Centering object on spinning platform

[segment3]

I've been trying to come up with a good solution to a unique problem. I have a spinning shaft with a plate attached, and i need to place an object on that plate and have it be centered.

The shaft spin is controlled via a laptop and is used for photography - the problem we haven't been able to find a solution for is trying to perfectly center that product on the plate. Right now when the shaft spins, if the product isn't centered up exactly it is very obvious in the photos.

Objects placed on the plate very greatly, sometimes they are cylindrical and sometimes they aren't even a standard geometrical shape. I'm looking for any suggestions on how to solve this problem - we've tried several different methods and haven't came up with anything good.

 

[IRstuff]

What is you definition of "centered?"

There could be a number of different meanings, center of gravity, optical centroid, etc.

TTFN

 

[GregLocock]

Put a sticker on the object placed where you want the 'centre' to be. Make sure that that sticker is on the axis of rotation of the platform. Remove sticker. Take film.

I suspect, as IRstuff hints, that in practice what you mean by 'centre' is not a definable thing.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MintJulep]

Do you really need the whatever "centered", or do you need its surface to be a constant distance from the focal plain of the camera?

 

[segment3]

I would defined 'centered' as the geometric center of the object being placed on the center of the spinning plate.

Since the objects we're taking photos of aren't always square or cylindrical, we can't really keep the lense a constant distance from the object. Picture something like a sculpture of a horse with a base and trying to take photos from every angle.

Bottom line is we just need these series of photo's to look fluid and pleasing for people. Whatever achieves that result is what we're looking for. Thanks for all your suggestions thus far.

 

[scastillo]

Can the plate be modified so that it tapers down to the center? Picture a very shallow cone turned upside down. Any object placed on it will gravitate towards the center as the plate spins.

 

[segment3]

Scastillo - I don't think that would work since the center of gravity on the object isn't always the geometric center. Interesting idea though - thanks for the suggestion.

 

[ivymike]

How about using a curved "scooter" or shepherds' crook to pull the object toward the center little by little as it turns - while you watch through the viewfinder and decide when it looks "centered enough."

 

[segment3]

ivymike - That's something similar to what we've tried - it's a very time consuming process attempting to adjust an item as it spins and make it look right. 1mm makes a big difference in this project and it's hard to accurately make such minute adjustments as the product spins. Thanks for your suggestion

 

[IRstuff]

You can buy software that finds the geometric center of an object. Programs like Matlab can do that with some programming. All that's required is an image of the object in a clear, or bland, background. Segment, or threshold, the object from its background and run a geometric centroid of the object.

TTFN

 

[GregLocock]

I still the the desired centre is not necessarily definable.

eg take the letter C. You want the centre of the circle, not its mass centroid, or its geometrical centre.

I'd think about a remote controlled x y table if this is a big deal.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MintJulep]

What is your definition of geometric center?

I can't find a decent one anywhere. However, I seems that not all objects have one.

Greg's suggestion of an X-Y table is where I was headed when I asked about constant distance.

 

[IRstuff]

A geometrical center should be the center of mass of the 2-D projection of the object onto a plane.

TTFN

 

[GregLocock]

Ok, how about a teacup?



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MintJulep]

Russian, English, or Japanese?

 

[IRstuff]

Wouldn't matter. The point is that the teacup would be photographed, rendering it into a 2-D projection onto the camera's focal plane. Each unique orientation of the teacup's image would have a different geometrical center.

TTFN

 

[GregLocock]

And yet I'm sure the OP would want the rim of the cup to be concentric as it rotates. So no mathematical procedure will find this desired centre, in general.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[IRstuff]

Desires are, in general, not mathematical.

TTFN

 

[ivymike]

here's a procedure that might speed things up:

* Use a small aperture so that depth of field is large
* put a large thin "x" on the table
* put vertical marks every 90deg on the edges of a "coaster"
* put the coaster on the x so that the marks align
* put your object on the coaster
* turn the table and take one photo each time the x has an arm pointing at the camera (4 photos)
* Superimpose pairs of photos from 180deg orientations so that they appear to show rotation about the "correct" axis
* measure the distance between the marks on the coaster
* move the coaster 1/2 that distance relative to the marks on the table
* re-check

 

[ivymike]

..how to know which way to move the coaster?

If you did your vertical marks in different colors for 180deg orientations (ie red opposite green), then you should be able to figure out the right way to move the object based on whether the red vertical mark was on the left or the right of the green vertical mark when you'd aligned the photos properly.