Simultaneous capture project

[agarbino 0]

Hello,

 

I'm trying to build a system that could allow synchronized recording of up to 4 video streams (+1 or 2 audio). I've looked at firewire solutions (point grey, avt) as well as analog to digital (going through something like a spectra8 )

The problem is that I need to get 24-30fps per color camera, good outdoor performance (Ie not washed out in sunlight) and keep the footage synchronized. Should I go with analog cameras converting to digital, or go digital from the beginning (eg firewire). Analog cameras are cheaper, but require a PCI board... and this would need to be portable (no rack-mounted stuff) and MUST work in outdoors/sunlight.

Can someone give any advice on wheter I should choose one path over the other?? Cost is obviously an issue, but both seem to be equally expensive.

Please also point me to forums/projects more specific to what I'm trying to do, I haven't found any that have done similar work...

 

EDIT: To clarify things, this is for research, not security/commercial. One of the trials we want to do is mount this on a car (think similar to racing cars, which have 3-4 cameras and 1-2 microphones recording simultaneously). Eventually, building this into a backpack would be neat for mountain expeditions (thus, I've been favoring something like a Mac mini w/ firewire devices, but this is not a requirement!).

 

- Alex

Edited March 5, 2006 by Guest

[KrCCTV 0]

Please provide extensive details in no less than 1000 words. Why, what, for whom and for how long. So far I'm leaning towards a recommendation of a home based DVR/DVD recorder like my DMR-EH50 fed by 4 quality analog CCTV cameras and one microphone through a four-way-split-screen quad with date/time-stamp overlay.

Edited March 5, 2006 by Guest

[KrCCTV 0]

Okay maybe 1000 is excessive - - 500 will probably work. Point is not to leave anything out.

[agarbino 0]

Ok, here we go:

 

I'm interested in doing some physiology research in peculiar environments/sports, mostly putting my interests (medicine) into some interesting work. I am particularly interested in fatigue/stress in sports environments, eg scuba diving (esp. cave diving), racing, flying, skiing. I'm building a modular system to acquire the relevant data in such environments. For example, I put together a basic 5-lead EKG (using RS232, now trying to switch to USB without the using serial-usb converters). I'm also capturing relevant analog signals and recording them (be they throttle position, altitude, depth, etc). But I think it would also be relevant to add to it video recording, and I'm not sure which way is suitable (analog cameras or all-digital).

Basically, I want to put together a "video module" that runs from a box and can record upto 4 video streams, with some audio as well, and synchronize the whole with other readings (eg EKG, etc).

The final product would thus allow me to play back video whilst simultaneously looking at whichever parameters I recorded.

Power is obviously an issue, but most devices run on 12v, which is what cars (and for scuba diving, DPVs) run on, so I'll have car-battery type power.

At the same time, I'm obviously trying to minimize costs, and I favor linux and embedded chips/microcontroller solutions (eg ucLinux).

 

Obviously, this is rather ambitious, but I think ultimately can be done for a 'reasonable' sum (and I'm not looking to commercialize this). Thus, I was thinking using firewire cards (eg the Point Grey research ones) seem best, since I can just plug them in and manage the data stream via linux to time-stamp, etc.

What I'm concerned about is the price and raw computing power needed to deal with the video stream, and size: whether I should go for analog system (which requires PCI cards, more power, etc but a PCI card could do the hardware compression) or fully digital (smaller size, parts count, but more processing required).

 

Let me know if this helps/more is needed.

 

Alex

[KrCCTV 0]

Thanks for the greater detail. Unfortunately I don't have much to offer (though I expect others will). My only thoughts are (all stuff you probably already thought about) that I think you'll need something with flash-type memory (low power consumption and ultimate vibration resistance) as opposed to a mechanical hard drive. If you find that you must have a mechanical hard drive (because maybe you can't get or afford the storage necessary in flash… then of course laptop hard drives are at least smaller, more vibration-resistant and smaller than IDE but I still see problems in this area – mostly power-related). I know folks who are working on single-channel solutions, similar to what you are talking about, for the military but of course they're not cheap. Aside from that though I would imagine that if they can do a single channel version then given more processing power, more electrical power, increased size and weight (and money) then what you seek is possible. As to imagers…“digitalâ€

[agarbino 0]

Can anyone else help?

[CollinR 0]

Why do you need the 4 cameras syncronized?

 

I assume a Geo card would suffice, send your serial data as an input overlay. I have never tried to syncronize across multiple inputs, I'm not sure how well it would work. Possibly add a heartbeat to your serial data, might be able to syncronize from that. If you don't care that cam 1 and cam 2 are 200ms off from each other it'll work depending on your data, geo was designed for interfacing serial reciept printers, it doesn't care though.

 

 

EDIT: Scratch below, I see what you want.

 

What does syncronize mean to you? +/- 1 second? +/- 1ms ? I think it will be pretty easy to an accuracy point but after that I dunno. I would think specific hardware controls would be required to get really precise.