I personally dont know which is better.
I prefer Rapid form.
Geomagic requires you to make planes and or surfaces to dimension to where rapidform you can just dimension the model.
We had that same decision here a while ago. It seemed like Geomagic was a little more user friendly, especially if you didn't use it every day, but Rapidform appeared to be more powerful, so we got Rapidform. I don't personally use it, but I haven't heard any complaints.
I have over 16yrs experience of 3d-modeling and high-level 3d-scanning in all industries and applications. Usually from a technical point of view.
We now have a consultancy offering services in 3d-scanning and specialize in putting together whatever the best hardware, software and project management is suited for any 3d-scanning application.
We sub-contract much work to the appropriate contractor and try to remain independent. This is important with the hardware as most systems still have characteristics that will be appropriate or not, depending on the job. However, with the software we usually point towards Geomagic, or more accurately Geomagic+CAD. Let me explain…
There is a plug-in that comes with Geomagic Studio software, which allows it to interface directly with Solidworks. You can then use Geomagic to (quickly and semi-automatically) create curves, 2d-features, 3d-features, sections, freeforms, extrusions, drafted extrusions, revolves, sweeps and lofts on scan data...then send them (individually or in a specified order) to Solidworks where they will be rebuilt DIRECTLY IN SOLDWORKS, with native history. The appropriate operations for each surface/solid (subtract, trim etc.) will also be applied as they come in.
Rapidform and Geomagic have fundamental differences in their approach to the MCAD market. Essentially Rapidform attempts to be a standalone modeler, referencing scan-data. Geomagic concentrates more on using the -usually millions of points- scan data and interfacing with existing CAD (i.e. Solidworks) to provide the CAD system with all the information needed to build natively. Hence we can deliver native Solidworks files where the model has been built IN Solidworks, not a Solidworks file that has essentially been built in something else. Nice!
I presume you're talking about reverse engineering rather than inspection.
Fundamentally, reverse engineering should involve a step where you can verify the accuracy of the reverse-engineered model. This is especially important if you are making a feature based CAD model because you are going to be making a lot of approximations when moving from scan data or point cloud into SolidWorks feature tree and you want to have a frame of reference as to how close the final feature-based model is to the original. Most of the discussion is about accuracy, and scan data useage and can help guide decisions.
When Accuracy in the Scan is important (Scanning Step)
In Reverse Engineering you start with a 3D scan and you always pay a premium for high-accuracy scan. If you you are reverse engineering something where accuracy isn't important then either a low or high accuracy scanner will both do the job (for example any of these middle bunnies scaled down to 1/24th scale will produce the same shaped gummy-candy from a mold).
When scanning for downstream use in functional parts and assemblies - be it a surface (like a helmet or a fender) or detailed mechanical shape (like a plastic housing or a connector, for example) - then accuracy starts to become more important because you need to worry about function and fit and more accuracy in the scanner is important.
When Accuracy in the CAD is important (Reverse Engineering Step)
Accuracy is also important in the case of reverse engineering from Scan to CAD (where scan data processing software comes into the equation). If you are extracting lofts or through-holes or estimating draft on extruded faces, then you are going to want to have both the CAD & the Scan in the same software so that you can validate the accuracy of the features you are creating (for example, does a nice round number like 10-deg draft on a wall or a 10 mm fillet on a corner deviate too much from the scan data). You should have an idea what should be hole feature, rather than an oval feature, and an extrusion feature, rather than a loft feature. So you can paramaterize your model into these feature approximations to capture the 'design intent'.
You want to make sure that your approximations have integrity to the original shape of the part. That is the approximations in converting the scan data into more easily used and edited CAD features (reverse engineering) are sufficiently close (are accurate, have integrity) in relation to the scan data (and thus the original part). A color-map is the most typical way to determine how accurate your (fillet or draft) approximations are and this is basically a difference-map between the scan and reverse engineered CAD.
If you are entirely planning to use the scan data for its as-is shape (no approximations) -- for example to develop a holding fixture for a series of identical turbine blades -- then high-accuracy "NURBS" surfacing are common and essentially stick, like saran wrap, to the scan data. There's generally no approximataion in this type of application, and typically the surfacing goal is to be as close as possible to the scan data with the surface patches.
In regards to the "saran wrapping" of a point cloud. Is there any easy automatic ways of doing this with any software. We typically want to build our solidworks parts to fit tightly to a surface. Many times we build our parts and the last feature is a "cut with surface" that uses the scanned surface we paid for. We would like to bring the scanned surface generation in house though.
Surfacing / NURBS surfacing / Autosurfacing / Shrink-Wrap surfacing are similar terms you'll see in 3D scan-based models that stick to the point cloud.
In reality, there are intermediate steps in the shrink-wrap surfacing process, including the question of whether or not shrink-wrap surfacing is the right choice for the application of the geometry. I explained the difference between as-is modeling vs design intent modeling. Below is an example of design-intent modeling to illustrate the difference and determine if shrink-wrap surfacing is really the way to go:
You can see that the scan data is not actually a point cloud, it is a mesh. Just about all 3D scan based reverse engineering processes involve first converting the point cloud into a polygonal mesh. Higher-polygon count models and more complex geometry is where you're going to start needing dedicated 3d scanning software.
Another good example of different model types is from 3D Scanning Labs, who have a few examples posted in this page including feature-based, hybrid, shrink-wrap and polygon models.
Reverse Engineering Variations
Many of the shrink-wrap workflows are fast but require a bit of training and of course the right software to handle the point clouds. Starting with a high-quality scan is important for shrink-wrapped modeling. For example, if you are in the business of custom hearing aids, you'd want a scanner that could at least capture all important details so that you could shrink-wrap point cloud. With lesser-quality point clouds, you'll be sanding and smoothing away the data and filling in holes, which are essentially approximations. If you get a good scan, then the scan represents the part well, and you create a tight surface, and then machine/fabricate directly from that surface. If you can fabricate directly from the NURBS surface, then shrink-wrapped surfacing is a good fit.
The design intent path is more interesting. If you want to develop design-intent models, you had better be experienced in SolidWorks in order to extract the desired design-intent from the scan data. This workflow requires training as well, and the right software. In design-intent modeling the reverse engineering softwares like Rapidform and Geomagic begin to differentiate themselves. Data quality is typically not as important in this case as there are tools (moreso in Rapidform) that allow modeling with rough or incomplete scan data to make high quality models. As mentioned in the post above, for design intent models, you'll want to verify the accuracy of the models, and a 2nd typical requirement is a well behaved SolidWorks model (parametric SLDPRT). Jumping between SolidWorks and a scanning software to update parameters to ensure accuracy has it's downsides in terms of intuitivness, complexity and speed. Rapidform software currently models parametrically and verifies accuracy of the modeled geometry and then transfers the parametric tree into SolidWorks -- meeting typical industrial/mechanical reverse engineering requirements in terms of deliverables.
Scan-to-Shrink Wrap-to-Design Intent (not recommended for complex forms)
Another approach for obtaining a design-intent model is to first use NURBS surfacing, then import the surface geometry into SolidWorks. This is a pretty tough and non-intuitive workflow based on my experience, and by the end of the process you're going to have done a lot of estimating and eyeballing in generating the model from a NURBS import. To appreciate the challenge of working with NURBS, just imagine how you'd align an imported NURBS body of even a simple shape (like an iphone) into to a coordinate system. Scan-to-Nurbs-to-Features is a tough workflow and it's better to jump directly from Scan to Features because of speed and final model accuracy.
Convert stl after ZBrush to SW (Use rapidform)
Message was edited by: Shon Owl
Retrieving data ...