[kstrauss]

I need to generate "point clouds" to produce replacements for some castings which are perhaps 6x12 inches. If you care, these are cast iron brake shoes and other components to be used in the reconstruction of an 1850 vintage railroad ore car. Using the Mach3 wizard produced by Fenerty and included in the Tormach version of Mach3 works but takes forever to scan a large area. There are a number of other choices -- some free and others not so cheap. What is this group's wisdom regarding a reliable, fast and reasonably priced alternative? The work is for a non-profit organization so funds are limited.

[rowbare]

How complex are the parts? Are they really 3D or are they 2.5D? Are you looking to make exact replicas or is a functional replacement sufficient? Can you post some pictures of the parts?

If they are essentially 2.5D, photograph them square on (or put them on a flatbed scanner if they fit) then auto trace the resulting image in a vector drawing program. You would probably want to convert to a high contrast black on white beforehand.

You could also take a hybrid approach where you would autotrace or draw the bulk of the part and only probe the tricky details.

Given the age of the part, it was probably drawn with a ruler and compass so if you are somewhat proficient with CAD software, you could probably redraw the part in less time than it takes to probe it. You could still manually probe to get your reference geometry.

Or using a couple of photographs you could do something like they do in this Fusion 360 tutorial: Create a razor design from concept images: Part 1 | Fusion 360 | Autodesk Knowledge Network

These are just a few ideas to get you started. I am sure someone will chime in with better ideas.

bob

[kstrauss]

Thanks for the comments.

There are several different pieces so I would like to streamline the processing if possible. The parts are actually 3D although functionally they could be 2.5D. The originals were (presumably) cast using wooden patterns so there is draft to allow releasing from the molds, nice convex curves, etc. The replicas need not be identical to the originals but I definitely want to retain the 19th century flavour of graceful castings. The originals are currently at the local museum so I can't immediately provide photographs. They'll probably be in my shop in a few days and I can then photograph them. Meanwhile, there are some online reports of the project: Ore Car Project | Cobourg Museum Foundation and Cobourg’s Crossen railway car comes home

[MichaelHenry]

Have you considered Tormach's CNC Scanner System?

Tormach CNC Scanner

Mike

[kstrauss]

Perhaps I'm missing something but does the Tormach scanner generate 3d scans?

[MichaelHenry]

No, as I understand it the end product of Tormach CNC Scanner is a 2D DXF file. The part is mounted to the table, a USB microscope is mounted to the spindle and the motion control system on the mill is used to create a fairly high res composite image of the part. You then trace the features of the part with separate software to generate a 2D sketch that can be used in your MCAD software to create the part in 3D or in a 2D CAD program for touchup and fine tuning. I've not used it myself as I haven't needed to use something like that yet.

The full-featured package is pretty expensive but they used to offer free software with more limited features and for use with off the shelf USB microscopes.

Mike

[sansbury]

I don't think there's any fast cheap way to do automated point scanning. Personally, I'd use the probe to determine critical dimensions and then do the rest in a solid modeler by eye. There are starting to be some hobbyist level 3D scanners meant for use with 3D printers but not sure they can handle that size. The problem is that all of these generate point clouds which can be difficult to work with, which is why I prefer building a model from measurements.

[pickled]

Yes it takes time to probe and generate 3D models with a touch probe...besides the use of a well planned systematic approach money is about the only short-cut available. It takes time with a 3D laser scanner too. The only way that I have found to reduce time is by prioritizing what areas require high resolution vs. low resolution while manually generating the obvious 2D geometry (Overall peak height, rectangular dimmensions...etc.) manually on a parallel path. I have a NextEngine HD with all the fancy software and I still end up performing a lot of common sense obvious type geometry modeling just as a reality check. Also realize that if you are working with worn, broken and/or used parts of unknown origin you do not know for certain that these are actually representative of the OEM specifications and or anywhere near the target values of those specifications. I wish you luck as you move forward! With planning and patience you will succeed.