How do I measure how steady my camera is?

1. emomartian
   Member

   

   Hi,

   I'm doing a school-related science experiment involving how steady a camera is under certain conditons.  I need to find a way to measure how steady the camera is though.  Is there any kind of software that you put the video into or a something i can attach to the camera.  During the experiment the camera will be attached to a a tripod and in a controlled enviorment.  I have talked to someone in the filming field and she said that there is a way, but she couldn't remember how.  Does anyone have any suggestons?

   Posted 1 year ago # Login to Send PM

2. Aspyrider
   Member

   

   You would probably need a instrument that measures vibration. Check with your science teacher, the school may have a device.

   Posted 1 year ago # Login to Send PM

3. ralck
   Member

   

   Hrm... you could put accelerometers on two sides (like left side and bottom of the camera) to get an idea of how much it accelerates in up/down and left/right directions (you can put one on the back too for forward/back).  Then using that and some calculus you could find out how much it actually moved.

   Another method, that probably wouldn't be as accurate, would be to film a special "scale" which is a series of vertical and horizontal lines.  Then you "measure" (count) how many lines you "move" in the video footage.  Depending on your camera's resolution and how fine you can make the lines, you can get decently accurate, but probably not as accurate as the accelerometers.  Since you'll be using the camera's footage in this suggestion, you may want to do both with the stabilization off and on to see the effectiveness of the optical or digital stabilization.

   A third way, that I'm not sure how well it would work, is to attach some lightweight rods (wooden dowels would probably work well) on the camera in the same kinds of places as the accelerometers I mentioned earlier.  Then using scales (rulers might even work well enough for these scales) measure how much the dowels move along the scales (the scales would be measuring the furthest point away from the camera on the dowels).

   Those are my initial thoughts... there's probably a much simpler way...

   Posted 1 year ago # Login to Send PM

4. emomartian
   Member

   

   Thank you.  Could you please explain accelerometers a bit more?  I like the sound of it.  I'm to young to understand calculas but i might be able to find someone who does.

   Posted 1 year ago # Login to Send PM

5. ralck
   Member

   

   Acceleration is the rate at which an object speeds up or slows down (more accurately, acceleration is the rate of change in instantaneous speed of an object over time... but I don't think we need to get quite so technical :-P).  An accelerometer simply measures the acceleration.  Depending on the accelerometer, it can measure it in one direction or multiple directions.

   These devices can get fairly expensive, and sometimes you need special connections to a computer to record the data.  There are many programs that can calculate the distance traveled and/or the velocity of the device (it's simple calculus, but if the program can do it for you to the amount of sigfigs you desire, why not?).  You might want to check with your science teacher to see if they have any.

   Posted 1 year ago # Login to Send PM

6. D0n
   Member

   

     I'm confused....you said:

   "During the experiment the camera will be attached to a a tripod and in a controlled enviorment."

     On a tripod your camera should be rock solid, and have no measureable shake. 

   so you'd wind up testing how stable you tripod is?

   or perhaps how stable the ground (platform, floor etc) is?

   or are you testing to find out if your cameras image stabilizer will induce shake on a stable camera? 

   Posted 1 year ago # Login to Send PM

7. D0n
   Member

   

    for what it's worth....

   If I wanted to find out how stable the floor/ground was I'd put my video camera on a tripod.

   I'd set up a target with a grid (ie grid paper with a bullseye drawn on it) and center it in the viewfinder.

   I'd use my cameras still image capture to snap a photo. (using remote).

   I'd run the video for about an hour. (using remote).

   I'd open the still in photoshop and cut it in half diagnally and leave half transparent.

   I'd open the file in my nle and overlay it on the video.

   I'd speed the video up.

   on play back, any jitter in the video should be noticable, and measureable.

   Posted 1 year ago # Login to Send PM

8. ralck
   Member

   

   Don,

   That's a much better expansion on my idea number 2!  The only thing I would add to that is, if you are putting the camera on a tripod, you might want to set it up so you can zoom all the way in.  That will make even little jitters more noticable.  Of course, if the jitters are too big that way, just zoom out a bit.

   Posted 1 year ago # Login to Send PM

9. Coreece
   Member

   

   I would highly recommend using the motion tracking feature within After Effects.

   If you don't have it, you can download the 30 trial at http://www.adobe.com/downloads/

   -Once you have imported your footage and dragged it into a new composition, go to Window>Workspace>Motion Tracking.  These buttons are located in the basic tool bar drop down menu.

   -The motion tracking panel will open.  Click the "Track Motion" button.

   -A small box with crosshairs in center will appear over your video called "tracking point 1".

   -Drag the box over a small stationary object in your video that has a high contrast to it's surronding colors.  Make sure the crosshairs are right at the edge of the border of the two contrasting colors.

   -Make sure the cursor in the timline is at the begining of the video and click the analyze forward play button in the motion tracking panel.

   -When the video is finished playing, twirl down the video's properties in the timeline.  You will see an item called "tracker 1;" twirl down it's properties.  You will see another item called "tracking point 1."  This is where all the tracking data is recorded.  

   -The data is stored in keyframes that are recorded in real time at the rate of 30 keyframes per second.  Each keyframe contains the X&Y coordinates for the positon of the stationary point you chose in realation to the movement of the camera at that specific frame.  It also gives you keyframed information on the accuracy or "confidence" of its tracking.  If you choose a good high contrasted point, there should be no problem obtaining 97-99% accuracy or reliablity.

   -The keyfames are located in the items named "feature center" "confidence" and "attach point"

   -You can than compare the X&Y positon coordinates in the "feature center" to determine the steadyness of the camera.

   -There are 1800 data sample reorded every minute, allowing you you determine the direction of the movement and the point in time that movement occured down to a 30th of a second.

   -If you don't have the ability to analyze every frame, it should be acceptable to compare only the position coordinates given at each second.

   -The tracking data is also visually graphed over your video for an additional reference.

   Good Luck!

   Posted 1 year ago # Login to Send PM

10. emomartian
    Member

    

    Thank you, those are very good suggestions.  D0n, i think i'm going to test out yours.  Thank you so much, I am testing how steady the camera is under conditions (in this case altitude) which will cause swaying of the ground below the camera.  Thank you so much.

    Posted 1 year ago # Login to Send PM

11. BarefootMedia
    Member

    

    I just thought I'd add support for the graph paper idea with a little twist.  Attach a laser pointer on the camcorder and aim it at the center of the graph paper.  You will have easily quantifiable results if you draw a spot on the graph paper that is where the laser rests at the start.  Now in any frame, if you measure the direction and distance between the spot & the laser point then you will know how far the camera has moved from the start or the magnitude of the movement.  By comparing the distance between the frames, you can calculate the acceleration.  It is easiest to use polar coordinates for measuring, but XY coordinates make calculating accelerations easier.  If you have trouble with the math, I can show you how it works.  Just remember, when you watch the video, the laser will appear steady and the graph paper will appear to be moving.  Measurements are made from the laser spot to the center target, using that value and knowing the distance between the camcorder & the graph paper, we can calculate the angle of the camcorder's deflection at any given point.  Sorry I'm getting carried away.  I really love math, especially when there is an actual purpose to the numbers.

    Have fun with your research.

    Posted 1 year ago # Login to Send PM

RSS feed for this topic

Supported video provider:

youtube, myvideo, funnyordie, gametrailers, collegehumor, dailymotion, glumbert, liveleak, redtube, googlevideo, sevenload, metacafe, clipfish, vimeo