Help me flesh out an idea: time lapse photography cross country trip

In My Humble Opinion
1

I’m going to be driving across the country from Cleveland to Las Vegas in a few weeks, and I’ve got a new camera that I can hack to do interval/time lapse photography. So I was thinking about mounting it in the car to a tripod and then turning the whole trip into a 5-15 minute time lapse video.

I haven’t done any time lapse though so I’m not sure how to go about figuring out what my interval should be. If we round the driving time off to 30 hours (a reasonable estimate since I’m only going to be recording while on the actual road and pausing for stops), that’s 1800 minutes. If I were to take 10 shots per minute, or once ever 6 seconds, that’s a total of 18,000 pictures. If I play back those pictures at 30 fps, that’s 10 minutes of video.

In effect, because I’d be taking 1 shot every 6 seconds and playing back 180 shots every 6 seconds (30*6), the playback would be at 180x real time, right?

The numbers sound reasonable - a 10 minute video sounds like a good target. I’m having a hard time actually picturing what 180x real time would be like - it seems like stuff in the distance like cities and mountains would gradually get closer in a visually interesting way, but the stuff on the sides of the roads would fly by really fast. Other cars, unless I matched speed with them for long segments, would fly by. I guess that’s just the nature of cramming 30 hours into 10 minutes. Do you think it would look good?

The other route I can go is to do more of a montage - I could use a higher rate like 1 frame every second, but only do it during the visually interesting parts of the trip, like approaching cities and going through mountains, and skip stuff like endless cornfields in Iowa. I could still aim for 10 minutes this way, and it’d be a more fluid and detailed video, but there’s something cool about having the entire route, start to finish, one one long video.

It’s a 12mp camera that puts out 3MB+/- pictures, so I wouldn’t do it at full res, since I’ve only got a 32GB flash card. But I can just record it in 1920x1080, which is what the eventual video would be anyway, which come out to around 400KB, which would allow me to take around 60,000-80,000 pictures. I think. Am I mixing up an order of magnitude there? 32,000MB / .4MB.

Would 80,000 pictures seriously wear down my camera?

As far as camera positioning - I’ll be in a regular sedan and not a convertible or anything so I can’t put the camera above the windshield, which would be awesome. Should I mount it in the passenger’s seat with a littel bit of telephoto so you see more windshield than dashboard and roof? Maybe angled a little bit to the right so that more of the side view is captured and not just straight ahead? Should I mount it in the back seat so you could see more of a total picture around the car including myself, the whole wind shield, and somewhat out the side windows?

A bumpy car wouldn’t be the most stable platform, but I figure I can tie a bungee cord around the seat and hook it to the weight hook on the bottom of the tripod to make it reasonably stable.

Another cool idea I had is that my camera has a built in GPS that can tag each picture with its precise GPS coordinates. I was thinking it would be great if I could somehow find some software that would allow me to put a little map in the corner of the video, updating the geolocation of the camera in each frame, so you would see my progress across the country on a map as I went. Is there any sort of software that could do this and help me by automating the process?

Give me any ideas you’ve got.

2

It’s been done. I would look up a few of those and see what methods they used. Here’s one: http://boingboing.net/submit/2010/07/time-lapse-of-15-hour-road-trip.html?mobify=0

Assuming you have enough storage space you might as well err on the side of more frequent shots.

At the very least I’d want a photo at the interval between each shot and a point not more than a quarter of the “apparent” distance to the road’s vanishing point.

I would mount it inside on the dash and keep your windshield very clean. Also make sure it has auto levels. The light levels will be constantly changing.

Good luck! Post the results.

3

Yep, something similar to that. Except more fluid video, higher quality, mostly in the daylight, and covering a more diverse part of the country. I figured I wouldn’t be the first person with the idea, but I could perhaps execute it well.

15s seems too long - the detail zips by too abruptly in that video.

4

Just thinking here… I enjoy doing replay on my DSLR and watching a couple of weeks go by in jumpy silent movie fashion. It seems that the top replay speed is about eight fps --replay starts out slow and gathers speed as you keep the button depressed-- and that is just too fast. It’s better when each image registers as an individual element.

Playback around 1/4 second per picture might be right, so an 8 minute video would use about 2000 pictures. For a 30 hour trip that’s about 1 frame per minute.

I don’t think it’s necessary or desirable to always use the same shooting rate. A longer interval could be used at night, a shorter one during daytime travel. Much shorter could be used when people are in the shot and doing something.

Obviously this approach doesn’t provide smooth, ultra-fast motion. It provides jumpy/fast-cut motion. It don’t think you’ll get smooth and fast without extensive editing and without wearing out your camera. Professional DSLRs come with shutter warranties around 300,000 exposures. Cheap DSLRs have warranties around 100,000 or have no shutter warranty.

Experiment with playing back your images on the back of the camera and see if you like the effect.

5

Stop motion animators use some sort of frame averaging simulated motion blur effect to make things look more fluid and less jumpy. See if you can find a similar plug in for your video editing software.

And again, better to cut down from too many base frames in editing than trying to film at the exact frame rate your video is going to end up in.

6

I think, in a way, the hypothetical ideal would be to record full 30fps video the whole way and then cut it down to only a frame or so every few seconds later. Of course, there are a number of problems with doing it this way so it isn’t really feasible.

But keeping that “ideal” in mind, I think you should shoot a pretty high frequency of stills if your equipment can handle it. At least 1 frame a second, I’d think, if you can.

Try to shoot it as stills (JPEG, I would think) instead of some stop-motion feature of your video camera. The frame-sizes might be the same but the compression and other algorithms are different. Full-res stills take up a bunch more storage space, but they’ll look a lot better. You can change your storage cards periodically as you fill them up.

Once you get into editing you’ll have a lot of big files to deal with so you’ll need a robust system to handle it. Eventually, you’ll have to use some sort of compression when you get into video mode–the trick is to keep the stills at full-res as long as you can and only compress them once toward the end. Do not compress already-compressed files unless absolutely necessary.

I recommend making use of QuickTime Pro’s Image Sequence function. There are a bunch of ways you could incorporate image sequences into this project so I can’t really recommend an entire workflow without knowing more details. I just think it would be a useful tool for you to look into.

If you don’t already have QuickTime Pro you should probably get it. I think it’s only about $30.

7

I agree with this in principle, but as a practical matter it would be a lot of trouble to suddenly change the camera’s frame-rate because you’ve come upon some really interesting scenery (especially while you’re driving!)

IMO, it’d work out better to record the entire thing at one relatively high frame rate (say, 1 or 2 frames a second), then cut it down later, with different frame rates for different situations in the final video.

PS-- --some random things I thought of-- don’t forget to take your shutter speeds, f/stop, ISO, ND (neutral density) and maybe other filters into consideration. Visualize your final product so you can plan ahead to have the best settings for what you are trying to accomplish.

sorry— ETA-- when i referred to “full-res” earlier I didn’t necessarily mean the very highest resolution your camera can handle… just as long as you capture in a resolution at least as big as your intended final video resolution.

8

I think I’d approach it the same way in the OP, yet if it’s not too much of a hassle:

Set a long exposure (ISO) and a slow shutter-speed if you can.

So, if you’re taking 6 shots per minute, if you can get a half to a full second long exposure per shot (maybe even longer?), that’s getting a ton more “time information” into each frame, than you would at typical high-speed photography settings. It should result in giving you a more fluid, albeit motion-blurred effect. This effect would fall-off with distance, so the stuff on the horizon should still be crisp, and the fluttering of local cars and objects would be mitigated by the long exposure.

Of course this would have to be adjusted as the intensity of daylight changed. Perhaps you can come up with predetermined exposure / shutter settings for different light-levels (dawn/dusk; strong daylight; cloudy or overcast; night-time).

Pretty tricky, but I’d love to see the result… good luck!

9

Funny they chose an OK Go song for this… as I was thinking about this video by OK Go (End Love), making tremendous use of stop animation and exposure settings (all shot non-stop over the course of 24 hours).

10

Definitely not feasible - 1080p video in the basic h264 codec it records takes up about 16gb per hour. But even then, what would the point be? I’d end up with 30 hours of footage that I’d want to turn into a 10+/- minute video I’d end up discarding 95%+ of the frames anyway, so you might as well only take the fewer frames to begin with.

1 frame a second for 30 hours is 108,000 pictures. Possibly doable but at 30 fps playback that’d leave me an hour of footage. I guess I could cut up seperate “through the rockies” and “to [whatever city]” videos, and still do a shorter video at a reduced frame rate, but I’m starting to worry about the workload taking 100k or even 40k pictures would be on my camera.

Yeah, I’m not sure how much tinkering the camera requires. I haven’t hacked the firmware yet and seen the intervalometer scripts to see how hard it would be.

I’m starting to lean towards leaving it at a high rate (1 fps or so) but only recording the parts of my trip that are visually more interesting. It’d end up being a cross country montage rather than a log of the entire trip from start to finish. But it’d focus on the interesting parts and save me some camera wear.

I can test this out but I don’t think it would work - there’d be so much motion blur in each shot at that speed that it’d end up looking like a sci fi movie where you go into hyperspace. That… and while I will have the camera on a stable tripod, the car will still be bumping around on the road, so a long shutter time isn’t really practical.

11

My thought was not to play back every single frame as part of the finished video, but to have that 1 frame from each second (at least of the daytime scenes) as raw footage to work from. Then, during the edit, you can decide which portions merit some 1 fps montages, and which parts might only be worth 1 frame every 10 minutes–or even no frames at all.

Because you’ll be concentrating on driving you won’t be able to really focus on whether the current view is worth capturing at a high or low frame-rate. My idea is to capture it all (or maybe just the daylight stuff) at 1 fps so you’ll be covered, and then cut out the bulk of the uninteresting parts later (but still leave in 1 or 2 frames per 10 or 15 minutes even during the lackluster periods–just to continue the overall tracking of the entire trip).

I agree that you would probably not want to use seconds-long exposure times. You would just get blurry messes of smears. I’m thinking in normal daylight you might try ISO 100 or 200; about 1/250 sec. Adjust those settings to leave you an f-stop of maybe 5.6, f/8, or f/11, Use your neutral density filter (if you have one built-in) when your are in really bright light (e.g. you are driving into the sun or traveling through snow, etc.) to keep approximately within the above settings ranges. (you could also go to a faster shutter speed if you needed to, but if you have a bunch of bright glare it’s best to use a ND filter if you can).

In very overcast skies and twilight you might be looking at around ISO 400-800 (maybe more); 1/60 sec., and you might have to open the iris up all the way (to f/2 or whatever you have). I’m not sure that 1/60th of a second will be fast enough, and I would think anything slower would certainly result in some motion blur.

[NOTE: Those numbers are pulled out of thin air. Ballpark guesses that could be a ways different than what you will actually end up using. But they include some “safe ranges” that you might try to stay within to manage the highest quality shots.]

Setting exposure: Personally, I tend to shy away from automatic exposure when I can, but for this project you won’t have any choice because you’ll be in no position to fuss around with exposure settings while you drive. You might try putting the camera on aperture-priority auto-exposure and do some metering before you start and again when you stop for gas (or whatever) in order to pick an ISO and f/stop that will keep the shutter speed in a good range.

On the other hand, it might be best to use shutter-priority auto-exposure and let the camera pick the f/stop while you keep the shutter at say, 1/125 or 1/250. I’m of two minds about this. If it were my project, I’d have to see the “real-world” conditions (including the camera specs) before I’d be able to make up my mind how I wanted to handle the camera settings.

(my CV (kind of): I’ve made a bit of money shooting stills for newspapers and other professional applications, and shot a lot of film and digital pix just for myself. But the vast majority of my professional experience is in videography and video/film editing. I’m considering your project largely on what I, as a video editor, would want to have to work with (in terms of raw material) in order to edit and deliver a quality video. Sorry I wrote so much. :smack:

12

Some digital cameras have versatile Program settings that might provide the best solution, although I don’t know about the OP’s camera.

On my DSLR you can set the minimum shutter speed and the maximum ISO and let the camera use its best judgment, including backlight compensation. I don’t remember what happens when the light is too low for your chosen settings. Of course a DSLR will provide better low light performance than a P&S.

Another possible filming technique. Shoot at 8 fpm (about 16,000 picture in 30 hrs.) and then in editing vary between jerky motion and brief interesting stills (held on-screen for 1-3 seconds.) You could edit major chunks of the trip out without harm… take out five minutes (40 frames) of travel here and there, and still get the full sense of travel.

13

A couple of years ago I saw Aussie band The Sleepy Jackson in Sydney. They are from Perth. So think New York and LA. Before they appeared on stage a video was shown that was basically what you are talking about. It was time lapse photography of them travelling from Perth to the venue that we were at. It was only short because they flew to Sydney.

But the main point is, it was very cool watching it once you worked out what you were watching. And I still remember it so I’m sure you will be able to achieve something just as effective.

If it were me I’d do a few local weekend shoots first to find out what works and, more importantly, what doesn’t. You know, going to the shops, driving the freeway, out in the country.

14

I should add that what was best about watching the time lapse photography was the recognition factor. You get a bit of a buzz as you work out “where” you are looking at. The time lapse effect driving around a city is both disorienting and intriguing as your brain tries to keep up.