Practice, Practice, Practice: An Amateur Filmmaker’s Journey

The old saw, “How do you get to Carnegie Hall?”  Answer:  “Practice, practice, practice,” is so true.  All of us are impatient: we just want to “do artistic stuff” and have it turn out like the examples that inspired us in the first place.  However, no matter how refined our tastes, our talents take time to develop.  How long depends on how much help or critique we get along the way, plus a lot of hard work.  What follows is a narrative example of and informal tutorial on making videos on a budget, with inexpensive equipment and open source software.

I’ve always wanted to learn to make video presentations.  I imagined I might want to record test flights in the homebuilt airplane project that has languished, unfinished, in my cluttered and sometimes soggy workshop.  Another  project is documenting our bicycle travels.  One obstacle was gear: quality video equipment is expensive.  However, all modern digital cameras have a video mode.  I started practicing several years ago, strapping my Fuji pocket camera and small tripod to the handlebars of our tandem bicycle, to document rides on the bike trails.  It was pretty terrible, amplifying the bumps and roots on the trail and the clicking of the gear shift, as well as not being very well attached, with the camera flopping around from time to time.

The next year, I got a GoPro Hero 3 (Silver–the mid-range model) point-of-view sports camera, and a handlebar mount made for it, a modest investment.  The  GoPro web site has daily videos sent in by users, showing off amazing feats of surfing, bicycling, motorcycling, scuba diving, parachuting, wing suit plunges, and all manner of dangerous sport, seen from a camera strapped to the forehead, chest, or wrist of the participant.  Some were exciting, some just plain scary, but all very professional-looking.

motelDSCF0614
The Mean Green Machine on tour in Michigan’s Upper Peninsula, with the GoPro mounted above the handlebars (and the headlight) in its waterproof case. While at an ideal Point of View height, the mount tends to vibrate a lot. I have a helmet mount, but my riding style involves a lot of head movement, which is distracting as well.

 

At first, I strapped the new camera on the handle bars of our Bike Friday Tandem Traveler “Q,” turned it on, and set off on a 35-km ride.  The result was better than the first attempt with the Fuji, but still shaky, vibrating, and endless.  OK, a bit of editing to show some particularly interesting parts, or at least cut out the really boring and really shaky parts.  But, a lot of time sifting through gigabytes of footage.  I eventually pared the hour and a half of “film” (I only recorded one way of the out-and back ride) down to 11 minutes of not-very-exciting or informative view of lake and woods drifting by at 15km/hr bicycle speed, plus a few moments of 30km/hr downhill bouncing and shaking.  The sound track was a muffled one-side conversation between me and my stoker, Judy, on the tandem, plus a lot of road noise transmitted through the frame, and the frequent clicking of the shifters  and hissing of the brake shoes on muddy metal rims.  A really round-about way of saying “We went for a really satisfying bike ride up the south shore of the lake, and came to nice waterfalls about once an hour.  Wait for it.”  Fast forward two years of trial and error…

After watching a lot of other people’s videos, and the progression in skill over the years of some of my favorites, like Dutch cycle tourists and videographers Blanche and Douwe, I have possibly picked up some hints of what makes a good video presentation.  I mostly publish on Vimeo.com, which offers a set of short tutorials on making videos., but I also recently viewed some good tips by Derek Muller, a science educator who makes a living filming short YouTube videos on various topics in science (Veritasium.com), and Ira Glass, host of National Public Radio’s “This American Life,” who published a series of four short talks on storytelling on YouTube.  Both agree that getting good takes practice.  Lots of practice.  Probably not as much as Malcom Gladwell’s tale of 10,000 hours of solid practice (in “Outliers“).  but a lot nevertheless.

The main point of Derek and Ira’s stories is: video is storytelling.  As we found, it is not enough to simply record the world as it goes by on your adventure.  The result has to tell a story: why you did it, where you went, what it was like, and what you learned, in a concise way that holds the interest of the viewer.  I know that most of my efforts failed, because of my viewer numbers on Vimeo.  Sometimes both of my loyal viewers watch a particular video, sometimes neither of them do.  Obviously something needs work.  Submissions to video contests garner a couple hundred views (compared with thousands for the winning entries, and millions for the viral baby, cat, and stupid human tricks videos on YouTube and Facebook), with no idea how many viewers actually watched all the way through.  So, we evolved over time, failure after failure.

First, I got rid of the “native sound,” because what the camera mic picks up isn’t what I focus on or even consciously hear while riding.  Instead, I find a piece of music that I think reflects the sense of motion and emotion in the ride, or one that at least fits the length of the film, or that I can cut the film to fit without making the visual too short or too long.  A fast ride needs a beat reflective of the cadence; beautiful scenery or glorious weather deserves a stirring orchestration or piano number, a matter of taste.  The next step is to trim the video clips to match the phrasing of the music, if possible.

I realized that, though I find looking forward to what is around the next bend exciting while on the bike, watching endless scenery flow by on the small screen isn’t particularly engaging.  Most of other people’s videos I enjoy have clips (scenes) of 7-10 seconds each.  Mine ran generally 20 seconds to several minutes.  Boring.  Furthermore, long takes don’t necessarily advance the story line, just as important to film as to the page, unless there is some interesting progression unfolding in the clip, much as a detailed sex scene in a novel is only necessary to define a key point in the development of the relationship between the characters:  mostly, it is sufficient for the characters to retreat to the privacy of the bedroom behind a line of asterisks, as a transition between scenes.  A video fade on a long stretch of empty road to the next bend suffices just as well.  We’re not promoting “bike porn” here: no matter how much we personally enjoyed the ride.  I’m beginning to appreciate the need for story-telling that doesn’t fall into the “shaggy dog” genre, i.e., drawn-out and pointless–suspense to boredom without a satisfying punch line.

Picking music is another issue.  At first, I shuffled through my library of ripped CDs (no piracy here,  just a convenient way to carry your music library with you, on the hard drive of your computer instead of a case full of plastic in a hot car).  However, even if the audience is small (i.e., myself and others in the same room), such usage violates the copyright on commercial recordings, especially on a public post on the ‘Net.  I’ve recently started re-editing some of the early videos I made this way, substituting from my new library of royalty-free music published under a creative commons license, and downloadable from several sites on the Internet, notably www.freemusicarchive.org, where musicians leave selections of their work as a calling card or audio resume, hoping for commission work or performance gigs, or to sell physical CDs in uncompressed, high-fidelity audio instead of downloading the lower-quality MP3 lossy compression version.

This is the wave of the future in a world where digital copies are easy: whether you buy a copy or get one free, play it, listen to it, use it to enhance your art, just don’t resell it whole.  That’s the idea behind creative commons.  Unfortunately, much of music publishing is still in the “for personal use” only, and if the pressed recording gets scratchy, buy another one, no “backup” copies allowed, and no sharing with friends: if you want them to hear a song, invite them to your house or take your iPod over to theirs: you can’t stream it or email it or share a copy on the cloud.   ASCAP blanket licensing for broadcast or use in video is still on a corporate price scale, intended for production studios and well beyond the reach of a PC user who just wants to add her favorite pop tune to a video of her and her friends having fun.  So it is that Kirby Erickson’s ballad of driving up US93 through the Bitterroot as background to a bike ride up US93 through the Bitterroot is gone, so viewers who aren’t familiar with his work won’t be tempted to buy the album the song came from, because they won’t ever hear of it.   Restrictive licensing actually potentially reduces sales in the Internet age.  By now, you can’t even upload videos if they have copyrighted music audible in them–Facebook, for one, matches audio signatures from video against a sound library and blocks them.

Although I see some improvement in quality in my amateur videos, I still have a long way to go.  For one, the handlebar mount for the GoPro camera introduces too much vibration, so the picture is hard to watch, and doesn’t reflect the experience of riding.  Some damping is needed.  We did get some better results with the camera mounted on our trailer, but we only use that when touring.  Some sort of counterweight to produce a “steady-cam” effect might work here, as the “real thing” is expensive and a bit bulky.

The story is more interesting when it shows the participants, which, for us, means using the trailer mount or some sort of “selfie stick” to put the camera to the side or front, or, as I did in one clip, turn the camera around briefly.  During my convalescence from heart surgery last summer, we did a lot of hiking, where I devised a selfie-stick approach to give the impression of the viewer being with us instead of sharing our point of view.   I’m a bit happier with some of those, particularly the ones where the camera boom isn’t in the view.  More practice, and experimentation.  I’ve been more satisfied with ones where we’re in the shot only when necessary to tell the story (an essential point, when the story was that I was OK, and getting better), and the scenery out in front when it was the story.

Now, the issue is to trim the scenes to the essential elements (who, what, where, when, why, and how).  To that effect, part of the re-editing process to replace the audio tracks involves cutting the video to synchronize with the sound track phrasing, as well as reducing the length to the minimum necessary.  To paraphrase E.B. White’s dictum on writing, “Omit needless frames.”

fireflyDSCF0730
This still frame says it all: who is reflected in the window, how is the bicycle, where is “Firefly Coffeehouse” in Oregon, Wisconsin. what is “bike tour,” and why is, well, we’re having a good time.

 

One of the issues with being the director, cameraman, and actor all at once is to keep the bicycle safe while planning the shot and operating the camera, as well as keeping the mission (travel) moving along.  We miss some good shots that way, but it is inevitable.  One popular technique is to set up the camera along the route and show the bicycle or hikers approaching or receding across or into the frame, which involves stopping and staging the shoot.  This is less intrusive where there are two or more cyclists, so it is a matter of setting up the shot ahead of or behind the other rider(s), but that isn’t an option with the tandem, and we’ve used it in limited fashion by propping up the monopod/selfie-stick along the trail.  We do have several sizes of tripods, but they aren’t convenient to carry when the photography is incidental to the main purpose of travel.  I’ve long since taken to filming short takes “on the fly” rather than just leaving the camera on to pick up everything, which involves anticipating some scenery reveals or events, and, of course, missing some.  But,  editing “on the fly” to limit scenes does shorten the editing process and save battery life on the camera.  We work with what we have.

Recently, we entered a video contest for a short travel documentary on the Newberry National Volcanic Monument,  in central Oregon, which seemed to demand some dialogue in addition to the usual soundtrack and titles, so we experimented with voice-over to add a short narration where appropriate.  This also wasn’t the best, since our microphone is the headset-attached variety, suitable for making Skype phone calls and video chats, but little else.  Good quality condenser microphones for the computer and lapel microphones compatible with the GoPro are simply not in the budget, along with professional video cameras with microphone jacks or built-in directional microphones.  Drones are all the rage, now, but one suitable for carrying a GoPro as a payload is stretching the budget, also, and presents safety and control issues for use in our primary video subject, i.e., bicycle touring and trail riding.

Besides finding a story in a video clip sequence, getting the story to flow smoothly, and finding an appropriate sound track to evoke the mood of the piece, the skill set also involves learning to use video editing software.  Microsoft Windows comes with a decent simple video editor, but we don’t use Windows.   We do have iPads, which have apps for making videos, but haven’t spent a lot of time on those, which also limit one’s ability to import material from multiple sources (the apps work best with the on-board iPad camera).   There are a number of contenders in the Linux Open Source tool bag, some good, some complex.  We chose Open Shot, a fairly simple but feature-full non-linear video editor, which gives us the ability to load a bunch of clips, select the parts we want, and set up multiple tracks for fades and transitions and overlays of sound and titles.   We also found that the Audacity audio recording and mixing software can help clean up the sound from less-than-adequate equipment.  ImageMagick and the GIMP are still our go-to tools for preparing still photos to add to the video.  Open Shot uses Inkscape to edit titles and Blender for animated titles.

Video is memory and CPU-intensive, so it helps to have a fair amount of RAM and a fast multi-core CPU (or several).  Our main working machine, a Zareason custom Linux laptop, has 8GB of RAM, an Nvidia GeForce Graphics Processor Unit, and a quad-core dual-thread CPU, which looks like an 8-processor array to Linux.  This is barely adequate, and often slows down glacially unless I exit from a lot of other processes.  The more clips and the longer the clips, the more RAM the process uses; often the total exceeds the physical memory, so swap space comes into play.  I’m usually running the Google Chrome browser, too, with 40-50 tabs open, which tends to overload the machine all by itself.

This isn’t something you could do at all on a typical low-end Walmart Windows machine meant for browsing the ‘Net and watching cat videos on Facebook and YouTube, so investing in a professional-quality workstation is a must.  Since we travel a lot and I like to keep our activity reporting current, that means a powerful laptop machine, running Unix, OS/X, or Linux.  Fortunately, our laptop “strata” is in that class, though only in the mid-range, a concession to the budget as well as portability.  We purchased the machine when we were developing software to run on the National Institutes of Health high-performance computing clusters, and is roughly comparable to a single node in one of the handful of refrigerator-sized supercomputers in the laboratories that have several hundred CPU cores and several dedicated GPU chassis each.

In addition to Open Shot, we also sometimes use avidemux, a package that allows us to crop and resize video clips so we can shoot in HD 16:9 wide-screen format and publish in “standard” 4:3 screen format if necessary, or crop 4:3 stills and video automatically to 16:9 format to use with other HD footage.  In addition to the GoPro, we now have a new FujiXP pocket camera that can shoot stills and video in 16:9 HD, and a Raspberry Pi camera unit, that is programmable (in Python), that we use for low-res timelapse and security monitoring.  The programmable part means we write automated scripts that select the appropriate camera settings and frame timing and  assemble a series of still photos into a timelapse movie, using the Linux ffmpeg command-line utility.

So it goes–gradually, the videos we turn out get slimmer and more to the point, if not technically better quality, something we need to work on constantly with prose as well, as an intended 500-word blog post ended up a 3000-word tutorial instead.