The new V3 beta will perform blend on unequal edges, unlike the V2 blend which needs equal edges to blend (I dont even think the V2 does multiple edges. You have to do one at a time)

I dont think pick order or area will do anything different. You can run "select naked edges" on the file and hit blend and get normal results.

There are a couple fragmented edges (could use merge to fix that. Select both objects and hit tab and type merge and hit enter). All this would do is have you miss one of those little edges in a pick, which would produce a faulty blend (all twisted in an area). If you get a good blend, there is also a start and end point on both sides that allows the adjusting of the blends meeting point.

Hi wim,

> Why does MoI not want to blend this 2 shapes?

You're probably using an older version than the most current v3 beta (version Aug-1-2013) ?

In versions previous to that, blend could only work between one single edge on both side, and in your case here you have many edges selected, not just 2 selected.

But this has been tuned up a lot in the Aug-1 beta so that blend can now work with a chain of multiple edges on each side of the blend instead of only between 2 edges like it used to.

- Michael

Hi mjs,

> Michael G - if you appear, is edge selection order a normally picky? I typically try to use reasonable starting
> positions when I am blending, lofting, sweeping, etc.

Some commands do pay attention to the pick side to decide on a direction, like _curve_ blend will use the end closest to where you clicked as the place to attach the curve blend, one rail sweep uses the end closest to the pick point for the start of the sweep. But for profile curves that are being combined together to make a surface, that does not depend on the click location, things like loft and surface blend automatically decide which direction to orient things by using a twist minimization calculation. If the automatic determined direction is wrong then in the final stage of the command you can click on a profile to reverse it there.


> but in this case it when it worked the second time for
> me I went in the opposite direction as before and started both selections in a different area than the first
> time. Fluke? Luck? Or does it sound like I ran into a rare issue?

Over here I can't seem to reproduce a failure using any type of different pick sequence. But if you accidentally did not select some piece, that could explain getting a bad result...

- Michael

Hi mjs,

> <update> Found a way to cause a failure in one way. Depending on how close to the ends
> of each edge you are when you select. This is something I don't normally run into as I always
> try to pick the near the points that I want to have lofted together. However, in the case of this
> file the two open holes have different numbers of edges. The smaller one has 4 edges while the
> larger one has 8 edges. If you pick closer to the "wrong" end of the little short edges you can
> get strange twisting or even a failure now and then.

There isn't anything in edge to edge blend that's trying to do different things depending on which side you selected on.

Is it possible that on one of those "wrong side" selections that you accidentally selected one of the other edges like this one here? :




Over here I've been trying all sorts of different selection orders and click sides and I haven't been able to get a single bad result...

- Michael

That sliver is an easy one to pick and not see. I bet that was it. That is a pretty small sliver and I don't think I was zoomed in enough to notice that as well as the actual edge to use.

Thank you all for your elaborate input!
I was not aware that same amount of edges should be used, so far it always worked, pure luck I guess.
I am not using the beta yet, I must rely on solid software for my job, perhaps this beta is solid too, but I used to work in a SW development company who sent out beta versions and therefore I am a bit careful with that.
I will d/l the beta and give it a try. but I am still very happy with 2.5 though, it does the job!

wkr
wim

Hi wim, I really try to keep the v3 beta stable and able to be used for regular work. It's got all the latest stuff in it, like in your particular case here if you would have been using it your original case would have worked as you expected it to.

- Michael

>I was not aware that same amount of edges should be used, so far it always worked, pure luck I guess.

No, you certain can Blend (between two un-touching selections) with mismatching numbers of adjacent edges.

Hi Michael, I was (now with this release) intermittently having an issue blending between two single edges that happened to touch on one or both ends.
This is how I would do the "scalloped" blend fix on the ends of my "transitioned recess surface" procedure.
Sometimes it still works, but many times it doesn't now, where I am still able to make it work in V2.
Do you have plans to see if a new bug was introduced?

Could it be possible to ensure that the Blend is still performed - in the case, if only two edges are selected, even if they are touching?
This way, the same original functionality of Blend would coincide with the new and improved version.

Hi Mike,

> Do you have plans to see if a new bug was introduced?

Yup, it's still on my list of things to look at before releasing the next beta, I'm going to get to that list once the hidden line stuff is complete.


> Could it be possible to ensure that the Blend is still performed - in the case, if only two edges are
> selected, even if they are touching?

Yes, this is actually what was intended to happen. There must be a bug that's preventing this from happening currently.

- Michael

Thank you!