Avoid Shaky Camera Movements With Gyroscopic Stabilization

Right after “no editing whatsoever”, the “shaky camera” is the bane of YouTube viewers the world over. [David] came up with a nice solution to the problem of shaky cameras that uses gyroscopes to even out the bumps of making a great movie.

Most cameras attached to moving frames – from the zip-line cameras at NFL games to police helicopters and aerial reconnaissance drones – have some sort of gyroscopic stabilization. Even though gyroscopic stabilization has been around for more than 60 years, the designs haven’t changed that much. [David] dug up a few patents dating from the 50s and set to work replicating the design.

[David] bought a pair of [Glenn Turner]’s fabulously heavy and expensive-looking powered gyroscopes and began bolting them onto a piece of sheet metal. Per the 1950 patent, the gyroscopes were mounted 90 degrees apart and bolted onto the camera.

From the video demo (after the break), there’s a marked difference between filming a stroll though a parking lot with the gyros on and gyros off. While the pair of motor driven gyroscopes look a little ungainly and are a bit too loud for our tastes, the solution is much less expensive than the $1600 professional gyro [David] based his build on.

[youtube=http://www.youtube.com/watch?v=Wh8-xxW1WgE&w=470]

40 thoughts on “Avoid Shaky Camera Movements With Gyroscopic Stabilization

  1. I can understand the intent in this and it is an improvement over the shaking but it isn’t really stabilizing much over the jerkiness of his steps.

    I think the problem is that he he holding onto a rod directly attached to the camera which will inevitably send every little movement into the camera. Try this again but suspend the contraption, by a cable or a few cables (you may need a third gyro for the third axis of motion if you do a single cable) from another rod you hold above it and see if it is better. Nonetheless, I like where this may be headed.

      1. I think that is partially from walking and him using DC motors. He most certainly should be using steppers for something like this to minimize noise and vibration but that becomes expensive rather quick so I gave him a pass on that matter.

      2. @Zac D: Steppers would suck for this use. Gyroscopes are supposed to be running fast. The precision needed is the one of the speed in RPM, meaning you know/set how fast it turns, but you don’t know/care where it is right now. The precision in steppers is in the steps, meaning you know/set the position and don’t know/care about how fast it reaches said position.

        So the right choice for a motor would be a brushless one. These things can hold a specific speed in RPM without much effort. See hard drives for example. They run at e.g. 7200rpm. All the time.

      3. @Zac D
        Yeah…. steppers are the worst choice for this. They don’t work well at high speeds, they are complex to control, and they make *more* noise than a DC motor, not less.

  2. Yeah, i see no improvement at all. Picture still shaking like a flag on a pole within a hurricane.

    And it’s damn obvious where the problem with this build is: there is no hinge-ish part at all. So this means he still holds the camera in his hands making the two badass and loud gyros useless.

    The gyroscopic balancing in those ropecams works, because they have a floating mount…no not the ropes. The mount for the camera consists of three major parts: the dolly, the camera mount and the floating assembly. The dolly is attached to the ropes, the floating assembly is then attached to the dolly and the camera mount is mounted on the floating piece of said assembly so it can move wherever it wants. The gyroscopes are on the cam mount and stabilize it.

    Exactly this assembly is obviously missing in the build of david.

  3. I think the reason the video doesn’t show improvement is that the device doesn’t work as intended. Two design flaws:
    1) The axis of rotation of each gyroscope must be 90-degrees apart, and FIXED to the camera mount. In the picture, both axis are in line with each other.
    2) Those gyroscope frames appear to have an angle measurement device in the frame, which leads me to believe that they allow the gyroscope to pivot. This won’t allow the gyroscope to resist the “shake” of the operator. The gyroscope must be FIXED to the camera mount.

    The “after” video doesn’t look better, because the device essentially doesn’t do anything.

  4. OK, I get the need for 2 motors spinning in opposite directions to cancel out angular momentum. So when you walk forward the gyroscopic action does not act twist the camera to one side. But would better stabilization be provided if the gyroscopes were mounted at the bottom of an “I” frame, the camera sitting on the top of the “I” frame and the rig being physically held in the middle of the “I” frame. Yes the frame would tilt forward if moved forward too fast, but the weight of the gyroscopes, if heavy enough, would help to counteract this tilting.

  5. I have seen a slightly simpler DIY solution to this problem: A tripod.

    BUT WAIT – not stood on the floor, folded up and held. The extra weight hanging under the camera acts as a damper and makes movements smooth and moves the centre of gravity. There are even some professional setups which are similar, just a weighted handle that attaches to the camera.

    Battery life is infinite, noise emission is zero.

    1. I may be late to reply, but thank god I’m not the only one that just uses a spare tripod for free-hand video. It’s heavier than carrying just the camera, but the benefits far exceed even what this hack offers. The added weight acts as a motion damper while the extra mass below my grip further prevents unwanted angular movement. All I can say for everyone else is to just try it – BEFORE you spend time recreating a gyroscopic mount that doesn’t work nearly as well.

  6. FAILED build, maker didnt understand how it works, made a loud mess :(

    There was another project posted here couple of months ago using 2 hard drives in a wooden box – that one was working flawlessly

  7. I concur with the other posters.

    Although HAD now has a “be nice” policy for comments, it does works both ways.

    This is obviously a bad build and as such it shouldn’t be on here because the only thing we can offer is to point out its obvious flaws and not offer any praise.

    1. Then again “be nice” is one of those relative terms. I my opinion a reasonable person wouldn’t read that to say criticism isn’t allowed, but how the criticism is couched. Criticize/discuss the build/ hack constructively, don’t attack the builder as a person. We will never be privy to any comments HAD does reject, but I’d have to say the criticism they do allow looks to be based on a standard similar to what I offered.

      1. You’re right on. Just be constrictive and everything’s cool.

        Oh, the comments you’re not privy to are either spam (over 4 million comments in the spam queue, most selling sneakers), and obviously incendiary comments that I would describe as “12 year old with an attitude”. Other than that, we’re pretty hands off.

    2. Other than the noise I never noticed a difference at all, or what I may have notice may have been so slight it was not of any consequence. Anyway like most hacks the builder’s satisfaction is the primary goal. Unless it’s somehow a physical threat to others, the rest of the world can take it or leave it, a world that contains the DIY/ hacker community.

    1. Exactly this. The page does not show any pictures of the cam despite the moving on in the header. If you look closely you will see, the ropes are going to the piece in the middle. THAT is the gymbal mount for the rest of the whole thing.

  8. Ok, that is way too loud. Most of the noise is probably being referred to it via the gyro frame to the camera body at the mount. So, I’d use some sort of rubber between the camera and the gyro. I’d enclose the gyro in something to dampen any sound that would be picked up by the mic. I’d also use an off-camera mic to limit the amount of noise. OH! Also, I’d just not use this. I would just use a steady-cam rig. And those can be made pretty easily these days. Seems like a lot of noisy engineering for something that could be fixed with PVC, and counter-weight.

  9. I wanted to do something similar with old disk drives (as a prove in), and sniper rifles. The guys on the helicopters would appreciate that, but apparently, it’s already been done.

  10. What would be much quieter is to take the blades off a pair or three of high speed brushless DC fans then attach a weighted disc to the remains of the hub.

    I have some 3.5″ really fast fans rated for around 50 CFM but they’re hellaciously loud at 40db. They have a very strong gyroscope effect.

    1. The problem with trying to make your own spinning weight is if you get the alignment just a smidgen off it’s going to make the whole thing shake like mad, you really need the preciceness of an old VCR head or (more readily available) the motor & attached platters of a harddrive.

  11. I have tried this of a bombsight gyro. It was not balanced, 2 fingers from each hand were needed to aim it and fight when turning with some of the grace of a single engine pilot. Noise was a problem even at 400 hertz,but no vibration. A tiny camera and sound off cam it could shoot very active work.

  12. Most of the gyroscopic effect is canceled out by having two discs rotate that are not moved 90 degrees in two axis. Also, the frame is very heavy and the motors are not powerful. But it could work with a better frame. I experimented with harddisks, and the configuration shown in the post has zero to no effect ;)

  13. Maybe using gyro + rod, making sure the gyro is close to rotating pivot and the rod adding to angular inertia decreases the stress on the gyro to.

    Maybe a harddisk with a camera on top and a rod going down beneath the harddisk would be a simple and functional setup?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.