3d Laserscan geometry

[Machiner32]

I was wondering if anyone has experience working with geometry created from 3D scanners. Our company is looking into purchasing a scanner, so the boss had a few 'test' scans done on a few things to see if we could machine something from it. One item that was scanned is an old copy mill pattern that was used to rough out a furniture leg that would then be finish carved by hand, so the scan is essentially a 100% organic shape. The laser scanner company provided us with a few different file formats of geometry that their software generated from the point cloud. They included a watertight mesh model as well as a STEP and IGES of surfaces/solids to demonstrate their software "Auto-surfacing" feature. 

So it turns out this is not at all what I was expecting with this autosurface stuff. I'm looking at thousands of patched together surfaces with flowlines going every which way and I'm not even sure where to start since I'm also still relatively new to the entire Multiaxis package in Mcam 2022.

Our customer provided us with a glued up chunk of mahogany for stock that they would normally start with to carve this out, and our goal for this test run is to see how close we can get to the original hand carved part, thus saving our customer lots of time in labor hand carving the entire piece like they've always done it. It doesn't have to be perfect, but I'd like to see how perfect I can get it.

We have a 4th axis rotary on our VMC and so far I've been able to get some 3d roughing paths rotated around the part. but now I'm really struggling to get some of the multiaxis toolpaths to work right on these odd surfaces. Flowline and multisurface won't work unless I painstakingly create my own surfaces. No luck with mastercams "surfaces from solids" tool. No luck using the new "overflow UV" tool. I tried to painstakingly re-draw my own surfaces on this, but then realized that kind of defeats the whole purpose of the Scanner Software "autosurface" feature. I've been exited to try out the new mesh modelling and toolpathing options in the new 2022, but I can't really get it to work right. Not entirely sure how to use a mesh model to drive a toolpath....

 I can attach a file if anyone is interested to help out with this or provide any advice, but its a huge file so I might have to trim it down a bit. I have attached screenshots of the part that show the mesh model and also the solid surfaces that were patched together. 

 

[JParis]

I would look towards creating a flat or elliptical surface above the part, using a 5axis Multi-axis and projecting it down onto those surfaces...

That's one way you can drive a clean surface and work with a multitude of different surfaces.

[Machiner32]

1 minute ago, JParis said:

I would look towards creating a flat or elliptical surface above the part, using a 5axis Multi-axis and projecting it down onto those surfaces...

That's one way you can drive a clean surface and work with a multitude of different surfaces.

I tried a 3D high speed project NCI of a 2d path, then convert to 5x with 4-axis output to add some tilt, and that seemed to work pretty good but I can't figure out how to wrap the toolpath it all the way around the part. If I do this with 5axis multisurface will it wrap around? I'm trying to avoid a toolpath mismatch, though its no big deal since its wood and a 5 minutes with sandpaper will take care of that.

[JParis]

1 minute ago, Machiner32 said:

I tried a 3D high speed project NCI of a 2d path, then convert to 5x with 4-axis output to add some tilt, and that seemed to work pretty good but I can't figure out how to wrap the toolpath it all the way around the part. If I do this with 5axis multisurface will it wrap around? I'm trying to avoid a toolpath mismatch, though its no big deal since its wood and a 5 minutes with sandpaper will take care of that.

Use a cylinder surface around the part instead....that'll get you all the way around...your tool axis control will be important to maintain a tilt as you want.

[Machiner32]

1 minute ago, JParis said:

Use a cylinder surface around the part instead....that'll get you all the way around...your tool axis control will be important to maintain a tilt as you want.

I have also tried using cylinder surfaces of various diameter, and this seemed to work well for the "gooseneck" and ball-claw section of the part, but approaching the "wings" section with the engraving which has more rectangular-ish geometry and is throwing it off a bit. 

[JParis]

4 minutes ago, Machiner32 said:

I have also tried using cylinder surfaces of various diameter, and this seemed to work well for the "gooseneck" and ball-claw section of the part, but approaching the "wings" section with the engraving which has more rectangular-ish geometry and is throwing it off a bit. 

So create a couple or a few, if that's what it takes, of ellipse's that loft into each other to change the shape as the projection shape changes 

You may need to experiment with several shapes to finally get what you need. There isn't a one button push solution on things like these

[Machiner32]

8 minutes ago, JParis said:

So create a couple or a few, if that's what it takes, of ellipse's that loft into each other to change the shape as the projection shape changes 

Still new to multiaxis stuff, but just wondering is it possible to 3D project the "multisurface" toolpath that I created on the pattern surfaces? Or do I use the "compensation surfaces" option? 5-axis convert/dropping?

I've found the scanner company provided a pretty lousy surface model in the form of an .IGS file but the STEP file they gave me has worked pretty well for projecting and other 3D paths so far, but I don't seem to have the option to select solid surfaces in Multisurface toolpath.

[JParis]

The compensation surfaces are what you want to project the tool path onto....you use your created surfaces as the drive/pattern surfaces

[Machiner32]

46 minutes ago, JParis said:

The compensation surfaces are what you want to project the tool path onto....you use your created surfaces as the drive/pattern surfaces

Thats what I tried to do. For some reason I thought you couldn't select solid faces for compensation... But there must be something else I'm not understanding because the path just stays on the cylinder.

[Machiner32]

Update: spent 2hrs today messing around with multisurface toolpath while my machine was ran a long Pencil finish op on the detail which turned out beautifully. But could not get multisurface path to compensate to anything when tool axis is set to a pattern surf. I tried both solid and surface for comp geometry, and different shapes for pattern, but the path stays on the pattern. Part is along the X on the rotary, 4 axis output pattern to cylinder , seems simple enough so maybe I'm just missing something obvious. I was able to get most of the other multiaxis toolpaths to work pretty smoothly so I kept the machine running with those while sitting there stumped. 

[#Rekd™]

If you upload a valid Mastercam file with your toolpaths on it someone will help you. 

[Colin Gilchrist]

For 5-Axis Toolpath Creation, you have to think differently than 3-Axis Toolpaths.

For 5-Axis Paths, you have the following:

1. Cut Pattern > what is the 'shape' of the path motion? (How do you want the tool to move through space, cutting your material). This is often "not your cutting geometry".
2. Compensation Surfaces (Collision Control) > what surfaces do you want the tool to contact' as the tool traces the Cut Pattern. (What shape do you want to generate with the tool?)
3. Tool Axis Control > how do you want the tool "angled" in relation to the Cut Pattern and Comp Surfaces?

With a 3-Axis Toolpath > the 'Cut Pattern' is implied with the choice of the toolpath. This isn't necessarily true with 5-Axis. With 5X, the "toolpath shape", and the "compensation surfaces" are often completely different.

 

[Machiner32]

Thanks Colin, I watched that video over the weekend it was very informative. I think I had the right idea but I was going about it the wrong way, so i'll keep trying on this. My other problem might be this part is a bit complex for a multi-axis beginner (me) to start learning on. The boss gave me a lot to chew on here... I'll try getting some other toolpaths working on simpler geometry.

I started learning mastercam on version 6.11 for fun at my dads shop when I was a kiddo, It's crazy to look back and see how far its come.

 

On 7/9/2021 at 7:31 AM, #Rekd™ said:

If you upload a valid Mastercam file with your toolpaths on it someone will help you. 

I made a zip2go, It might still be too big but ill try to upload it. 

tried deleting some stuff since the file grew into a disorganized mess in my frustrations, i will upload to dropbox and share a link

Edited July 12, 2021 by Machiner32

[Machiner32]

https://www.dropbox.com/s/x96i3aqv99vvdyi/test scan2.ZIP?dl=0

There are different geometries the scan company provided, i put them on different levels. Just wondering which works best? the STL looks to have more defined edges than the solids do. I tried slicing some of it up since there are flat areas on the stock wood glue up that dont need to be milled, and also trying to avoid hitting the rotary and tailstock.

[Aaron Eberhard]

I just now had a chance to play with this, which is WAY more fun than my real job

To deal with this, as mentioned above, create a "clean core" as the kids nowadays are calling it..   The big issue is that you don't want to establish your tool axis control based on a mesh (you can do it directly with the Triangular Mesh toolpath, but it's a PITA).  You want that clean body to drive the toolpath from, then comp it out to the STL.  You also are not going to want to use MultiSurf.. That's a pretty old toolpath and it's not very flexible.  I'd use Unified so you can try many different patterns to get what you like.

I simplified your part to keep the file size & regen times down, and I'm using level #5 to drive this toolpath.

I prefer to work in solids as there's some trial and error and the associative updating of the body is really nice.  I created a spline through your shape using a Wireframe > Manual Spline in the top plane.  Then I used Wireframe > Edit Spline to manipulate the shape:

Then I created a circle in the side plane and Solid Sweep to sweep that circle down the spline:

This is where the Solids associativity is nice, that circle was too big, so I used Analyze Entity to make it smaller:

Next, from the Top, I created the basic outline of the "main" body I wanted:

Then Solids > Extruded it as a "boss" so it was part of the "neck" that I already made:

Finally, I did some Solid > Fillets to make everything nice and smooth:

 

Next up was a toolpath.  Because it's almost lunch time, i just went with a "Plane" that was set 90° to the Z (Previously I would have used the Multiaxis Parallel strategy, but since I'm in Unified, the parlance has changed a bit).   A Morph or Guide would also work fine here.  Note that both toolpaths attached are set a VERY COARSE step over and tolerance.  Please tighten this up before running the program... I didn't want to miss lunch due to regen time, that's on you!   It also kept the file smaller...

One big note is since my drive geometry has a lot of long, sweeping flats, you'll want to set up your advanced options to force a vector density that you like:

Tool Axis Control is set to 4 axis, follow the surface, and no collision control.  This is what it looks like:

 

Now, on the Collision Control page, all you have to do is tell it to Retract and point it at the STL on lvl 5:

To make your life easier, I'd recommend turning on "drop tool" in case your core surface isn't fully inside of the mesh:

 

That'll give you this:

Hope this helps!

 

Get the file here:  DropBox Link

[Machiner32]

Wow Aaron, thanks!! I can see now exactly all the things I was doing wrong. I was confused about machining/avoidance selection in collision control for starters, but now it makes sense what Colin was saying before about the ''motion of toolpath vs what you want to leave behind''. This is a much cleaner path and a lot less work than slicing up my solid model and/or creating dozens of planes, in fact I might not even need the scanner softwares 'auto surface' as much as I had thought. Looks like it comped out to the STL very nicely and with the new Edit Mesh functions in Mcam it should make my job a lot less frustrating! 

I've spent hundreds of hours in the last month learning all this new stuff on my own time, and your example just tied everything together. We plan on doing many similar parts like this one so I really can't thank you enough, I should be buying you lunch! 

[Aaron Eberhard]

2 hours ago, Machiner32 said:

Wow Aaron, thanks!! I can see now exactly all the things I was doing wrong. I was confused about machining/avoidance selection in collision control for starters, but now it makes sense what Colin was saying before about the ''motion of toolpath vs what you want to leave behind''. This is a much cleaner path and a lot less work than slicing up my solid model and/or creating dozens of planes, in fact I might not even need the scanner softwares 'auto surface' as much as I had thought. Looks like it comped out to the STL very nicely and with the new Edit Mesh functions in Mcam it should make my job a lot less frustrating! 

I've spent hundreds of hours in the last month learning all this new stuff on my own time, and your example just tied everything together. We plan on doing many similar parts like this one so I really can't thank you enough, I should be buying you lunch! 

Glad to hear it   Yeah, you'll probably be able to control the surface finish better with tolerance and step over than over-smoothing the mesh.

Like I said, you may find that Morph or Guide may be a good way to go as well,  and you can easily change to that in the Unified interface.  If you're not familiar with it, check out this thread to see my "how to use Unified" videos: Mastercam 2022 - Multiaxis Improvements

If you're ever in the area of HQ, let me know, I'm always happy to meet for lunch with people and you can get a tour of the headquarters.