Stabilising in a confined space.

[Neirin Jones]

Hello all.

Ok so, doing a shot and we have epic and prime lens (ultra prime - small as possible) but we want to mount on a conveyor belt and let it go through a machine unmanned, maximum space is 2' x 2', which is fine for camera on plate locked off at a set angle. Problem is we are getting a vibration of the conveyor belt and we don't have the space to put in a fully stabilised head! Plus it would have to be stand alone as we can't connect any cables at all. Obviously we can take some vibration out in post! But, we would like to dampen the vibration as much as possible in camera.

Anyone got any ideas for this sort of problem?

Is there a small shock absorbing rig or something that does a job, if not properly stabilising at least reducing the issue?

Thanks.

[Chris Millar]

Not sure how much vibration you're taking about, but something cheap that has both spring and damping characteristics is sponge. Are you just sitting the camera in the conveyor? If so, that's your mass.

 

Google 'mass spring damper' ;)

[Neirin Jones]

Thanks Chris,

 

The vibration - hard to describe, but there is a 2 - 3mm spacer between every roller on the conveyor belt. Every time the camera hits a new roller, it bumps. This is displayed in camera as a small but visible and consistent vibration of the lens. We are just sitting the camera and a plate at a lock off position and letting it run through.

 

 

We are looking into sponge and some other shock absorbent materials. Thanks.

[Satsuki Murashige]

Have you tried putting the camera on a larger platform like a 1/4 apple or a pancake to spread out the bumps? That plus a bean bag under the camera should help smooth out the movement. Also, frame a little wider than you need and give yourself some room to post stabilize the footage.

[Neirin Jones]

Thanks Satsuki,

 

We will use a large base and a wider than desired frame for stabilisation in post. Guess all that remains is to find the right material to stick in between.

 

Anyone tried memory foam before?

[Chris Millar]

From an engineering standpoint, you want to use 3 points of contact as the conveyor belt when modelled as a plane is defined by only 3 points in space, hence any further points of contact - for example: 4 (a common choice for some reason) - will only over define the system. Each point would be placed as far apart as the following factors will allow: the length and width of the the conveyor; the effective length of the conveyor used in the shot; and the shot itself (i.e. keeping apparatus out of shot). In doing so any motion at the camera (in the middle) is effectively levered/averaged into a smaller amplitude. Each point of contact would ideally have a really nice linear bearing on it and a way to add adjustable springs and dampers, although that is the expensive and very likely unnecessary part, as mentioned earlier: it turns out many cheap products already have a good spring/damper characteristics - you just gotta experiment and find the right one. Memory foam why?

[Gregg MacPherson]

Four contact points will cope better with unusual loads. A rectangular piece of foam might allow a distribution of contact points with lower loads. Soft foam. An easy way to separate the uneven vertical loads of the conveyor from the camera is to make the camera unit heavy, with high mass moment of inertia. As a notion, both the linear and angular (rotational) disturbances could be visible on screen, but my bet is that the angular disturbances are the most noticable. The higher mass moment of inertia is easy to achieve by placing masses in the camera system as far as possible from the centre of mass (CoM). Place the film plane/sensor at the CoM.

 

The contribution of each mass m, at distance d from the CoM, to the mass moment of inertia is md^2. There are clearly 3 axis for this rotational behaviour and the most critical will be the pitch axis (nose up/down). On the conveyor one can easily rig a system with separated fore and aft masses to give a high moment of inertia in the pitch axis.

 

Easy to try. Mount cam to centre of a curved piece of plywood. Laminate it or bend and staple it to match the conveyor? Layer of soft foam. Have a range of foam densities on hand to try. Attach masses in a crusiform pattern to the base.

 

Inertia should easily be enough. Mostly rotational as described. Easy to experiment with it and find out.

Have fun.

[Chris Millar]

A camera isn't an unusual load - we put them on tripods all the time ( :blink:)

 

4 points only 'work' because the materials involved deform, 3 is the physical reality, at least at the level of mechanics.

 

Foam is I guess a bazzilion points, with very high deformation.

 

I would have thought that talking in terms of mass moment of inertia implies a system that is free to rotate - i.e. 'frictionless' or non-constrained - putting a camera on um, anything is well, not that (?).

 

As we've both pointed out, it can be discussed in an engineering context, but the reality will be wood, alu channels, glue and your mom's favourite cushion :D

[Satsuki Murashige]

I think memory foam may be too soft to hold the camera solidly, the camera may shift if it registers a bump. Why not use a camera specific bean bag like the Cine Saddle? It works really well.

[Gregg MacPherson]

various stuff

:D

 

Not to be an argumentative mofo, but.....

 

Unusual loads meaning something unexpected from the conveyor. I don't get how you might think I'm concerned about loads imposed by the camera.

 

Three points are the minimum definition for a plane. This is a conceptual reality, and has a few literal physical applications. How many three wheeled cars do you see on the road?

One could make a three point version of my suggested rig, but one would need more care. It will probably be built by people who are not engineers. Using a rectangle seems a safer route to me.

 

To clarify the linear inertia, rotational inerta relative to a camera system. I could be way off here, so if someone has designed or analysed inertial stabilization systems please speak up. Linear disturbance of the camera, say a vertcal bump, may be visible on screen. Small angular (rotational) disturbances will be more easily visible. But these small angular disturbances can be damped simply with (angular) inertia if the mass distribution is good. I think the pitch axis needs especially to be inertially damped, and this is achieved simply by shifting the mass elements away from the centre of mass (CoM). Like a generic steadicam camera with sled, the mass is concentrated at the camera at the top and the batterys, monitor etc at the bottom, with a good distance between them.

 

For the conveyor belt rig, if the inertial masses are sitting against the foam, will that soft constraint stop them from working. I didn't think so, but one could try a smaller piece of foam and or simply extending the masses further away from the CoM. Having the masses extend further from the CoM than the foam may be the key to it.

[Chris Millar]

Conversation better had with white board, couple of beers and a papaya salad.

 

3 wheeled car:

 

http://www.topgear.com/uk/videos/clarkson-tips-over-reliant-robin

 

 

Note that it is fitted with a 4 point safety harness

 

I love that car by the way (!), although I'd probably spring for the Messerschmitt KR200, given the choice.

[Gregg MacPherson]

@Chris M.

I didn't get to the Top Gear show (internet slow), but I enjoyed being reminded of the little Messerschmitt. It's a sweet little car.

 

Four point safety harness? Like in a glider? Some have a fifth anchor point just below your crotch as well.

[Dom Jaeger]

As the 3 wheel car idea illustrates, 3 points are fine until you add motion! My personal favourite odd- ball car used only 2 wheels combined with gyroscopic stabilisation:

http://www.aqpl43.dsl.pipex.com/MUSEUM/TRANSPORT/gyrocars/schilovs.htm

Awesome! But probably not much help with this question..

[Chris Millar]

When you get a chance please do watch it - from wikipedia:

 

"In the 18 February 2007 episode of Top Gear (Series 9, Episode 4), a Reliant Robin was used by Richard Hammond andJames May in an attempt to modify a normal K-reg Robin into a reusable space shuttle. The booster rockets separated cleanly, but the fuel tank did not detach, and the Robin crashed into the ground and exploded. This launch was the "largest non-commercial rocket launch in European history."^[6] In a subsequent episode of Top Gear (Series 15, Episode 1), a 1994 Reliant Robin was used by Jeremy Clarkson to drive 14 miles from Sheffield to Rotherham. He described driving it as dangerous as "inviting your mum 'round for an evening on chatroulette," and that "(the Reliant Robin) wasn't funny, it was a complete menace." During the segment, Clarkson rolled the Robin at least six times due to the odd-sized wheels and the weights which were attached to the car to allow filming, before having front support wheels mounted for safety.^[7] The following two episodes featured racing driver The Stig and Ken Block on their Test Track in Robin, which neither of them could finish a clean lap, and rolled over like Clarkson. The cars used in the these episodes were extensively modified. All the reliant robins on top gear were fitted with bigger wheels on the passenger side and front to make the unstable, this is why the car usually rolled on the drivers side."

 

V entertaining.

 

I wasn't much of a Top Gear person (at all) until my partner showed me these two clips.

Edited July 3, 2014 by Chris Millar

[Chris Millar]

But back to the topic at hand - as I imagined it the motion is mostly up and down and not left/right, fore/aft...

[Gregg MacPherson]

...My personal favourite odd- ball car used only 2 wheels....

 

 

That's like a large motorcycle with added gyros. Very cool.

[Gregg MacPherson]

But back to the topic at hand - as I imagined it the motion is mostly up and down and not left/right, fore/aft...

 

Meaning motion of the camera? Up and down meaning orthogonal to the conveyor movement? The small disturbances orthogonal to the conveyor path are what I assumed were the issue. Unless these loads pass through the CoM of the camera rig they will produce a moment, hence an angular displacement, very visible on screen. Hence the idea of reacting those moments with inertia and cushioning with foam.

[Gregg MacPherson]

 

Still sans beer or salad but here's my as-if white board.

 

This is what I tried to describe. The left and right masses (m3, m4) may not be needed.

Edited July 3, 2014 by Gregg MacPherson

[Gregg MacPherson]

Just realized that m1 is probably in shot. Masses m1, m2 could be shifted vertically to fix that, or, better, shift them, with their support rails, to the rear. As m1 gets near to the camera it's function is achieved with m3, m4 instead. Chris will be happy, there are then three masses. I think three masses is the minimum required to achieve inertial damping about all three rotational axes.

[Brian Drysdale]

You may find that you need to use a sandwich of foam with different densities, which will damp vibrations at different frequencies.

[Neirin Jones]

Wow, great response guys. Thank you.

 

I have to say some of the theory is a bit beyond me! :blink: Plus I don't think I really have the language skills to discuss the finer points of these issues with my Brazilian Grip! ha. Pictures on imaginary white boards for fathomed Bob Sled teams are much more accessible :lol:

 

Thanks Gregg and Chris.

 

One question though Gregg, at the (I believe we have concluded in this particular scenario) 3 contact points, what material would make the contact there, would I use foam again as used under the plywood?

 

Also thanks Brian, we are stockpiling foam as we speak.

[Satsuki Murashige]

#Nerds.

[Gregg MacPherson]

...., at the (I believe we have concluded in this particular scenario) 3 contact points, what material would make the contact there, would I use foam again as used under the plywood?..............

 

 

I'm not sure I understand that. My idea was a board hard mounted to the camera base with a foam cushion between that and the plywood "bobsled".

 

The simplest thing to try would just be the camera hard mounted to a heavy plank with a foam cushion on the conveyor. Probably have to shift the camera forward, then add mass at the rear. One could add additional mass at the front and rear. The plank could be (guessing) 1.2 to 2m long

 

When I say cushion I mean a layer of foam that has that function, not a cushion from the furniture, but who knows what you use in the end. I would initially try for the softest possible cushion under the whole plank or bobsled.

 

To be fair, we didn't use any numbers yet, but I agree, it is quite nerdy

[Chris Millar]

Would have thought that with the CoM is above the pivot point and with those masses at those levers arms it only has a stable equilibrium not much beyond:

 

'careful, careful !'

 

 

smash

 

 

The solutions to that of which I agree are entirely possible involve too much presumption about the conveyor and so on...

 

Without being there and assessing it, or should I say 'feeling' it (the liberal arts side of me kicking in) we can discuss our solutions to the (individually interpreted) problem to solve, debating at cross purposes and abstraction until we say enough, and suggest beer (yet again)... but that's already happened :)

 

With a bit of reading between the lines an observer could at least arrive at a boolean intersection of: use foam/sponge "with different densities, which will damp vibrations at different frequencies."

 

Nice picture though !

 

Here is mine:

 

 

 

[Gregg MacPherson]

@Cris M

 

You're just a fuzzyist (artist) pretending to be an engineer aren't you!

 

In my drawing what is the "pivot point"? I think it will be close to the system CoM. Could lower that relative to the conveyor if one has to, but looks OK sofar. If the the foam connection between the camera and the bobsled gets a bigger plan area to, as if stabilize the mass system, then the moments imposed on the camera from the conveyor will be higher. The idea in the drawing is ti isolate the camera from those. The mass system should be stable. If there was a tendencey for the viewing angle to diverge I would have a long, slender tube attached at the lens axis and let this slide through the operators fingers.

 

It's not our fault if there isn't enough info to give actual concrete solutions. But on the bright side, the more generic concepts that come up due to that are perhaps more interesting.