Optical measuring experiment.

[skypilot]

by setting the camera on the cross slide you could use it to " measure "existing parts. It work great on a lathe that was equipped with a DRO. You could do the same thing on a mill by mounting the camera on the head.

[Toolmaker51]

Tying measurements to what produced them is logical. Measuring via camera [physically viewing] very dependable, providing aim is at correct feature.
But my preference would attain verification using something not dependent on same instrument; ie gauge rods with micrometers.

[skypilot]

I was thinking about measuring something you were trying to make a copy of, for instance a mill collet.

[skypilot]

I often need to make parts that bolt to something with an odd bolt pattern. I could lay the existing part on the mill table and use the camera and the DRO to measure the location of the holes in relation to each other or some conmen reference point. I can do the same thing with a dial indicator mounted in the drill chuck but it's a pain.

[Toolmaker51]

I often need to make parts that bolt to something with an odd bolt pattern. I could lay the existing part on the mill table and use the camera and the DRO to measure the location of the holes in relation to each other or some conmen reference point. I can do the same thing with a dial indicator mounted in the drill chuck but it's a pain.

Understood. You are correct, a mill can sub as a coordinate measuring machine. The meaning of my post is checking a part, a little different than reverse engineering. It all hinges on selecting correct reference datum, determine tolerances, and utilizing that to produce either a twin or modified as intended.
But good as optical measuring is, software or properly handled mathematics are only guarantee of success.

[PJs]

All good points of use for Olderdans system and agree that hard measurement is still a foundation for in process. To me it's not about making his system into a "All seeing" end all to producing things on the mill or lathe...the point is getting the most out of a cheap and cheerful build, cheap and cheerfully...ie., not trying to build a CMM or high end vision system but something that can be used regularly for lots of ops and even some rudimentary QA with relatively good resolution...for lots of HMT's folks and others.

I checked with my son and his system did use an x-y table as the parts were tiny and needed 2 views (length profile & Cross section for ID/OD) with datums to measure and record. The software used a crosshair with feeds from the x-y coordinate table and push buttons for datum and measurement recording. So that's not going to work for this, imho.

I still think a Gif/PNG or Tiff reticule/graticule might be doable...but would need to spend some time with it...something I'm a bit short of for the next while as a big video project just dropped in my lap along with about 6 other large things going on. I'd be more than happy to generate the Gif or Tiffs if someone wants to help figure out how to use it with the cross hair or stand alone, then test them. When I get a few or in need a fun diversion, I will try the crosshair program...its an MSI install and don't know the details of how it ties in with webcam drivers but could see.

Love the idea of this! Inspiring OlderDan, Thanks again!!

PJ

[clavius]

I will echo everyone else on the nicely done implementation here. Very handy. There are many USB microscopes out there now for very reasonable money that can put this capability within the reach of pretty much anyone who wants it. I messed around with this concept a bit some time ago but never really went all that far with it. May have to revisit that now. I do have an old machinists microscope with an x-y measuring stage, perhaps it needs a camera now...

Related to some of the other discussion here, there are a couple of other software packages that some of you may want to check out that have application here. One is called "MICAM" and can be downloaded for free here:

Science4all - Microscopy and Photography

This allows you to do all sorts of related stuff like add calibrated scales to your images, relate pixels to real world measurements, do measurements on your images and such. It takes a bit of messing around with to get used to it, but can do quite a lot for a free SW package.

Another is called "ImageJ" and quoting from their website:

"ImageJ is a public domain Java image processing program inspired by NIH Image for the Macintosh. It runs, either as an online applet or as a downloadable application, on any computer with a Java 1.4 or later virtual machine. Downloadable distributions are available for Windows, Mac OS, Mac OS X and Linux."


It was developed initially by the US National Institute of Health and is in the public domain for anyone to download and use. It has extremely powerful image processing capabilities but the caveat is that the learning curve is quite long. It is mostly useful for captured images so not quite as good for real time "comparator" type stuff but I figured some here might find it interesting. You can read about its capabilities and download it here:

https://imagej.nih.gov/ij/docs/intro.html

Anyhow, just some FYI and thanks for sharing your work!