Extracting Points From Curve for Line

[[Sh...]]

I often use curves to establish a point cloud in which I recall feature points to create lines. This runs pretty smoothly about 75% of the time. The other 25% of the time I get the annoying purple highlight on the line and it can't recall the points. 

The solution so far has just been going back into "recall feature points" and redraw my box. 

Is there a way to stop this from happening?

[[Ma...]]

I am quite lost in translation and your problem, but we can solve that selection box.

I saw this behaviour whenever your measured points are outside of that box:

• your alignment is not stable or
• scanned area is further due machining or part deformation

Because i am dealing with forged parts, which are not straight and so on i am using this method:

1. Scan area in one element.
2. Make a bestfit alignment from that scan
3. Now use another element which will recall all scanned data and it's feature alignment will be that bestfit
4. Now you can use element where you want that selection box:
   1. Use recall measured points
   2. Select element from 3rd step
   3. Right mouse button and select "Add selection"
   4. From combobox select "Feature's alignment"
   5. Now fill XYZ of that selection box
   6. Your box will now move by that bestfit accordingly

[[Ri...]]

I've never been a fan of using the box tool for Recall Feature Points as it can easily break. Most of the time you need to manually change the size of it (usually the thickness of the box). 

I would strongly suggest you use Point # instead - it is more robust - also make sure Curve result points is checked - this will use the act to nom corrected points instead of the raw points (it also makes finding the correct point number much easier - just use the nominal point numbers 🙂 )

 

[[Sh...]]

Please sign in to view this quote.

[[Sh...]]

Please sign in to view this quote.

I will give this a try, this is not selected in my curve feature. 

[[St...]]

When you're establishing your box, it defaults to .0039".  As someone already stated, if the scanned points lie outside that box it can't calculate - it has no data.

If you manually increase the size of the selection box by typing in a bigger value (like,say .025") - or selectively increase by typing in x,y,z value different than the one it defaults to, the recall will be less sensitive to part variation (or other problems which would cause the scanned points to lie outside the selection/ recall box).

[[St...]]

Also, based on that pic: a 2d Curve is trying to spline the points together.  Corners or abrupt transitions won't calculate correctly because the software is trying to best fit an angle via a curve.  You will always wind up with a "fillet" of calculated points.  

You can manually filter those transitions out by selectively masking them out.  Depending on when you do it, tho, it will alter the point count and screw up previously selected point number recalls.

[[Pi...]]

I also noticed 1 thing - that might solve your issue. I see that you have 80 nominal points (those at least are visible on picture) but you have only 12 points measured. You should have at least 240 of them (approx 3 times nominal usually gives decent results).

[[Ma...]]

Please sign in to view this username.

 Hi, that's not necessarily needed here.

Anyways that curve has data only for sloped lines. For lines in Z axis there are no green dots representing measured point ( or evaluated, i am not sure ).
I don't know dimensions so i would reccomend increasing search distance - this can lead to unexpected missing scanned data.

Unless you have segments only for those slopes.

With no scanned data you can not have calculated 3 2d lines in Z axis, so no calculation of intersection also.