Monday, February 20, 2012

Promoting - Finish Line Camera Test

I don't own a slit lens camera, one that makes those funky finish line pictures. At Bethel we rely on a more traditional camera with a frame by frame look at whatever sits in front of the lens.

Since the finishing hill slows things down nicely, usually a fast finish is only 25 mph, maybe 30 mph tops. I know I've won field sprints while crossing the line at about 25 mph, with a max speed of 35 mph at the bottom of the hill.

At any rate, with some of the new rules coming into effect (like Cat 5s have to finish ten races to upgrade, not just start ten races), along with the realization that our current finish line camera is non-HD (albeit a very, very nice non-HD camcorder), we had to improve our place picking abilities.

In 2011 we used an iPhone (HD) as an experiment, set up in parallel to the Canon GL2. We found the race bib numbers much easier to read in HD, but with the low frame rate we couldn't tell who crossed the line first. We could read the numbers, yes, but we couldn't get the placings.

Using the low-res high frame rate camcorder allowed us to pick the order; the HD iPhone gave us the number.

Logically this led me to an upgrade plan for the following year. I decided that for 2012 we'd move into the HD world.

Initially I thought this meant buying yet another expensive camcorder. I knew I needed a relatively high frame rate, and I thought I wanted 1080p resolution. If we could get away with a slightly lower resolution (720p) then I had an even better idea - ContourHD cams.

I should point out that we've dropped every single camcorder we used for Bethel. Not just a drop from a table, we're talking a drop from the ten foot high tripod we use. A teetering, tottering, "Oh snap!" kind of pregnant moment, followed by the smash of the camcorder.

We've dropped three of them, resulting in three somewhat crushed cases. Fortunately the units still worked, so, for example, the latest one, the GL2, has soldiered on successfully with a crushed mic boom, some case damage, and a ruined and illegible LCD screen.

On the other hand, I've fallen with the ContourHD cam once, at 30 mph, and had no damage to it. It survived with just a few cosmetic scrapes, some quite deep, but functionally it worked fine. I figure that if we drop a Contour from 10 feet up chances are it'll still work fine.

The Contours are also very light, about half a pound each. This beats the one to two pound camcorders, and the even heavier GL2. With less weight and virtually no wind profile (it's only marginally thicker than a tripod leg), a Contour won't be quite so attracted to the ground.

The big question for us was resolution. We knew 60 fps would work great, but we weren't sure if 720p would be enough to read a number.

I remembered that when SOC got fake-swiped by a truck we could easily and clearly read the truck's plate. That was in 720p, and if I could read a plate of a truck passing us at 20 mph above our speed, reading a five inch high number shouldn't pose a problem.

I had only one way to check this - a test.

Luckily I had a few cycling friends over the other day and they helped me test. Initially I was going to give each rider a number and have them sprint up this short rise past me at 45 mph. Because, you know, they're uber-sprinting machines.

Uh, okay, I'm exaggerating just a bit.

I'd have them start down a steep descent leading into the complex where I live, pedal a bit to hit at least 35 mph, and fly by me, pretending to throw their bikes and such.

Holding my two cameras, one set at 720@60fps, the other at 1080@30fps, I'd record the "finish".

Then we got caught up in talking and such and suddenly it was time to go and we were still in street clothes. We quickly changed the plan to "let's tape some numbers on a car and drive by the camera."

One of the guys that helps with Bethel volunteered to drive his car, with numbers taped on it, past me and my two Contours. I set one at 720@60fps and the other at 1080@30fps (the max fps for each resolution respectively). Since I have one camera in each hand, both hands next to each other, there shouldn't be any argument over "differing conditions".

The other guys made smart comments and such.

"That's 35 mph? That seemed pretty fast."
"He should drift around the bend."
"He should do a burnout when he launches."

I posted the two tests to YouTube since I was having problems attaching a 14 MB .mov file to email to all interested parties. Now everyone can see my test results.

720 at 60 fps:

1080 at 30 fps:

To put things in a bit more detail, here are some pertinent stills from the two clips. I'll start with the 720@60fps, two frames taken one frame apart.

First frame. 720p@60fps.

Second frame. 720p@60fps.
Note the peanut gallery in the background, Joel and SOC. David is driving.
And also note the awesome Renault car back there. Belongs to a neighbor.

You can see, from the oil stain on the road and the wheel position, that the car only moves about half a number worth of road between frames. That's pretty good, considering the car just blew by us.

A single frame of the 1080@30 fps.

After looking at all this stuff, I think we'll be going with the 720@60fps set up. Obviously if there are major problems we can go with 1080@60fps, but with at least two Contours on the tripod things should be okay.

I should point out that the Contours feed into a Mac. I found that the video appeared inconsistent when viewed on a Windows machine. In the Mac Quicktime lets us review clips a frame at a time, quickly and without any fuss or muss.

4 comments:

Anonymous said...

Sorry to burst your bubble, but that can't be a true test for Bethel.
1)one of the numbers has to be upside down
2)another number has to be crinkled beyond recognition, or completely folded over
3)the last number has to be taped to the other side of the car, cuz you know, "that's my other left side".

Aki said...

I knew I forgot some testing stuff. We're on it, don't worry.

Anonymous said...

Another way to do this, if you are just trying to read numbers and not really worrying about placing, is to use an SLR with a high speed shutter. Then you an use your old system for placing.

Aki said...

I thought of that too, but to get 10 feet up, aim it, hold it steady, etc, would be difficult. With a relatively top heavy set up, it'd be a bit more topple prone.

Finally, because it costs more, it would be hard to have 2-3 cameras.

One thing is that I need to work on the assumption that the camera will hit the deck, if not while on the tripod, at least in the process of transporting, loading, unloading, etc. To have a durable camera is key.