Page 1 of 1

Center Line

Posted: 16 Jul 2019, 14:52
by macxike
Hello All,

Is there a way to accurately locate the center line of a 3D scanned object (such as a car hood or bumper), so when you mirror the data across the axis in a 3D modeling program it is symmetrical?

Thank you in advance

Re: Center Line

Posted: 16 Jul 2019, 18:01
by Micr0
macxike wrote: 16 Jul 2019, 14:52 Hello All,

Is there a way to accurately locate the center line of a 3D scanned object (such as a car hood or bumper), so when you mirror the data across the axis in a 3D modeling program it is symmetrical?

Thank you in advance
Center line relative to what?
I'm not trying to be clever but you have to think about it from the computers perspective. To us it's easy to see features like say a hole and think there is a hole, and we can easily give a good approximation of the center even if it's not round. But in a scan, no matter how accurate a scan, there is always error point to point. A scanned cylindrical hole is never perfectly round, so how does the computer find the center? Programs like Geomagic will do a best fit by taking an average which can come close (with some margin of error), but they are expensive and still require a bunch of manual CAD work. Especially when surfaces are involved. I've done a bunch of reverse engineering and when it comes to things liked machined features, I always measure them manually (CMM Calipers etc) and build them in CAD, using the scan mesh for the swoopy bits that have lower tolerances.

Re: Center Line

Posted: 17 Jul 2019, 17:11
by macxike
Micr0 wrote: 16 Jul 2019, 18:01
macxike wrote: 16 Jul 2019, 14:52 Hello All,

Is there a way to accurately locate the center line of a 3D scanned object (such as a car hood or bumper), so when you mirror the data across the axis in a 3D modeling program it is symmetrical?

Thank you in advance
Center line relative to what?
I'm not trying to be clever but you have to think about it from the computers perspective. To us it's easy to see features like say a hole and think there is a hole, and we can easily give a good approximation of the center even if it's not round. But in a scan, no matter how accurate a scan, there is always error point to point. A scanned cylindrical hole is never perfectly round, so how does the computer find the center? Programs like Geomagic will do a best fit by taking an average which can come close (with some margin of error), but they are expensive and still require a bunch of manual CAD work. Especially when surfaces are involved. I've done a bunch of reverse engineering and when it comes to things liked machined features, I always measure them manually (CMM Calipers etc) and build them in CAD, using the scan mesh for the swoopy bits that have lower tolerances.
Thank you for your reply! For example, if I scan a full car bumper, how do I find the best estimate for the location of the center line relative to the left and right side of the part. I understand both sides will be slightly different, but is there a way to find a good approximate average instead of eye-balling it? Any advice is appreciate, thank you.

Re: Center Line

Posted: 18 Jul 2019, 15:57
by Micr0
macxike wrote: 17 Jul 2019, 17:11
Micr0 wrote: 16 Jul 2019, 18:01
macxike wrote: 16 Jul 2019, 14:52 Hello All,

Is there a way to accurately locate the center line of a 3D scanned object (such as a car hood or bumper), so when you mirror the data across the axis in a 3D modeling program it is symmetrical?

Thank you in advance
Center line relative to what?
I'm not trying to be clever but you have to think about it from the computers perspective. To us it's easy to see features like say a hole and think there is a hole, and we can easily give a good approximation of the center even if it's not round. But in a scan, no matter how accurate a scan, there is always error point to point. A scanned cylindrical hole is never perfectly round, so how does the computer find the center? Programs like Geomagic will do a best fit by taking an average which can come close (with some margin of error), but they are expensive and still require a bunch of manual CAD work. Especially when surfaces are involved. I've done a bunch of reverse engineering and when it comes to things liked machined features, I always measure them manually (CMM Calipers etc) and build them in CAD, using the scan mesh for the swoopy bits that have lower tolerances.
Thank you for your reply! For example, if I scan a full car bumper, how do I find the best estimate for the location of the center line relative to the left and right side of the part. I understand both sides will be slightly different, but is there a way to find a good approximate average instead of eye-balling it? Any advice is appreciate, thank you.

That was my point. If both sides are slightly different there is no true center line for the computer to extract. Therefore a best fit is all you can hope for. There are ways of doing it by extracting intersection curves, but in the end it's not going to be perfect and eyeballing can come close without spending big $.