Need to put together a system with GPS and such

[JeepinDan]

My apologies if I'm posting this in the wrong place and please feel free to toss this elsewhere if it is considered inappropriate but I could use some expert advice in a video-dependent matter.

What I need to do is to take a low-lux video camera such as SuperCircuit's PC164CEX-2 and end up with a time-stamped video with great precision in the timing. I also have to be able to split the video with something like a Pyle PLV2 so that I can view the feed from the camera to make sure I'm pointed at the right star while the video is acquired.

The application is astronomical - not surveillance which means I want as great an accuracy in the timing as possible.

As best I can tell, this is typically done with the aforementioned camera or something like the WAT 120N+? They take the signal and run it through a Video Time Inserter and then to a DVR which is typically recording into .avi format.

Does anyone know of a reasonably low-cost way of doing this?

I don't think that I can save much on the camera end - the PC164CEX-2 is probably the best choice on that, although there are some arguments presented for going with the older version of the camera.

In a few weeks I'll likely be able to get a VTI for under $300. A monitor I can get locally, and a DVR I can also get. But the price and complexity of set up starts to mount.

So I'm wondering if I can reasonably get a good DVR/GPS combo or maybe a Camcorder/GPS which would do the timestamping and recording from the camera?

Or maybe figuring out how to hook up a GPS unit to a computer and acquire the image and timestamp into the computer?

Is one video imaging format going to be superior to another? I'll admit, however, that in this case I may not have much leeway for change. I think that the software I'll need to use to analyze the video uses AVIFile, DirectShow, or AVISynth (tries to use them in that order). The people who do most of the analysis tend to use .avi (so far as I can tell). But I keep wondering if MPEG-2 would be more precise if I could get the analysis to work on that type of file?

And if it is of interest, the particular application is to time occultations of stars by asteroids and by the Moon (maybe do some transits as well). Some occultations are measured in just the tenths of seconds total - so you can imagine that the more precise I can get, the happier I'll be.

Thanks in advance.

[Maccara]

What is "as accurate timing as possible"?

On a windows computer + gps you can get around some milliseconds accuracy, modified linux can get to microsecond range accuracy (according to some NTP studies), but that probably requires stable environment (i.e. no temperature changes etc) too. Nanosecond accuracy will require more hardware.

If within a second accuracy is enough, a networked computer with just NTP software can do that.

There's also specialized hardware for getting accurate timecode synchronization, f.e. this (http://www.meinberg.de/english/products/gps164.htm).

[JeepinDan]

I'd take the most accurate timing possible. As best I can tell, the most limiting factor is the frame rate. With currently available low-lux video cameras (at least ones I know of and can afford), the best frame rate I can actually hope for is 30 fps. So realistically I'll likely never really get accuracy better than 1/30th of a second. If there were some sort of affordable trick available to get much better accuracy I'd be delighted - but I don't really expect it.

But accuracy to within one second would be unacceptable. There are some occultations of interest which last for only about 0.3 seconds - so 1 second accuracy would encompass both the disappearance and reappearance of the star (as occulted by the asteroid) and would be fairly useless.

That hardware to which you linked looks amazing! It also appears to be unaffordable. . . I also forgot to mention that I have to go to where the occultation is occurring which means I have to supply the power from batteries, so 12VDC is preferable although I can use a modest-sized inverter.

That modified Linux idea looks really cool, but at a frame rate of 30 it's probably overkill. And in any case your caveat about temperature change is most relevant since temperature change is virtually guaranteed.

My dream world solution would be to figure out how to use my Droid X to acquire the video from the camera while using the Droid's GPS to insert a timestamp into the the video as it were acquired. Unfortunately, I strongly suspect that even if I learned the appropriate software that there would be a lot of difficulty with getting it to timestamp with sufficient accuracy.

I may be dreaming the impossible dream. I'll likely need to stick with the aforementioned camera, an available VTI, and maybe use a miniDV to store the video.

[Maccara]

Hmm. Since your limiting factors are actually hardware, isn't the best you can hope for a steady stream of data? By that you have ~33ms between frames +- jitter of camera/capture.

Do you actually need as accurate as possible time of day, or only durations? (I would think the latter, but I'm not sure)

I know practically nothing of the measurements you'd like to do, so I don't actually understand what the issues are (as I see it "anything" can achieve what you need easily - given that camera choice) so I hope maybe someone else can give better information. :)

[JeepinDan]

On durations vs. time of day? Exact time of day is very important. By determining the exact time at which the star "disappears" and then "reappears" you can help to determine star positions, the orbit of the moon or asteroid, and the contours of the solar system object. Most valuable if you have a bunch of people doing the measurements from somewhat different locations.

And honestly, I appreciate your efforts to be helpful. I really didn't expect anyone would have an amazing insight into how to do this better than is already being done - but there are so many bright people on this forum that I figured it would be a mistake not to ask.

[Maccara]

Ok, thanks for the clarification.

Nevertheless, since you're limited by your capture rate (assuming 30fps here) to 33ms resolution, you don't need much greater precision for time. Even a standard windows machine with GPS + NTP will do ~10-20ms accurate time.

For example, if your occultation lasted 1 frame, the best you can say is that it took 100/3ms (or less). Therefore, it doesn't matter if you can measure time to nanoseconds, as you don't have the "resolution" for that anyway (and VTI doesn't do any better than this either).

PS. I took a quick look how this "works", and it seems you rarely would need better than 100ms resolution.

[JeepinDan]

Thank you.

I'm going to check into the GPS+NTP thing a little more.

The help is appreciated.