces. If you’re going to have someone talking on camera, add a backdrop. There’s a big difference between seeing a bedroom behind the talking head and seeing a bed sheet behind the talking head.

Finally, use the best video connection you can. There are lots of different video connection types. In the analog world, composite is the lowest quality, then there’s a small step up with S-video, and then there’s the pro-quality component connection. In the digital world, things are rather more confused. In general, you want as few steps between the digital sensor and your computer’s hard disk as possible, and you want to use the widest “pipe” possible. The smaller the pipe or the medium, the more compression the camera will have to use. So, capturing to SD memory cards or live over USB is likely to give lower-quality results than capturing to tape, or over FireWire or HDMI.

You might be wondering why I mentioned HDMI. This connector, becoming common on newer cameras, is primarily meant for connecting the camera to an HD TV for presentation. However, this connection is usually a path around the camera’s built-in compression, so it’s a way to get professional quality results from consumer equipment. It comes at a price, of course. You need an HDMI caputre card (e.g. Blackmagic Design’s Intensity), a fast RAID to capture the fat video signal without dropping frames, and a willingness to tether the camera to the computer while shooting. If you can tolerate all that, you can get professional results on the cheap.

Compression Overview

The human brain is good at filling in missing detail in things it perceives. We humans don’t need to see the whole message to figure out what that message is. Clever people have taken inspriation from this and invented many ways to “compress” media files by throwing away some of the information in the file, relying on the human to fill in the blanks, as it were. These methods work on a sliding scale: the more you throw away, the smaller the file gets, but the more likely a person will notice the loss or the compression artifacts. Almost every digital media device sold — cameras, DVD players, MP3 players, etc. — uses some sort of lossy compression. The smaller the device can make the files, the less storage it needs, so the smaller and cheaper the device can be. You can crush media files down to arbitrarily small sizes, as long as you’re willing to accept the loss in quality.

I use compression in four stages: capture, edit, archive, and deliver.

The compression method used in the capture stage is often dictated by the tool I use to do the capture. If it’s a DV camera, part of the DV spec is a compression method, about 5:1 relative to uncompressed video. HDV and AVC cameras use even more compression, typically 20:1 and up. For a screencast or direct camera capture, I use a very light form of compression, usually either QuickTime Animation or Apple ProRes. Other good choices are TechSmith EnSharpen and MJPEG, each set for very low compression.

Video editing programs work better with some video compression methods than others. Even if your video editing program claims to be able to edit your camera’s native format, it will often work better if you recompress to something lighter. This is especially the case with today’s HDV and AVC cameras. With screencasts and direct-from-camera captures (i.e. not recording to tape first) I have the freedom to use something that will make the video editing program happy. If I’m using a camera with removable media (e.g. tape) I keep this as a backup, so it doesn’t matter if I recompress for editing. I can always get back to the pristine original if I need it.

Once I’ve got the video edited down, I save out an archival version of the final edit. Depending on how important the video is to me, I will save with varying levels of compression, but never more than about 10:1. So, I might save out in the same uncompressed or light compression method as I used for editing, in which case the only difference between the versions is that the archive file is a flattened version of the edit timeline. Or, I might compress it harder for something I don’t expect to go back and edit, ever. In that case, I might use some flavor of MPEG set to a quality level higher than that used by DVD; 15 Mbit/s MPEG-2, for instance, with a standard-def video, or more like 40-50 Mbit/s for HD.

Finally, I recompress the final edit for delivery, usually crushing it pretty hard. Depending on the project, I either do this directly from the video editor, if I keep the edit files around, or I recompress from the archived version of the final edit. You want to use the highest quality source video you can for the final compression because video compression artifacts add up through the stages. The better the source, the lower the bit rate you can use for the final compression while maintaining a given perceived quality level.

Final Delivery Compression

You need to use a lot of compression for web video. Even with the small frame size and decimated frame rates I use on this site, uncompressed video would be way too big to tolerate. 320 × 240 at 15 frames per second, uncompressed, is about 18 megabits per second, too much for a typical broadband connection.

Part of the reason YouTube video looks bad is because the submissions aren’t always great quality to begin with. But mostly, it’s because YouTube has a huge economic incentive to crush the hell out of the submissions. If you had their bandwidth bill and their crappy business model, you’d crush the videos mercilessly, too.

So, the first rule for delivering high quality video on the Internet has nothing to do with compression, per se: host videos on your own site, or use a paid content delivery network. If you use a free hosting provider, they get to dictate the video quality. If your videos become popular, this freedom to set your own quality level can get costly, so you’ll want to think about how you’re going to pay your hosting bills in advance. This site runs on a mid-tier hosting plan, paid for with proceeds from my parts store. Ads work for a lot of people. Donations usually don’t work out unless you’re willing to beg for them in the videos, and often not even then.

The second rule is, experiment with bit rates, frame sizes, and frame rates. The more bits you can let the video compression program have per frame, the better the quality it can give you. The more frames you have per second and the bigger the frames are, the more bits per second you need to keep a given quality level. If you cut frames per second in half, you can cut the bit rate in half as well. If you scale the frame size to 50%, you can cut bit rate in half again. (Yes, only half: the pixel count is quartered, but bit rate per quality doesn’t scale linearly with respect to pixel count.) You need to find a sweet spot where you can live with the quality of the video at the same time that you can live with the bandwidth bill you get as a result.

Third, use a high-quality encoder with lots of compression settings, and take the time to study and experiment to learn what these settings do. It’s possible to get a 2:1 bit rate savings without losing quality just by using a better encoder with well-chosen settings. There’s no reason to use anything but the most efficient setting for online web video. It’s common for video encoders to have several modes, some less efficient than others; the less efficient modes exist for other purposes. Learn what they are, and avoid them when making web videos. You might have to pay more for a better encoder, and you’ll certainly have to spend some time to learn how to drive it well. These costs will easily be offset by a lower bandwidth bill.

The Craftsman’s Tools

Think of this section as a set of examples, not a list of things you must go buy to get quality video. There are lots of things that can work equally well, or better. This is just the stuff I happen to use.

Microphone: Currently, I use a RØDE Postcaster on a cheap Heil boom. (Just listen to the videos, and you can easily figure out where I switched to it from the headset I used previously. The difference is not subtle.) This mic has several nice conveniences. First, it is a USB mic. This means you don’t have to add an analog mic interface to your computer, or put up with the poor quality of the one built into most computers. The USB also makes it easier to use portably. Second, it has a headphone jack on it, so you can monitor your sound quality as you record; because there’s delay in the USB chain, it’s far better to monitor at the mic than plug headphones into the PC. Finally, it’s extremely well built; it might be considered a lethal weapon, club class, in some jurisdictions.

Camera: I’ve used two different consumer-class cameras for my non-screencast videos. The first one was a standard-def DV camera, not worth naming. If you already have it or someone’s giving it to you for free, by all means, use it. If you have to buy a camera, though, get an HD camera of some sort. Unless you happen to have an old professional SD camera, a new HD camera will likely outperform it. This is even true on the web, where you often want to scale the video frame size down considerably; the higher resolution, scaled, still shows up in the crispness of the final video. Compare Tangent Tutorial #15, shot on a Canon HV20, to Tangent Tutorial #4, shot on my old DV camera, for instance.

You don’t have to get a professional camera. Lights, a good mic, and a tripod will do more for the overall quality of the presentation than getting a better camera.

Tripod: A good, sturdy tripod keeps the camera still and lets you point precisely at the right section of your work area. You’ll most likely be pointing the camera and then “locking it off” — that is, no camera moves or zooms in the final edit. If that’s the case, even a cheap tripod will do all you need.

Screen capture software: There are lots of choices here, especially in the Mac world. I’ve used both SnapZ Pro and iShowU for my screencasts. Both programs have their plusses and minuses, and if you don’t like these, there are two or three other choices, at least. I can’t say too much on the Windows side. I’m only aware of the semi-abandonware CamStudio and the expensive but awesome Camtasia Studio. If you know a program that’s under active development, somewhere between these two in cost, email me.

Video editing software: Most computers come with basic video editing programs built in these days. Unfortunately, all of the inexpensive stuff I’ve tried gets too much in my way. I’ve had an opportunity to try lots of video editing programs, and have yet to find something inexpensive that doesn’t make me tear my hair out when doing even relatively simple videos like you find on this site. I found older versions of Apple’s iMovie and Pinnacle Studio usable for basic things, but the current version of both apps goes in the wrong direction (“easier”) if you want to create your video’s look from the ground up like I’ve done here. Even simple layered effects like the transparent meter overlays in Tangent Tutorial #6 can be a hassle to put together in the low-end video editing packages. I ended up going with Final Cut Pro, but that was more because I wanted what is bundled with it rather than because I needed the full set of pro features to make these videos. For much the same reasons, Adobe Premiere Pro is really nice, but not actually necessary. If money’s a little tight, I recommend going with one of the semi-pro apps, such as Final Cut Express, Sony Vegas, or Ulead MediaStudio Pro. Your barber will thank you for avoiding the low end programs.

Graphics software: In producing these videos, I used several different graphics packages. I could name them all, but in point of fact, none of them have features so unique or valuable that they were key to the development of these videos. All I really want to point out here is that you shouldn’t look at your video editing software as a hammer and try to turn every problem into a nail. Be willing to step outside the video editor to build certain pieces. Pro and semi-pro video editing programs are really good at pulling pieces together from many sources into a single, coherent package. You’ll find uses for 2D vector and bitmap graphics packages, animation software, and various sorts of 3D graphics programs. When you start to pull together all these assets, you will appreciate having a multi-track video editing program with keyframing and full alpha transparency support.

Sound editing software: You will be able to do basic audio editing inside the video editing program, but usually you need to step outside and use a separate program to do some of the heavier lifting. Noise reduction and pop/click removal are just some of the things you don’t want to attempt inside the video editor. Even if you’re lucky enough to be recording your video inside a soundproof booth, you’ll still want to do a little audio sweetening. Remember, sound is more than half the value of videos like these.

More Ideas

If you want more ideas for tools to get, look for reviews put up by podcasters. The podcasting phenomenon has probably done more for consumer awareness of high-quality A/V production tools than anything since the advent of the garage band. What I’m doing here is basically the same as podcasting, from a production perspective.