Petr, you obviously did not understand my last post, or perhaps you confused my post with a different post. A scanner is not useless.
The tools in Rhino work for me to section and contour meshes from scan, create curves where needed, and create surfaces using the setions/contours/curves. The match of the surfaces to the scan data is then refined.
Questions for you:
Have you modeled a new shape similar to the ones which you want to “reverse engineer”? Knowing how to model an object is a key element of creating a model from a scan.
You claim that Rhino has tools for selecting contours from the mesh and further simple operations ,like extrude, revolve, sweep, boolean and etc, but the Rhino toola don’t work in this case.
What is the basis of this claim?
Have you tried to “reverse engineer” using the Rhino tools to section and contour meshes, and then create surfaces from those curves? What did not work for you? Did you reach out to this forum for help in making those tools work?
@Petr_Tikhonov
I’ll leave this as an example for a curve from two views. Try to understand it by deleting the 3D curves, because to recreate them, you just need to take the side and top curves and use the “curve from two views” command.
also erase the surfaces and try to redo them from scratch, it will be preparatory for you to learn
I leave you two projects within this forum that I did in my free time just modeling nurbs.
the gun is a finished project, while the car I’m far from finishing because I currently don’t have much free time to dedicate to modeling beyond work.
this is the only case in which the modeling is mesh, done in zbrush, with a small contribution from rhinoceros
Thank you VinPo for helping me. I just meant that if we anyway has to redo the form almost from the beginning the scan it is not necessary. Three photos made on smartphone and measurements will save money and time. Because the process of scanning is not easy and take time. Thank you guys, for your help and advising ,it is very important for me. I am new in this field, and most of my questions can arise for you stupid. In the past, all my questions I’ve directed to Rhino team, they let me know that the better way is to ask for help on the forum. I was very embarrassed to look like an idiot, especially since English is not my native language. Thank you for understanding and help!
In principle, your task is to use your 3d scan as a reference, and use all the available rhino tools to essentially re-create and idealized version of the object mostly manually. Tools like Geomagic automates some tasks - making it perhaps faster - but there’s nothing you cannot in theory also do manually in rhino - and automate some tasks. To have success in your task is to really learn how to use rhino. This is a time investment - but the good news is that there are plenty tutorials online.
Project from 2015.
22 foot long wooden boat. Scanned using photogrammetry (Metashape software). The boat was sitting on a trailer against the back wall of a shed, surrounded by other boats.
The photogrammetry scanning resulted in a point cloud with 25.5 million points. A mesh was created from the point cloud and decimated to 2 million faces. The mesh was imported into Rhino.
In some situation, tracing profiles from images doesn’t provide you full info to model, for example, some area which is hidden from the view, cross section shapes, etc.
So 3d scan data still works as a “direction” to provide you the data that you won’t get from the images.
Combining use of both 3d scan data with images is the best
Thank you Brv very much for your detailed answers! You own projects are very impressive. I am still digesting all the examples you gave me. I already tried something, but I can’t boast of success. As I understand, you outline the scan and on base of curves you construct the surfaces. I have divided the mesh on contours, then tried to loft them and my comp said bye bye. On that example that you advised me to do cross section the shape on the second view more rectangular. That is my fault, I didn’t gave you right picture. I mean that the tip of that thing are rectangular .That means that I need to add different shape, or make more picture from the front. Thank you very ,very much for giving me a lot of info, that is not easy to find!
You can get rid of the noise and might loose some of the details. It’s your task to put the control points in the right place. Same thing might help when working with subd. Either you work with high face count / control point count and end up having a hard time editing the result or you rebuild the model with lower point count with the risk of loosing detail…. The solution is likely somewhere in between
FitCrv may be useful when rebuilding sections/contours of meshes. I usually need to experiment with the tolerance.
Another approach is to use Rebuild with a large (perhaps 20) points, and then use RemoveKnot to simplify the curve by removing control points where they are not needed.
PointDeviation is an essential tool. Create/simplify a curve, then check the deviation from the input. If the deviation is too large UnDo and try again.
@Petr_Tikhonov Hi, I recorded the video of how I position a randomly scanned object, but the file is too large to upload. If you send me your email in private, I’ll send you the video. I can only upload the video of the 2-view curve command, as it was short. In the positioning video, I used the “Orient object: 3 point” command, in case you’re wondering while watching it, and it’s just a simple rotation.
if you reconstruct the contours, you will be much more precise, and if the customer requests it, you can make all the necessary changes.
It often happens that the client wants to make a change of one-tenth on one part, three-tenths on another part, and so on, in small portions of the curve, while leaving the rest of the curve unchanged in terms of its mathematics.
