Let's say 10hz clicks.  100hz?  Which is best?  How would this work?  Pseudocode please?  I'm not convinced.
Anyone else agree?

XBee is about 4x that speed.  What hardware can we use to compress the video and send it serially to XBee?
Small in size?  Low power?
Would be cool to see 10fps!

My DS1307 drifts a few seconds per day.  So if you require 1 second accuracy, sync with PC every hour.  Do the math.
How do you correct your PC clock?  By default mine seems to sync with time.windows.com once a week.  I have Windows Vista.
I don't see how to change this period.  It is off by more than 1 second. 

It CAN be done with XBee instead of an SD card.  The SD card is even easier, with a few lines of code.  It takes a few seconds to send a compressed 640x480 image at 9KB/s.  These results can be achieved with an Arduino and ZERO coding.  The hardware comes with a PC application that displays the image.  Just connect the modules and go.  The Uno is acting as a power supply with no sketch running.  The JPG camera talks directly to XBee.  This proves the concept is possible.  Then you can write the code to customize it for your application.

Frame rate?  That's for video.  1 "fps" is plenty fast for still images.  Low expectations?  We expect to see an image.  Yes it's 640x480.  Our expectations of the quality are realistic.  It's amazing to see whats happening 300m away without wires.  It can also detect motion or visual changes.  It's amazingly simple code if there is no movement normally.  It can ignore some leaves blowing in the distance as well as changing shadows due to the suns movement.  In fact it's only 1 line in my sketch!

It will probably not be reliable because the helicopter makes a lot of noise.  The plane does not.  This interference is why I suggested sending only from the craft using a clock to do it at a defined interval.  It is not so loud at 2m away.  I like the idea of a radio link!  What circuit can I use?  My first thought is XBee, but there is a variable delay in receiving a single byte.  It has to be low mass.  

@goforsmoke
I was thinking of making only 1 short click every 1 second.  Possibly 0.1 second.  You seem to have a different idea.  Please explain.  The distance wouldn't change much at 100hz, from one cycle to the next.  I assume we're talking 1-way transmission now, without a clock?

@Magician - Great idea!  Where have a I heard that? 

You're correct, the timing between pulses will always be 1s for a slow moving object in my example.  But it will be shifted compared to the clocks by 30ms at 10m, at the receiver.  All I have to do is keep the clocks in sync.  Or measure how they drift consistently without any movement/distance.

It will be easy to guess the temperature to get within 1-2%.  34cm/10m is about 3%.  34/5m is 6%.  Mostly I want to know if it's moving further away or closer to keep it at a constant distance.

What is the advantage of optical compared to 2-way sound?  The disadvantage is that light is much harder to point in the right direction from a moving rotating object.

I'd like 1 circuit to be simple and very light.  I'd like to measure distance up to 10m with 34cm accuracy.  Will this work?

http://arduino.cc/en/Tutorial/Ping?from=Tutorial.UltrasoundSensor
But different.  Here is my modification...

My idea is to put the sender in the small airplane or helicopter.  It will simply beep exactly every 0.1 or 1 second.  The receiver will store the milliseconds and a number of milliseconds it's drifting per ping when it's 1m away not moving.  This might be +-0.52ms, for example, which could be determined by waiting for 100 cycles or 100 seconds.  All I need is an accurate clock that's small.  Obviously the 2 clocks do not need to be in sync.  One can even drift by 1%.  As long as they are consistent.  Can I use a 555 for the sender?  Arduino Uno timer for receiver?  Do you think I will get 34cm accuracy?  Can I do better?  Can I share the main helicopter battery?  Or watch battery.

Sound is 343m/sec.  So 1ms is about 34cm.

I realized this will not work for long.  The code will have no way of knowing if it slowly moved away to 10m, or drifted 30ms over time.  It will have to be reset to a known distance every few minutes?

http://www.ti.com/lit/ds/symlink/lm555.pdf
Disregarding voltage and temp changes, it should continue to be accurate within 1%?
How much could it vary from 1sec to the next?  I think much less than 1%.

Any better ideas?  Please.

Update: Can I use the hardware in the first link together with a Nano or Picaxe to receive a ping and send one back after a short constant delay?  Is there a complete piece of hardware that already exists to do this for the consumer market?  Which idea do you like best so far?

Bueller?

2 cheap consumer MP3 players are much easier, and less expensive!

I did road defect measurement in the 90's.  A USB host works great with RAZR phones.  Not sure why you'd want to get rid of the Arduino.  Why can't you remove the phone instead?  Why do you need the data on a server in real time?  Why not at the end of the day?

http://www.dealextreme.com/p/usb-tri-band-gprs-modem-cell-phone-radio-gsm-900-1800-1900mhz-12057

I enjoyed building Crossroad's PCB!  Cheap price.

Or already built $10:
http://www.ebay.com/itm/Arduino-Nano-ATMEGA168-Mini-Module-nano-mini-168-5V-16MHz-/180745921091

http://www.ebay.com/itm/130613520642

Which might be roughly about 3 sec / cm with a pen as a pulley?
Perfect for timelapse!

An Arduino can control 2 MP3 players.  Just have it push the play/pause button on each at the same time.  You might also need the rewind-back-a-track button.  A player can be bought for less than $20.  To get really fancy make a single song playlist so it stops automatically.  Or figure out the exact time of each track and stop a second earlier/later.  Then sync with rewind/forward.  This should keep them in sync within milliseconds.

I understand the gearing.  I did not expect it to be fast.  But maybe speed is the issue I should be concerned about after all.  Using a pen as a pulley, how long would it take at full speed without a load?  1m/minute?

Thanks for the warning!

For focusing it should work when the beam is large and spread out, especially when there are humans in the picture.  When I am trying to focus on a subject ~100m away then I will want to focus the beam some.  Still it will be many inches in diameter.  It is much faster when it is tightly focused but this is not needed.  It will be wide when being used as night vision.  How can I control this?  A servo moving the lens?  Any ideas?