Hi Michael,
how are you doing?
I hope things are proceeding well for you.

I have an observation.
I use the "sweep" tool a lot and I have noticed that in V2 the "sweep" is much cleaner and no problems vs the shame shape (BTW, these are shapes designed by me with the Freeform tool, almost let's say ovals or heart shapes) in V3 beta (which I use daily).

Basically,
the V2 sweep result is cleaner than the V3 result.

Has somebody else found this?

Can you plse tweak it, expecially if I will go for the V3 final release when due.

Best & Thank you,
Rudy

Hi Rudy, I'm not really aware of any changes specifically to sweep between v2 and v3...

Do you have an example file that I could test with to see the difference?

Thanks,
- Michael

Hi Michel,
thank you for taking the time.
I Include a 3dm file and two screen shots, the sweep clean one in V2 and the odd one in V3.
Rudy

Hi Rudy, thanks for posting the example file. That seems to be something going wrong with corner mitering. I'll see if I can track down why the miters are not behaving well there in v3 but it's a generally difficult area to intersect 2 surfaces that come very close to being tangent to one another but are only a few degrees off, which is what you have here. The segments of your curve there are close to being tangent to each other but are just not quite smooth, about a 1.7 degree difference between them.

The sweep mechanism sees that is not a smooth juncture there, and it extends the surfaces and tries to intersect them with each other, and it's not all that unusual for that kind of barely grazing almost tangent type of intersection to be difficult to calculate.

If your other cases are similar with strange long sections poking out, those are likely to be the same problem - if you get the curve to be smooth in those areas rather than just slightly kinked it should get rid of those kinds of problems.

- Michael

Hi Rudy, here I've attached an updated version that has the pieces tuned up so they're smooth to one another. I did that by turning on control points and collapsing the endpoints and first inside points of each curve to be in a straight line. Now that they are smooth to each other rather than at a very slight sharp corner it should behave properly in both v2 and v3.

I'm not entirely sure why v3 did not like your file as much as v2 did - I do seem to remember some changes in v3 to miter intersections to use an intersection tolerance relative to the size of the individual curve segment rather than the whole big curve. That then fixed other cases that were not intersecting properly but seems to have made this grazing tangent intersection case you have here behave not quite as good. The "near tangent" type intersections are a problematic area in general though, it's usually best to set up your geometry in such a way that they're just avoided.

- Michael

Thank you Michael.
yes, tweaking the shape through control points will work definitely better.
Thanks for your input.
Very much appreciate it.
Enjoying MOI very much.
Also, in my last project I had a similar shape and builted a ring around it...
Basically, at end of project I had an area underneath that did not closed (somehow though, it did boolean union OK....).
As I grow these models as RPs, I have discovered Netfab basic (STL viewr + repair) and that software took care of that little area.
So, basically I am squared.

Thank you Michael.

No problem Rudy. So is this thing with sweep behaving differently from v2 something that you've also seen on a lot of other cases as well or was it just this particular one that it happened on that surprised you?

- Michael

Also I've checked and it is indeed different in v2 versus v3 because of the changes made to intersection tolerances. Those changes were made about a year and a half ago.

The changes were a result of this example model posted here: http://moi3d.com/forum/index.php?webtag=MOI&msg=5266.1 if you try sweeping that file, you'll see that v2 makes the weird result in that case, and v3 does it better.

Unfortunately the same change that made that file behave better made your particular case here behave worse...

- Michael

I did see it in couple of other occasions, after trying older files , builted in V2, related to this last project ( which I do in V3).
In other words, these odd shapes were done all in V2, except this last one.
This is where I tried the odd shapes that I did so far, in both V2 & V3

From what I have seen, V2 did a better, cleaner sweep (without the tweaking of control points), where V3 failed.

Rudy

Hi Rudy,

> I did see it in couple of other occasions, <...>

If you happen to have those files handy can you please send them to me at moi@moi3d.com ? It could be good for me to check if they're the same kind of "almost tangent but not quite" type situation or something different from that.

- Michael

Hi Rudy, I've got this tuned up so your original file there will sweep ok in the next v3 beta.

If you do happen to have any other cases that you might have saved that also had problems, please send those over to me at moi@moi3d.com so I can make sure the tune up fixes those cases as well.

But anyway the tuneup should help with mitering these near tangent type cases.

- Michael

Hi Michael,
Thank you.

actually cannot find any other files with this different type of behavior in V2 and V3 beta. Thought I had one more, but it looks like not....if I find more, will send surely.

I usually make the "odd shape" half way, then mirror it, and delete the apex point (for sweep). It usually works in every case, except in this one.

Anyway Michael, pardon my ignorance, but I did not understand the changes you made at the shapes using control points....cannot find any visual difference (control point wise) from file I have submitted and the file you fixed. Can you explain, if you have more tolerance & time about this, the passage below

"...... Hi Rudy, here I've attached an updated version that has the pieces tuned up so they're smooth to one another. I did that by turning on control points and collapsing the endpoints and first inside points of each curve to be in a straight line. Now that they are smooth to each other rather than at a very slight sharp corner it should behave properly in both v2 and v3. "

Thank you.
Rudy

Hi Rudy,

> Anyway Michael, pardon my ignorance, but I did not understand the changes you made at the
> shapes using control points....cannot find any visual difference (control point wise) from file I
> have submitted and the file you fixed.

The difference is quite small so it's hard to see visually - it's changing some points that were at at about a one degree angle difference to each other to instead be in a perfectly straight line.

What I did first was to use Edit > Separate on the curve to break it into it's individual segments that make it up. That way you can hover your mouse over each one to see where each segment ends.



Then I turned on control points and zoomed in quite close to the end, there is a small cluster of control points at the end so you have to zoom in quite a ways to get just the last 2 points at the end of the curve:




Then I selected these points here - that's the end point and first inside point of each curve:




Then to true these up so the curves are smooth to one another, those should all be on one single straight line, to do that I grabbed the corner of the edit frame and moved it over until I got "flat" snap:




Hope that helps explain the tuneup!

Probably at some point in the future MoI will do something like this automatically to fix up curves that are close to being tangent to one another but just a couple of degrees off. I haven't done that quite yet because it's a little bit hard to judge how much of a change should be permitted.

- Michael

Thank you Michael,
really appreciate it.
Rudy