Physics question: Maximizing moment of inertia.

[fuzzybabybunny]

Problem: When you handhold a camcorder and walk with it, you get a lot of jitter.

Solution: Camera stabilizers / gliders such as these smooth out the motion such as side to side rocking and forward and backward rocking.

http://www.creativeplanetnetwork.co...eoGraphy/Videography_News/GlidecamNAB-1bg.jpg

They operate on increasing the moment of inertia of the system. The handle in the middle is the pivot point. There is a set of weights down at the bottom and the camera sits at the top.

Now, I have a very very lightweight camera. Due to its low mass, it is actually quite difficult to keep it stabilized. Any small movement can knock it around. Also, the stabilizer is simply smaller, so the distance from the weight to the pivot point is shorter. Increasing the distance from the weight to the pivot point + increasing mass maximizes Moment of Inertia.

So my question is:

In the image above, provided that all three are perfectly balanced, which would be more stable while moving and which would be more accepting of operator error, such as bumping it around and the operator's arms swaying and footsteps bouncing up and down?

My thought is that the biggest setup would be far more stable because even if all of these setups are perfectly balanced, the added mass and distance from the center of pivot would make #3 more resistant to rocking.

The small one is "made for" lightweight cameras like the one that I have. But I have a feeling that if I get the big one, add weight plates to the top where the camera is, to drastically increase the mass, the setup would be harder to destabilize due to the added mass than the lighter setup, despite it being "made" for the lighter camera.

 

[Ken g6]

the added mass and distance from the center of pivot would make #3 more resistant to rocking.

I think that's true. But are you sure you want a hardware solution for this? Mass is heavy, in case you hadn't noticed.

I gather there are lenses that will compensate for motion. You'd know more about them than I would. They're probably too big for your little camera.

There are also software tools that will compensate for motion in video.

 

[fuzzybabybunny]

I think that's true. But are you sure you want a hardware solution for this? Mass is heavy, in case you hadn't noticed.

I gather there are lenses that will compensate for motion. You'd know more about them than I would. They're probably too big for your little camera.

There are also software tools that will compensate for motion in video.

Well a few issues with software and lenses:

1. A stabilized lens can only compensate for so much movement.

2. A stabilized lens isn't available for the focal length and camera body that I have.

3. Software stabilization will crop out areas of the frame in order to smooth the motion, which means the resulting video is more zoomed in and lacking in detail.

4. Sometimes software can't stabilize certain motion - it's good at smoothing out random jitter but is not good for keeping everything straight, for example (ex. horizon is perfectly horizontal)

5. Software takes longer to process in an already CPU and time-intensive situation.

Basically, it's often better to just get it in real life from the very beginning.

 

[DrPizza]

Wow, talk about an over-engineered solution. (I'm thinking of the extra weights added to my bow - far less complex & performs the exact same duty; plus, with the bow, precision of aim is far more important than with a camera.)

I'd go with more weight, but don't forget that there are more axes for it to rotate around. With all that weight at the bottom, your camera could still roll from left to right fairly easily (I'd think.)

 

[SKORPI0]

Why not DIY Stabilizer?

http://www.howtogeek.com/99267/build-a-diy-camera-stabilizer-for-ultra-smooth-video/
http://vimeo.com/10315887

 

[Phoenix86]

I'd go with more weight, but don't forget that there are more axes for it to rotate around.

Sound dangerous. :hmm:

Yes, I know it's correct.

 

[exdeath]

This is what we have computers for. Welcome to the digital age.

Use a imager that is higher resolution than the advertised "X megapixels" so you have a guard band, then you have a motion tracking algorithm that works the same way a 99 cent optical mouse does that tracks the image movement and still produces a X megapixel output from anywhere on the larger imager surface. The target shifts around on the imager, but the tracking software follows it and outputs the frames following the target resulting in no motion.

This can be backed up with feedback from accelerometers and a compass that aid in knowing when the user is moving the camera on purpose, or just by noting that the image is actually shifting completely off frame and isn't just unwanted slight motion. Image deltas that show up as a "vibration" about a center are considered unwanted, while continuous trends in one direction for a certain amount of time are a result of purposeful user movement. These deltas are ignored passed a certain threshold: if you are wobbling while you are walking, it will remain stabalized, but if you shake it on purpose you can exceed that threshold and "trick" the motion compensation filers. User accessible variable setting to control the damping as desired.

The DSP image kernels/filters and technology to track image motion is identical to that of which both a 99 cent optical mouse and a missile IR seeker achieves. It's trivial to implement in a modern SoC or ASIC.

This is one of the reasons that almost all "X megapixel" cameras are "X megapixel" stills and "X-n megapixel" video capture.

Welcome to the 21st Century... (though we still use 50 year old mechanical storage devices...)

 

[gorobei]

^ wow talk about overwrought.

in film and tv you just use overscan and bump the resolution you shoot at. matchmove software like boujou can clean up the path.

@ OP:
you dont need to add weights to the camera end, that just counteracts the stabilization effect. if you want a dead solid vertical with pitch damping you just need to extend the length of the counter arm(lower part with weights). the longer the arm and the more weight below, the more stable it will be.

if the small one can extend the shaft length enough, the inertial moment will be maxed enough for your purposes. adding the kind of extra mass you seem to be talking about will just make it harder to carry and more unstable as your arm fatigues from carrying all the extra weight.

http://littlegreatideas.com/stabilizer/diy/
johnny chung lee is the guy who modded the xbox kinect for custom uses and did some of the early surface projection user interface stuff at mit. for $20 you can get a ballpark idea of how these things perform for your uses.

 

[exdeath]

Isn't that what I just described (overscan and track)? Except real time in an embedded consumer device eg digital image stability.

 

[Murloc]

Sound dangerous. :hmm:

:biggrin::biggrin: thought the same. I just found out that despite the same spelling, they're pronounced differently. Irregular spelling wins again.
http://www.thefreedictionary.com/axes
interesting

anyway I think that if the camera weighs less it's better but I'm not sure.

 

[gorobei]

real time digital image stabilization will always have limits. you still need a steady cam rig to get you in the ballpark of usable footage. the post matchmove stuff just makes it feel like it was shot from a dolly.

look at all the handheld footage of stuff in the sky(bird, plane, ufo) no amount of digital processing will ever get the shakes out when you are at that level of zoom.

 

[fuzzybabybunny]

real time digital image stabilization will always have limits. you still need a steady cam rig to get you in the ballpark of usable footage. the post matchmove stuff just makes it feel like it was shot from a dolly.

look at all the handheld footage of stuff in the sky(bird, plane, ufo) no amount of digital processing will ever get the shakes out when you are at that level of zoom.

Another big downside is simply decreased field of view. The lens optically only casts and image that is so wide. Digital stabilization crops away part of the frame in order to achieve the smoothing effect, meaning my field of view is diminished. This can be unacceptable in certain situations