Building issues, advice needed

[adloram]

I can't seem to be able to do anything I put my mind to when dealing with Alias.

I'm a beginner, but for the first time in some I felt like I got a bit of Alias finally right and decided to build two models that I had in mind for a while.

However, no matter the effort, the different ways and build strategies, I always get to a point where it seems to me that it should be something that alias should be able to deal with very easily, but instead hangs had and can't seem able to give me the surface quality I'm searching.

Since this has happened before in other cad software, is there something wrong in how I build my models?

Maybe it's some finesse, maybe it's the complete building strategy that's wrong, but it seems to me I'm not doing anything space-rockety. I just don't know what to do at this point, went through all the beautiful foundation tutorials and then some on youtube.

Thank you for your advice.

(in the second image, there are two deformations in the surface, visible when switching to zebra analysis)

[adloram]

Ehmm, is this an active forum?

[duaxiong]

---

@adloram wrote:

Ehmm, is this an active forum?

---

Sometimes...

There is no easy way to answer the questions you have, especially without the data.  Yes, the surfaces appear fairly simple, but to get to an acceptable solution requires fairly deep knowledge.

My advice:

A

Whenever possible, it is better to use a natural boundaries instead of a Trimmed Edge.  In this case, you would do better to Rotate the Sphere 90 degrees in the Z axis and Detach it where you want to cut it.

Even then...if this is the default Sphere, it will be degree 3;  This means internal discontinuities along that Detached edge, so you'll want to rebuild to degree 5 or just live with G1 instead of G2.

B

Transistion surfaces are much too heavy;  the chances of it having ripples or waves in the surface are very high.  Build it to lower spans and use Align to the Sphere.

C

Not a big issue, but this surface is heavier than it needs to be, assuming the shape is flat.

D

G1 continuity means that you should not expect G2 for your transition surfaces.

E

Again, G1 contunuity...here, requires careful control to get acceptable shape.

If I were to do this, personally, I would first rotate the Sphere, and make it degree 5.  Detach it at the centerline...not where you cut it.  Detach it again where I want to break the transition surfaces.  

Delete the surface at "E", and build a new surface what will bridge to the opposite side in one surface and not two.  Rebuild it so that it will match parameters with the corresponding Sphere slice.

Build the two transition surfaces with Freeform Blend or Square with G2 on the ends, but G1 in between each other.  Even then, continuity between the two transition surfaces will be tricky and may require some CV pulling.

See below for an example: (Note, not perfect, but mostly there.)

[duaxiong]

Your second question is an easier one.  This is a two-step process...example below:

First, define the boundaries of your transition.  The vertical cylinder is a natural boundary.  The cut on the lower cylinder is a Trimmed curved projection.

Build the transition surface making sure that it aligns G2 to the vertical cylinder, matching its parameterization with Explicit Control.

Fix the lower transition with Align; Notice I needed to make it heavier to meet my tolerance settings.  If I want it to be lighter, I can detach at the internal isoparms before Aligning...but I usually don't bother because it can complicate things if I need to make a change to this area down the line.  Alternatively, I can also just ignore the G2 continuity warnings because at this point, there will be very little visual difference between a G1 and G2 transition.

[adloram]

First, thank you Duaxiong for all the time and effort in going through the model, point by point, everything is clear. There's still something that I can't seem to get straight in my mind, I guess it'll take some more time.

On my own I tried going through the first model again, trying a different build strategy.

I drew a "scheme" that seemed realistic in the attempt of understanding how many surfaces I needed if I wanted to do it more cleanly (class A surfaces, in theory all G2).

 With these results:

The zebra is really looking weird...

So, still not there, but at least some progress.

Here's file if you want to give a look: (on dropbox as I couldn't get it attached through the forum)

https://dl.dropboxusercontent.com/u/471039/Funnel_MMPP.wire

Thanks again for the huge help.

[duaxiong]

I am more bothered by your practice of Aligning G2 to perfectly flat surfaces; This creates flat spots where you may not necessarily want them.

I don't think the zebras look particularly weird, but your Hulls are not good.

A and B

The irregularities in these two rows of CVs should not be there if you want a nice smooth surface.

C

The distribution in your CVs is strange, given the somewhat regular shape; It's not technically wrong, just not good practice. I would make this degree 2 (or at most 3) and it should have a much more even CV distribution.

D and E

Aligning G2 to these linear surfaces will cause flat spots, as is evident in the gap at E.  You can also see it in the curvature plot. If you mirror your surfaces and put the zebras on it, you will see the flat spots.  

So, if this is not desirable, you should not build it this way.

F

I don't think you need this whole row of surfaces.  Aligning right to the Sphere would give you much more flexibility and produce a more flowing surface.

I don't know why you have a deviation at your ??? location, but a Colinear Align with Curvature will fix it.

[adloram]

Great, thanks, I'll fix these points later today.

Thanks for pointing the flats out, I couldn't find this piece of knowledge anywhere, and I've wondered if it was a visualisation error or not. I guessed if G2 "applies" then it should be fine, and I guessed wrong.

Some hulls are displaced weirdly as some technical surfacing videos talk about accelerating the surfaces before making transitional ones. But the weirdness comes from keeping the curve as I wanted + some acceleration.

[adloram]

Now I can't even close it, should something be implied curvature/tangency?

Here's the file:

https://dl.dropboxusercontent.com/u/471039/Funnel_MMPP3.wire

[adloram]

Something I forgot to mention, I need the F surfaces to make sure that the sphere is "sphere" until at least that point, for other stuff to be placed in that go past the vertical diameter.

[duaxiong]

G0, G1, G2, or G3 continuity levels do not define whether surfaces good or bad....they are just different levels of continuity, that's it.  It's up to you to decide what level you need or find acceptable.

In most cases, you only need a very little bit of acceleration.  

You still need to Rotate your Sphere...notice where the pole is.  Yours is in the wrong place.

There's something you should know about the Sphere...it's not really "circular";  It's close, but those are not arcs, so if this is a bearing surface, you should use a rational Sphere.  If you put a comb plot anywhere on that Sphere, you will see that the radius measurements vary.

That said, the normal Sphere is easier to work with than a rational Sphere...so unless you really REALLY need it I would not bother.

Lastly, on your transition surfaces it is not absolutely necessary, but it would be preferable to get your hulls to line up a bit more (Yellow lines).  If you don't you run the risk of the surface folding onto itself as it stretches to Align to the differing edges.

The intersection at "A" is G1.  This means you cannot have G2 continuity at "B" and expect to not have a gap at "C".

D

Surface is too heavy for what it is.  This is giving you extra CVs that do not do anything...and may actually be distorting things or causing ripples at "E".  Notice how the Hull zig-zags...this is what you do if you want waves in your surface.

F

This surface is also too heavy.  Furthermore, the weirdly uneven CV distribution is unacceptable for a surface that is supposed to be simple and smooth.

Overall, your surfaces are much too heavy and the weight is causing complications.  Below is a comb plot on your "sphere"...the original Alias Sphere does not look like this, so I don't know where this came from, but it should be fixed.

[duaxiong]

Just for fun, I took a proper stab at this.  The results are attached;  Took about 15 minutes...it was not easy though, required some manual CV pulling.

Wire file attached for your reference.

[adloram]

Yeah, I've read about rational circles/spheres, but thanks for pointing it out. Everything I (sort of) know comes from Alias Workbench (aliasworkbench.com) which is just another round-up of the official learning material that Autodesk offers on the Alias help.

Thank you for the file, but I'd like to know more how you solved the many issues I'm having 😞 (as you probably understood, I'm not here for the file)

I can see a different disposition of the major driving curves, how did you achieve those?

Is it an isocurve of the sphere? But then, did you operate a freeformblend and snapped onto it, or did you build the blend curves? (I'm going now through the file you attached)

First of all I'll try again rotating the sphere as I didn't get that from reading above, my bad.

Most of the curves with that weird curvature are Fit curves on the trimmed edges, revolved to have the same number of CVs along the horizontal "band".

Also, I've heard about it, but every time I try to manually pull CVs and Hulls to clean stuff myself with the tool named "xfrmcv" Move CV, first Alias reminds me that this will break the construction history (and the aligns, and the continuity) and then even If I somehow manage to create a decent flow of hulls that seems clean enough, I never reach a decent continuity, and trying to align again screws any work I've done manually.

You're being of incredible help Duaxiong, thank you so much.

[adloram]

For example, what's the workflow here?

Did you mirror the top curve on the other side? Stitched the surface into one? or made the two verticals and then freeformblend?

Sorry for the quantity of questions.

[duaxiong]

There are many different ways to build it, but this is what I did:

The "tube" part is an extrusion of four curves.  The profile curves are in there somewhere...maybe Templated.  The upper and lower rounded ends started as a degree 5 circle and were detached in half.  The vertical flat part started as a straight line.  Everything is Aligned and properly shaped and then Extruded to create the "tube".

The spherical part started as a Sphere primitive.  It was rebuilt to degree 5 in both directions and then Detached at the appropriate locations.

For the blending surfaces, the upper transitional surface started as a linear Skin and then simply Aligned colinear at either ends.  The side transitional surface was a Square, I think...with G2 Alignment (colinear) to the Sphere and flat part of the tube.

This left a discontinuity between the two transitional surfaces, so this had to be "massaged" into shape by moving the CVs around on both surfaces until the highlights looked good and the continuity checker turned green.

Sorry, probably difficult to follow.

Some tips:

- I make use of the Symmetric Modeling Tool everywhere I can;  it's very good a guaranteeing that things match from side to side and the History means you only need to worry about making changes to just the one side.

- Don't be afraid to delete History.

- Don't always rely on Square's continuity tools to achieve your alignments.  The Align tool is more flexible and much better at it.