Re: [CAD_CAM_EDM_DRO] Re: some more milk digitizing ideas

Graham,
You're spot on about the difficulty of relating the incremental volume of milk to the incremental milk surface height that results. Philosophically, one can say "if we know enough about the surface that we can assess that increment/increment relationship, then possibly we may know so much that we don't need to scan it in the first place". That sounds a tad negative, but it's a truth that has guided me to study principles of determinism and strategies of opportunism in 3D digitizing.

An even worse situation arises with geometries that have the potential to trap a volume of air that the milk won't fill. And of course, areas that do fill with milk but are invisible to the sensor due to undercut/overhang structure, create a nightmare involving erroneous assumptions of "where the milk went".

In assessing surfaces for a scan, experience in a former military-related task that involved stereoscopy from high altitude has taught me a great deal. But to this day I struggle to find any decent, universal technique for what seems like a practical, 'doable' task. I think the Milky Way (hehe) is loaded with fun and promise for uncritical artistic scanning. I loved the Roland ultrasonic needle-probe and sobbed quietly when they canned it. And none of us can afford nuclear magnetic resonance (NMR / MRI) or other e.g. X-ray penetrating methods that allow lovely tomographic computer models to be built, ready to be reproduced in your powder-builder.

The entire scanning subject is huge - too large for the forum - but ideas like the Milky Way of scanning are so thought provoking! Imperfect yes, but amazingly useful and worth the mental effort to implement.

For so many scan tasks, if I take the view "my method gives pleasing results that look great to human observers", then I can ignore many of the technical faults that could be measured by a comparator. Sort of a "glass half full - of milk" attitude.

Mark Bingham
http://www.fourth-axis.com/rotary-4th-fourth-axis-pix-30/

Graham Stabler <grezmos@...> wrote: --- In CAD_CAM_EDM_DRO@yahoogroups.com, Kevin Martin <kpmartin@...> wrote:

>
> As long as you know how much fluid you are adding each time, you
>(meaning the program, not you personally) can calculate how thick

each >"slice" is. As long as there are no sudden changed in
cross-section >(i.e. no surfaces parallel to the surface of the milk)

An interesting idea, I've been reading about this sort of thing,
stereology, the best way I can think to do it would be by using a pump
with a very constant flow (such as a syringe pump) then capture video
of the filling process. That way you don't have to construct a
machine vision system.

The problem you highlight (horizontal surfaces) would seem to put a
dampener on the idea, as many parts have them.

Graham






---------------------------------
Building a website is a piece of cake.
Yahoo! Small Business gives you all the tools to get online.

[Non-text portions of this message have been removed]