Hi Michael,

Yes, it is the original produced like.

>>>Most likely tiny segments like that would be the result of the original curves coming really close to being tangent with one another, but not quite being really tangent.

I will then have to look at the fillet on 2 curves.

Hi Steve, there are a couple of problems with curve in these areas:



There are little fragments around 0.002 units in size at those areas.

Also the one on the right actually has 2 little fragments there, and that is the start/end of the curve. The little fragments are overlapping at that right-side spot, making this actually be an open curve that has a small zone of self-intersection/overlap instead of being a closed curve.

Here is a zoom in and some separation between the pieces on the right to illustrate some more:



If you have an earlier version of this, could you also post that so I could see if filleting introduced these little tiny segments?

- Michael

Hi Michael,

Yes, sorry,
I did post the result of such a offset. (sorry, long day).

I am attaching better correct explanation?

First to show original joined curve (1) and to show offset curve (2). Model at that stage attached.

Hi Steve, it does appear to be related to very small differences in the tangents (like around 0.035 degrees), which get magnified to a certain extent during offset.

Do you happen to have any earlier versions of this before the curve filleting went in place?

Also, are you able to finish this model now, or do you need help tuning up this particular case?

I think that the fix will be to true up tangents that are really close before doing the offset. I'm also planning on similar tune-ups just in Join in general, and if I have a test case where the curve fillet is not accurate enough that would help to improve that as well.

- Michael

Hi Michael,

>>>it does appear to be related to very small differences in the tangents (like around 0.035 degrees), which get magnified to a certain extent during offset

Is that then just a connection problem on the offset curve (just interested)?

>>>Do you happen to have any earlier versions of this before the curve filleting went in place?

No, as you will see, I was very early in construction at that point.

>>>Also, are you able to finish this model now, or do you need help tuning up this particular case?

Certainly no problem from me, I did find a little problem, but whatever, I will still finish. I was only trying to show a problem I found.

I will certainly look more at this, more as a possible problem for others (it will probably show again~ possibly from a user who does not know what to look for) and not everyone will look back as I do.

Michael,
Please, I am not complaining, I know such as deviant as "like around 0.035 degrees" is not a problem, but knock on effects such as ability to sweep/merge can/will fail (it is why I backtracked~ then posted).

Regards,
Steve

Hi Steve,

Don't get me wrong, this is a good concrete example for me to focus on to tune up, thanks for posting it.


> I know such as deviant as "like around 0.035 degrees" is not a problem,

Well for a lot of things it would not be... But the offset mechanism is kind of sensitive to this particular amount of deviation.

Things right around this level of deviation create offset segments that are just a little bit too far apart from the 0.001 tolerance, so they don't get the sort of automatic snap-ends-together applied to them, they will get filled in with either a small connector if there is a gap, or if they cross each other it will try to intersect and trim them. But it is quite difficult to get a proper intersection on things that are very close to colinear...

I think the best bet to fix this up is for me to actually try and adjust curve's shape just very slightly to true up the tangent there. I've just got to be a bit careful about how this is done, I don't want to mutate things too much.

For now I will put this on my bug list of things to tune up.

Please let me know if you continue to run into it in other situations.

- Michael

> Is that then just a connection problem on the offset curve (just interested)?

Yup, basically each segment is offset - if the segments were exactly tangent with each other, then the offset segments will touch each other exactly and can just be easily joined.

But if the tangents are different between the 2 segments, then the offsets will have either a gap between them, or cross each other.

Gaps are filled in by putting in a kind of extension piece - this is what is controlled by the "Corners" option.

The part that tends to be more difficult is when the pieces cross - then they have to be intersected and the excess part trimmed away. Intersections of things that are small and/or nearly colinear tend to be difficult to calculate.

- Michael

Hi Michael,

I will keep a check to see if this happens again. I have made a number of similar constructions but have not seen this error again.

I thought I did see somewhere a script for incremental save but cannot find this, does anyone know where that is? (or was I mistaken?)

Regards,
Steve

Hi Steve,

> I thought I did see somewhere a script for incremental save
> but cannot find this, does anyone know where that is?

For this one use IncrementalSave as the command name in the right-side column of the shortcut key.

Also if you would like to have a button on the bottom toolbar for this, there is a UI update here that will do that:
http://moi3d.com/forum/index.php?webtag=MOI&msg=1139.21

- Michael

Not being a fan of key strokes, the Button is a joy to use, regularly and often!

Hi Michael,

>>>For this one use IncrementalSave as the command name in the right-side column of the shortcut key.

So thats why I could not find a script, lol

Many thanks,

Steve