bitingduck

I'd be interested in your results if you want to post them. Does the contour have adjustable shutter speed? At 1/2000 and 30 fps it's easy to read numbers in standard def as long as the light is ok. We ended up putting a couple PAR64 cans at the finish line so we wouldn't have to have all the overhead lights ($$$) turned on.

badhat

hey dudes

would this cam be a decent entry level finish cam?

in conjunction with a laptop
https://www.mediacollege.com/equipmen.../dcrtrv50.html

bitingduck

yep

botto 

iirc, somewhere in the 5-10K range, but that was including chips for 200+ people (at the venue that the KNWU used as a test).

botto 

very.

carpediemracing 

I think I will, it'll be interesting. I have to remember to do it soon.

badhat

zombie bump cuz i'm back here again

any PC software out there that will let me review video and capture at the same time?

bitingduck

Some kind of PVR software might let you do that -- most PVR stuff lets you continue recording while you rewind stuff that you missed or want to see again. I don't really do PC stuff, but there's a ton of PVR stuff out there and probably most of it will let you do that.

iconicflux

A Canon HV30 and a macbook seem to work very well. The important thing seems to be mounting it on a tripod about 8-9 feet up so you can get a better view of the racers. Oh, and do yourself a favor, make sure the refs remember to check the racers at the start for the numbers facing the wrong direction, being on the wrong side, etc. At the beginning of a season it seems that new racers are especially prone to having the numbers wrong and in some races you may have 636 and 939 or similar in the same race.

carpediemracing 

The early season races get to deal with this stuff. Our officials won't score someone who pins their number on the wrong side, wears a vest over, etc. I think a Cat 3 missed a 5th place because of a wrong side pin.

waterrockets 

Bump. With the new GoPro Hero 3 having a 240fps frame rate in WVGA means this is actually a potential solution. Vertical resolution at 480 pixels, but I don't think that matters much with a horizontal resolution fast enough. So, the question is, is 240fps fast enough?

Here's a thread I've been following over on POTN that has some free software for pulling slit functionality out of digital video, and a "bike" scan from a Hero 3... but at a pretty slow speed.

https://photography-on-the.net/forum/...6#post16228406

I ran the numbers at 38mph and discovered that a rider would be moving 7cm between frames. I guess another problem is that most automatic video cameras record at a shutter speed equal to the frame rate so the video looks right to the eye. If there is a way to tell the gopro to use a faster shutter speed, then maybe one shot every 7cm would be enough. Not sure how the Hero 3 video would work, but I'd like to see what a finish looks like going through this software. My thought is that since 1/240 shutter speed is not fast enough to stop motion (for 38mph, I'd be looking at 1/1000 or faster, probably 1/750 minimum)

Thoughts?

bitingduck

7 cm is a good fraction of an official race number size, and probably wider than a digit, so that's not enough to use the 240 fps as a linescan camera. Even with short exposure times if you're only looking at one line you won't be able to extract numbers most of the time. It's a cool concept though. I looked a few years ago (before I stopped promoting) and linescan modules faster than 1 kline/sec with GigE output were starting to be reasonable. You'd still need to get a good lens for it, and some kind of software setup, but I suspect with GigE output it wouldn't be hard to get the data out and stacked across a page.

You shouldn't bother with faking a linescan from an array detector-- you're just tossing out useful data. We did pretty well at the velodrome (and SCNCA at crits) with a MiniDV camera (standard def) that could be set to fast exposure times and just take full images to a mac, and scroll through the frames to pick out the places. If you're working outside you need a polarizing filter and need to adjust it as the sun moves to remove glare.

MiniDV + mac (and iMovie) is good because you can leave tape rolling while you review finishes in iMovie. That's especially useful for things like points races, but also for primes at crits. We'd get maybe one dead heat every few race days, where a race day has 25-40 times finishes have to be called (lots of intermediate sprints in points races). In a crit it's also useful for capturing groups as they go by so the officials can track riders every lap and make sure everyone is still there (yes, officials do that as much as possible) and whether they're gaining or losing laps, etc.

As far as shutter speed: 1/2000 s was recommended to me as optimum, and we ended up having to get a couple of 500 W PAR 64 cans to light up the indoor finish line to get that. 1/1000 was generally good enough to stop motion well enough to read numbers and pick places. 1/750 is probably marginal, but might be ok. at 1/500 you're looking at 3.5 cm motion during the capture, so the numbers will likely smear out.

waterrockets 

Yeah, that all makes sense. Would be cool if you could sync three Hero3 cameras to shoot between each other, similar to this: https://petapixel.com/2011/08/09/how-...h-canon-dslrs/

Of course, if anything other than a line sensor is a waste, multiple array sensor cameras are a bigger waste. I guess we'll just wait for prices to continue to fall.

Looks like Magic Lantern has a solid exposure override for Canon DSLRs that will let you move all the way to a 1/8000 shutter speed, but unfortunately only 65fps!! You can have one, but not the other. That would be a good solution, as the T2i is getting pretty inexpensive on the used market now, and with ML installed, is a very powerful little camera.

bitingduck

It's not at all true that anything other than a line sensor is a waste, just that you're throwing away useful data if you downsample a slow array detector to use it as a line sensor. A 240 fps camera would be awesome at just about anything local/states/regional if you can identify numbers and step through the individual frames without having to do any processing (i.e. just capture to the computer and start stepping). When a few cameras started coming out with those kinds of speeds ~4 years ago it looked like they didn't use frame based file formats, which makes them hard to use as a finish camera. Even a 30-60 fps camera works for most purposes if you can get a 1/1000s shutter speed. I have a $150 MiniDV camera that works great for local races.

But here's a page full of linescan cameras (need lenses still) that would mostly work and be less than a finishlynx. Least expensive is about $1500 and has 2048 px (plenty to put across a road) and ~50 klines/sec: https://www.machinevisionstore.com/C...gory=1&group=2

Not quite cheap enough for hobbyist territory, but close if you're going to start a finish line business.

carpediemracing 

I'm pretty camera illiterate. Can you put a link for a lens appropriate for that camera?

Personally the ContourHD seems to be okay, 60 fps, enough resolution to get the whole road.

I ran into problems with a night race. I had 1000w+1600w of spot lights along with a trailer generator/light thing. Not enough light and the Contour just slowed the shutter speed so I couldn't see numbers. It was a disaster. I contemplated some of the 10"-20"-40" LED light bars - I think 40" of bars will throw a lot of light and I won't come close to melting the power cords from the generator (they were so hot that we had to wait for them to cool before we could handle them).

However first I bought a much nicer Panasonic HD cam. For day races we have the shutter speed set at 1/10,000, I reduced it to 1/2000 at the Friday night race just to test it (I wasn't the finish line person and I'd just happened to read bitingduck's 1/2000 remark) but I haven't looked at the results yet. I think I spent $2k on it, I don't remember now, and in some ways I wish I spent more and got a replaceable lens one.

I would like to have a slit lens type camera, aka "a real camera". It seems the most legit even though we've never had a problem with the cameras we've used. It may be a luxury to think of it now but thinking about it is free.

bitingduck

Basler Racer 2048 px camera: https://www.machinevisionstore.com/C.../850#Resources

You have to pick an adapter from the accessories page:
https://www.machinevisionstore.com/Catalog/Details/863 (this is the F-mount, which is among the more expensive, but you'll probably find more suitable lenses with F than with C mount)

Most of the lenses on that site are super wide angle (12 mm and shorter) for machine vision applications. There are a few fixed focal length 35 and 50 mm. So I got lazy and looked at Lynx. They sell 12.5 to 75 mm zoomable, as well as 24 to 85 mm zooms. You could probably pick up one of those and it would work (might have to double check that it illuminates the whole sensor, but it should-- the racer has a ~14 mm sensor, which is on the small side). Or you could get a fixed focal length from that site:
https://www.machinevisionstore.com/C.../112#Resources
35 mm is a decent wide focal length that won't have a lot of distortion. When you get into super short focal lengths you get major fisheye distortion (which you can undo mathematically, but probably can avoid with just a longer lens).

As far as lights--what kind of spots did you have? You can get pretty narrow beam spots, and PAR 64 cans are ~$20 (the bulbs cost more than the can). And you have to mount them high and pointed down at the line (basically along the line-- probably above and behind the judges so you don't blind them) and test it before you do it live. They're cheap enough you can always add more...

And yeah, for me it was never worth the $$ and time to experiment with a DIY system when I had a $250 solution that did most of what I needed, but didn't look as cool.