So as luck would have it, the film I was going to do a how-to on, The Prayer Tree, got into two film festivals – the International Poetic Short Film Festival, running Monday evenings at the Nuyorican Poet’s Cafe, with a different lineup of poetic shorts each night; and the Seguin Arts and Film Festival in Seguin, Texas. So here’s a not-so-brief rundown on how the short came to be. Hopefully it’ll serve as an inspiration, if nothing else.
Starting Without An Idea
Usually, I do things the way you’re supposed to: I come up with an idea for a story, write down a premise, expand it into an outline, then write a few drafts of the script. At the end of this grueling process, I have a script that’s well-formed and tight. Apart from the fact that this process can drain most of the fun out of writing, it also tends to make me think in a certain way. So I wanted to do something more like collage – start with pieces and see if I could still a story together out of them.
From the beginning, I had an idea that I wanted to combine different layers of video and stills. I also wanted to do something dirt cheap. The Prayer Tree ultimately has five layers or elements:
- water tank footage of me drawing something with pastels – shot on HD on the Sony Z1
- drive-by footage of a beach – shot out the window of a moving car with my Canon Powershot still camera, in video mode
- A still photograph depicting a woman standing in front of a tree – shot on the same still camera
- A music track – put together in Acid
- Subtitles that tell the story and tie it all together
All of these elements were shot months apart, sometimes for other projects or without any specific goal in mind. So this “how-to” will necessarily be a little non-linear.
Water-Tank Footage
I first used a water tank for a short film I did called Mournir/Mournen. It’s an old-fashioned technique used to create atmospheric/water/oddball effects. The basic principle is simple: you fill an aquarium or other large clear-walled container with water. I bought a a plastic container in Chinatown, then waterproofed the edges with silicone. You set the water tank on a flat surface, and put the camera facing it from one side. I lit the tank with sunlight from the window, plus a little fill from my desklamp.
For The Prayer Tree, I stuck a piece of paper inside the tank, on the wall opposite the camera, and drew some figures and lines/shapes (mostly abstract) with pastels. The water and gravity caused the pastels to smear down the page, and the water created a hazy effect. I shot the footage on my friend Ben Wolf’s borrowed Sony Z1 camera (HDV). I shot it to tape at 1080/60i, and downloaded it to my computer via FireWire to Video Vegas (more on that below).
The Beach Shots
I happened to be riding in the passenger seat of a car driving along a beachfront down in Virginia. It was pretty, somewhat desolate, and flat. The model I have – the Powershot A590IS – has a video mode that shoots 640×480, at 20fps. So it’s not the camera you want to shoot your next feature on. But it’s perfectly good for scouting, test shots, auditions (it records sound through a built-in mic), docs, “stealing shots”, etc. The video can suffer from the “jello” effect that’s common to CMOS cameras. The best way to deal with this is to put the camera on something steady (a tripod, monopod, block of wood, etc.).
At the time I shot the material, I was unaware of CHDK – a software hack for Canon cameras that enables you to do a lot of great stuff (save to RAW files, do real time-lapse work, and control the video compression).
I took the video off the card and opened it in QuickTime Pro. You can also use VLC Media Player (free). Canon saves the video using a Motion-JPEG codec, wrapped in an AVI file. The best thing to do is to convert your video to uncompressed SD. In QuickTime Pro, use the Export function. Set the Compression to None (see screenshots), make sure the quality is set to Best, and let it rip. NOTE that this will produce a very large file. Saving the file as uncompressed video helps keep the footage from suffering recompression during editing and compositing.
The Still Image
The still was taken with the Powershot at the Bronx Botanical Garden. I opened it in Photoshop, did a dozen different things to it – adjusted the brightness, brought the blacks up, saturated the colors, and applied a couple of filters. The idea was to give the photo an otherworldly look. I saved a few different versions of the photo to JPG and saved a master (with all the different altered versions on different layers) to a PSD file.
When working with stills in video, it’s important to note that most cameras take fairly large images (about 4K by 3K) and at higher resolutions (180dpi). The Sony EX-3 shoots 1920×1080 video, at 72dpi. So you’ll have to decide if you want to resample the photo down in Photoshop, GIMP or whatever your still photo editing program is; or let the video software do it. For The Prayer Tree I decided to see what would happen if Vegas did the work. You may also want to leave it at a higher frame size so you can do digital zooms or pans.
The Soundtrack
Sadly, I’m not much of a musician – some recorder lessons, and some drum lessons, and that’s about it. But thanks to looping tools like Acid, Live, Soundtrack, and others, I can lay down pre-made music loops and come up with some decent-sounding tunes. A real musician could come up with something better, but I think I did okay here.
A few tips: you’ll want to change the sample rate to 48KHz from 44KHz. Try putting loops together with the same bpm setting, otherwise they may fall out of sync with each other. And don’t lay down too many tracks – I usually start with 10 and get down to 4-6. If you have dialog in your piece you’ll have to be careful of both the sound levels on your music tracks, and also the frequencies of the loops you use. Just listen to the loop while playing back some dialog. Does the loop interfere with being able to understand the dialog? If so, you’ve either hit a happy accident, or (more likely) you need to filter the loop or use a different one.
Laying Out the Layers
I brought each of these elements into Vegas as separate tracks. So I ended up with three video and one audio (stereo) track. The first thing I did was order the layers so that the water-tank material was on top, then the drive-by material, then the still photo.
Still looking for a story, I tried various things. Cutting between the stills, the water-tank material, and the beach. Putting in nonsensical text cards. Super-fast (3-frame) cutting. Eventually I realized the solution was to layer one video element on top of another. I selected the most interesting bits of each element, butted the bits together with some dissolves, and then changed the opacity of each clip at certain points. This resulted in something pretty close to the final film. Sometimes the water-tank footage takes over. Other times the still photographs bleed through.
Next I realized that the water tank material was too fast for the music. I slowed the clips down, in some cases by as much as half their original speed.
When I was done I had a piece that was about 3:41. I went through the usual editing process, until had a version that ran about 3:18 with titles. Perfect.
An Idea Hits Me
At this point I put the project away for a couple of months. In the meantime, a lot of other things happened in my life. This was good – the time away from the project made it possible for me to look at it afresh. I realized finally what the story would be about – a car ride on the beach taken as an adult would remind the narrator of drawing during car rides as a kid. The tie-in would be the relationship the narrator was in the middle of as an adult. I had a story, of sorts.
I jotted down a mini-script, then tried recording it. My delivery wasn’t particularly good, so I started talking to some voice-over actors. But then I realized that telling the story in subtitles might be more powerful. So I added another video layer, put in the subtitles, and made some minor trims. And I was done!
Lessons Learned
There is no right way to build a film. Most of the time, the standard method of idea-premise-outline-treatment-script-shoot-cut-mix-release works well. It’s a very rational way to control costs and it’s well-understood.
Having now done several shorts in the opposite manner, however, I’m convinced that you should always leave room in your process for experimentation. It was only by using my tools experimentally – trying the camera’s video mode, fiddling with Photoshop filters, etc. – that I found the story. While every shoot requires some experimentation just because things never go quite according to plan, you’re far better off if you plan to experiment.
This can be done in a rational way. The Prayer Tree all told cost me next to nothing, so I had a great deal of freedom. If you’re shooting your first feature, you may only get a chance to play around during the rehearsal period, or when you’re doing your storyboards. But whatever the situation, cherish and make the most of your experiments. Most of them will be failures, but the ones that stick will be your greatest successes.