I was quite surprised to find out that apparently Rapidform considers Alias one of its main competitors (by the way, I’m not asserting anything, I’m just quoting a gossip which is a wrong thing to do, information should be checked before wrapping opinions about it, but I can’t help myself). It is hard for me to judge why so since I haven’t had the time to try out Alias for reverse engineering applications. Alias is an Autodesk product, highly sophisticated surfacing package that is capable of processing point clouds and meshes and use them for modifications of current design. Beyond that I know nothing of it.
So far I’ve been concerned with comparing Rapidform only with Geomagic for the simple reason that these two packages are available to me in their full glory. Autodesk Alias is available for free trial period download but whether it has its full promised functionality I unfortunately have no clue as of yet.
To be honest, with Rapidform and Geomagic doing what they are made for I see no point for involving a surfacing package (which is not a properly dedicated point processing, thus reverse engineering package) in the process. If Alias is extensively used within the organisation it makes sense not to invest in different package. But when you have a choice between Alias or Geomagic, or Rapidform and you don’t require A-class surfaces, what will you get?
I’ll drop Alias out of conversation for now and will focus on Geomagic vs. Rapidform battle. Actually, both packages are not really the same thing and probably aren’t true competitors but the task I need to handle can be tackled in either of them (with some obvious and not so obvious disadvantages).
I was given air inlet scans (not of very good quality) scanned by a handheld laser scanner to reverse engineer and preferably come up with an editable CAD model. When you have both Geomagic and Rapidform at your disposal there are several ways you can do that. I don’t have all the time in the world, so I didn’t try them all. Besides, some of them seemed like not a very good idea. To explain simply - the air inlet component consisted of a rectangular base with mounting holes and a hollow section that had rectangular shape at the bottom which transforms into something like oval at the top. This might sound very confusing, the picture below is the closest thing to it I can find as I’m not allowed to show the actual part. Our part had a bit more regularity to i’s hollow section and, to be honest, seemed like a very simple thing to remake in Rapidform XOR. The part was given back to the customer before I started work on the scans so all I had to work from really was a stack of points that were obviously a bit faulty. I wasn’t really upset about that, because, as I mentioned, it seemed pretty easy.
(this picture gives a pretty good idea of what is an air intake though our part had different geometry, the pic comes from http://www.600rr.net/vb/showthread.php?p=1370856)
So the first method is to do all of it from start till finish in Rapidform. Obviously, I didn’t do it because I have no luck in creating a mesh from scan data in Rapidform. I still haven’t figured out why and I’m still writing it down to lack of experience. Besides, I already mentioned that I prefer Geomagic mesh editing tools to Rapidform’s (despite Rapidform positioning itself as more user friendly and familiar to CAD users, I’m very surprised how “unfriendly” its mesh editing tools are; for example, mesh borders which are highlighted in Geomagic – a blue mesh has green highlighted borders which turn read when user chooses – but not in Rapidform which makes it very hard to find all holes that need to be filled in massive uni-colour mesh where “Fill all” option is not applicable; if it all comes down to some silly display settings, I’m sorry to be so picky about little things though I think mesh borders should be highlighted by default). Thus, all methods were mesh manipulations could be done in Rapidform were avoided. With my faulty scans I went straight into Geomagic Studio and tried to save what could be saved. When I call my scans faulty I mean they are not very accurate and even the most awesome mesh software in the whole universe won’t help to fix it. What would fix it however is to introduce the actual part to my callipers (I’m very proud of my callipers, it’s a great present from ex-colleagues) but the part went back to the customer. A lot of guessing got involved which made me really unhappy with the resulting part but more on the end model later.
Combining scans, cleaning it, meshing and editing is not worth discussing anymore – with Geomagic it’s like a second nature, click the icons without using much brain power. The route is always the same – first multiple scans are aligned using Local Alignment, then Global registration. From here there are two ways – either merge the scans and produce a mesh in one step or combine the scans without meshing. I prefer the second way because that gives me the chance to do some cleaning on the scan (like deleting outliners, cleaning noise and most importantly – fixing normals). I’m very pro-normal-fixing because without it the resulting mesh can have some serious defects which can only be fixed clean by deleting parts of the mesh and filling them like a hole (during which obviously original information is lost). What can happen is layers of mesh form on top of each other so the mesh model has sort of “flakes” over the surface. They can be eliminated by “filling holes” or “rewrapping” but often it results inside surface flipping on the outside and it’s something very hard to get rid of. Therefore, my advice – fix the normals.
(a layer of stray mesh can be seen inside the part. This defect resulted from scan error not from normal problem, but similar smaller layers appear if I don't do anything about normals)
I made a number of meshes – first was a reference mesh with all the features intact, but smooth, with clean edges and hole-free (apart from bottom and top opening, of course), second mesh with only free-form-ish part of component with bottom base and all mounting holes deleted; third and fourth meshes were the same as 1st and nd but thickened – the original part is a follow carbon fibre air intake which was scanned only from the outside. It looked to me as if the thickness was pretty much constant through the whole thing but the part was taken away before I could actually measure the thickness (which is only my own fault). I didn’t keep the carbon fibre texture as it wasn’t necessary.
While I was still in Geomagic I decided to make a surface model of air intake in case all my efforts in Rapidform fail. I used thickened mesh with all features intact and “Exact surfacing” option. Honestly, I really think I should get a grip of “Parametric surfacing” but what looks so easy in tutorials with parts specifically selected to demonstrate benefits of the option isn’t as easily done with my little training on randomly shaped parts. Anyway, surfacing gave the result I expected – regular features got robbed of their regularity and would be no good.
(Can you see the mounting holes on the bottom of the base? They got seriously distorted during surfacing. Tinkering with the mesh might get a better result but it adds to processing time).
What used to be done is to process data capture info in Geomagic Studio and send surfaced components as IGES and STEP to customers who then (if they needed a CAD model) would use it as a template to rebuild the component in whatever CAD package they use (sending .stl is not always an option, as not all CAD packages can open them and even the ones that can aren’t capable of performing any actions on them apart from showing a point cloud in random space; there are add-ons that can help but they have very limited capability).
I also surface the fourth mesh – thickened but without features. I had a funch that the freeform part of component might not be so easy to rebuild in Rapidform manually and it will need autosurfacing and as so far Geomagic proved to be better at it. The idea was to import surfaced featureless body into Rapidform, align it with reference mesh and rebuild the rest of the features around it.
After covering all the possibilities in Geomagic Studio, I finally abandoned it with a pack of meshes and couple surface models and went on to explore what Rapidform is capable of.