Video for a Still Photographer
Posted by Tom Benedict on 26/11/2010
I’m a still photographer. I’ve known this myself for years, but I’ve also had this pointed out to me by others. The first time someone made a point of mentioning it to me was right after I’d finished a short course on broadcast video production. I think their exact words were, “You’re terrible at video.” Yeah. Thanks.
My class partner and I actually did make a video that year, and oddly enough it was shown on TV far too many times. It’s not because of its stellar production or camera work. I’m certain it’s because they were short on material and this one took up a full thirty minute slot. We’d shot video of a Physics Circus, an extravaganza of physics lecture demos all jammed together into one extremely entertaining event. Anyone who thinks physics is boring has never seen a pencil blown through a sheet of 3/4″ plywood by a fire extinguisher, or seen the kind of damage you can do with a giant Tesla coil. Physics is FUN! It’s also highly photogenic.
It could have been a lot more photogenic. Unfortunately one of the two people throwing the show was an uptight dweeb who didn’t want cameras in his face. So we were forced to keep our cameras firmly attached to tripods at the back of the room. Of course this didn’t stop someone else from using their camcorder handheld at the front, and getting all kinds of good footage of the event. When they reviewed it, their comment was, “There’s no motion to the cameras. You’re terrible at this.” Yeah. Thanks. That’s because you made us not move. The irony was lost on them. And as awkward as our video was, it still got more screen time than anyone else’s. Go figure. I honestly think it’s because ours was edited to length.
I’ve made some videos since then, but comparing those to my still photography, it’s still achingly apparent that I’m not good at video. I look at a really well produced sequence and see they have these sweeping crane shots or these dramatic focus pulls, so I try the same effects and it’s invariably bad. Comically bad. But that’s ok. Everyone who cares knows I’m a still photographer. When I make a video, it’s ok for the audience to laugh or groan at inappropriate moments. It’s all good.
Until now, that is. The final product of the PUSH N8 project is a video. A video. In the infamous words of Charlie Brown, “AAAAAAAAUGH!”
So I’ve been doing some homework. I went on a bunch of review sites to find out what the favorite, least expensive video production software was for Windows. Turns out I already had it installed as part of the software package that came on my computer. Score! Now all I have to do is learn it. No small thing, but at least I have a plan: In addition to being a still photographer, I’m also a masochist. And for a masochist, the best way to learn a thing is to do it badly, realize I’ve done it badly, reflect on that, and do it again less badly. Wash, rinse, repeat. So instead of doing one video, I’m planning on doing one video per week for the duration. The hope is that by the time I get to the final video product, I’ll at least be familiar with the software. Ideally I’ll also be a little better at video production, but I’m not holding my breath. At least by then I’ll be more intimately aware of my shortcomings, so I’ll be more practiced at feeling bad about them. But I hope I get more out of it than that.
To pull off the first video I want to do, I need to make a time lapse sequence. 60 seconds of screen time is 900 frames at 15fps, or 1800 frames at 30fps. Just about perfect for what I have in mind. I’ve done time lapse sequences in the past, typically at 60-120x speedup, and have bracketed them with a title card and credits and tossed them up online as being only slightly more interesting than watching paint dry. (Hey! Now there’s a time lapse idea!) So I know I can do time lapse videos. But a couple of things make this one different: First, it’s faster. The thing I want to video will be moving at human-perceptible speeds already. Past about 15x speedup, it would be just nonsense images crammed together. Second, I finally have video editing software that will do full 1920×1080 HD video! So instead of editing my frames down to 640×480 or even smaller for NTSC, I get to edit them to 1920×1080 pixels.
Obviously this presents several opportunities to screw things up. And since the action I want to make the video of is more or less a one-off (think: dropping a glass and watching it shatter, whoops no more glass!) I can’t afford to screw up. So I’m making a test video first: a glass of ice water melting. Ok, fine, so I’m back in the category of watching paint dry. But at 30x normal speed, that might actually be interesting! And even if this is boring as sin, it’ll at least let me prove all the steps necessary to get from point A (no video, no technique, no nothing) to point B (a working video and the knowledge that I can make this work). You get the idea.
I started by writing a CHDK script for my Canon A650IS that simulates holding down the shutter button. This generates a stream of 4000×3000 JPG files taken at a rate of about one per second. Next, I got a glass, filled it with ice and water, and got busy. No sweat!
As it turns out it wasn’t that simple. It took me four sets of images before I got one that didn’t have insurmountable technical issues, and even the fourth had bad white balance. See why I try to test these ideas offline before I try them on the real deal? R&D is great and all, but I like to use tried and true techniques when my ass is on the line. (Note to self: Camera loses some settings when power-cycled.) But I persevered and came up with a good set of images to work from.
Next I used Imagemagick to re-size them to 1920×1080 using the following command:
for i in *.JPG; do echo $i; convert $i -resize 1920×1440 -gravity Center -crop 1920×1080 HD/$i; done
These were then run through Adobe Photoshop 7 in batch mode to correct for the bad white balance. The bad white balance was due to operator error, and now the operator has learned better. Since I don’t plan to do this in the future, no documentation for this step. Besides, this was just plain embarassing.
Next I ran them through VirtualDub, which I’ve used in the past to do video de-shaking. Turns out VirtualDub will also do time lapse: Using VirtualDub for Time Lapse The codecs I have in VirtualDub don’t support compressed 1920×1080 video, but since these sequences are short there’s no real penalty. I saved it as uncompressed video and ate up several gigs of disk space in the process.
Title cards were added to the first and last images in the sequence and saved as 1920×1080 JPG images using Photoshop 7. In this case I used the Batik font, but I’m not sure what my plans will be for future videos. This part, at least, is old hat since I did something similar for my SPIE 2010 poster.
Finally, I ran all this through CyberLink Power Director to make the final film, which I then posted to Vimeo. Watch it if you dare. Dare to be bored, that is.