Is there any way to model a perfect sphere that can be extremely squashed and stretched in any direction with out any creasing (that is a result of typical sphere making methods)? Right now, any method that I try in which all the CPs connect to eachother results in problematic shading on the sphere.

Here are some gifs-

• Wire
• No Wire

uh... are these being rendered in preview? from the modeling window?

If so, use page up and page down keys to change resolution on your preview... my avatar in progress is a sphere made by:
1. lathing a 3cp line
2. deleting the lower 2 sections (leaving a circle in the top view)
3. rotating the circle 90 degrees along the x axis
4. deleting left half of circle.
5. breaking line between ends of half circle
6. moving remaining arc to right one "bump" with right arrow key
7. lathe arc.

this will give you a sphere with a slightly open top and bottom.

I believe it is squashable with no creases...

but, I haven't tried it yet.
HTH

<edit>
re reading your post.... I see...



that may be the problem... I don't think you absolutely must join all the cps... but if you do, you may have to adjust the bias of the cps that are causing the creases... there is a tutorial that quickly shows this... I'll find it and report back... here it is: <edit off>qvids clcik on the green sphere in the lower box... larry does it on the left, I do it on the right... he uses a 2 cp spline... i did 3 I dunno why, I just did.

Hi:

I'm glad to see that that video tutorial is still being used..

I have no idea why I do the sphere on the left.. Just seems more natural.. I wonder if that has someting to do with left/right hemisphere brain functions.

It is a miricle that the tutorials were created in teh first place! I was beginning to loose interest in finishing em.

Anyways, enjoy them.
-Larry B

of course, they're being used... they are quick reminders for us oldsters! and the only thhing missing for a newbie is how to model a dragon with 5 cps... and get a job at super mega studios, too!

Thanks larry... which leg should I put in my pants first.... I forget?

Well, the videos do move a little slow for people who know the program already..

Unfortunatly I forgot about the 'make dragon' button. One of these days I'll get around to finishing the tutorial that reveles where THAT button is! Untill then people have gotta wait.

Now you wer asking about puting on pants.. I guess you HAVE forgotten.. you put both legs in at the same time, Stand on your head, and do the hip-dance untill it all shakes into place!

Anyways, you all have fun, and enjoy the videos

-Larry B

Correct me if I am wrong... I think you guys may have missed the point.

TheWoz is looking for a sphere with "equal" patches all around the surface, instead of the "squished poles" of the traditional lathing method.

I use to think there would be some sort of mathematical magical bias formula to do this..

... Unfortunately you always always end up with those 3 point patches at the "corners"... like a rounded cube (I have a spherical Rubik's Cube... like that). Nothing you can do by hand would ever give you a completely smooth sphere.

This "Holy Grail" of splinage is not achievable I think... I tried once... for a couple of days I tried to create a smooth closed sphere... I came to the realization this can't be done.

I think it has to do with the speed of light and PI... not sure.



Vernon "!" Zehr

Vern says it can't be done... it probably can't.

Maybe I did miss the point... if Woz wants to squash and stretch in any direction... perhaps the holes at the poles would make the sphere pucker under stress.

And, I can think of one or two instances where a sphere might have to be "perfect" so that it could deform in an animation... I guess this is one of those things where, as animators and "artists" we have to fake reality in post?

But, if a sphere is squashing and stretching, it is elastic... and the puckering would only show for a frame or so ... making it nearly invisible, no? I'm overthinking it again... but why is it so critical?

There are ways around it... but...

...in a perfect world... it would be kind of nice to be able to have a smooth... uniform sphere. This would aid in animation and texturing.

Distorting a lathed sphere can cause creasing and puckering at either ends of the poles.. a wobbly soap bubble for instance... or a slightly deflated kick ball.

... if a perfect world were possible... this would be number 689 on my list... I haven't really encountered too many situations that couldn't be worked around.

...dang! I never tried using porcelain on any of my sphere experiments... that might smooth out those 3 pointers at the corners!

Vernon "!" Zehr

It wasn't very critical... I mean, the way I will be squashing and stretching the sphere, the distortion will be noticable, but I may just use a prop for that. I always wondered if and how it could be done though. Thanks for the replys, and if anyone ever does manage the impossible, let me know.

Oh, btw Vern, I just tried porcelain... Not really that effective from my experiments. Thanks though.

I've been working on a related idea, which is a sphere whose CPs are distributed more or less evenly across the surface. This was an outgrowth of trying to run SimCloth on a sphere, because if you make a lathed sphere into a cloth group, the poles act like weights due to the higher density of CPs.

So I tried modeling a "basketball" sphere (shown below left). I can tell you how it was done in another post if you need to know. The key is that when you have to connect up two of the diagonal longitude lines through the middle of a segment of a vertical longitude line, you select one spline in the vertical longitude and press Y to create a new CP at the midpoint.

The basketball sphere doesn't exhibit creasing at the poles when you distort it, but it still doesn't work well in cloth simulations because the poles are slightly heavier than the rest of the mesh. Perhaps the real holy grail would be a "volleyball" model (below right). This is a screenshot taken from Lightwave, IIRC. I can't figure out a way to model something like this in A:M with geometric accuracy. Anyone have any ideas?

julian,

at the volleyball's three-spline intersections, make them 3 cp loops, so each patch would be a 5-pointer.

you could extend the splines out to control the curve without resorting to bias handles, but you'd have 8 very small holes in your sphere.

not sure if it would work, but i might give it a go when i update my space globe background.

thanks for the leads!
-jon

Julian, that type of sphere seems perfect. I feel stupid now for not being able to come with something like that on my own. Is it possible for you to post a quick tut. or mdl for me to see any bias handles you have set? When I hit 'y' to add a cp, the spline looses its "sphere-ieness". Thanks

I've tried Julian's solution a billion times, however I always seem to fail somewhere, because it never gets 100% round! Or anywhere close to 100, for that matter...

However, Julian's seems pretty damn close, could you show us a shot from the front so that we can see the tough splineage from the side?

Julian, Vern....


What would happen if you used a distortion map (of a grey scale sphere) on a grid plane? Would that give you the cps/patches of a 1/2 soccerball? That could then be copy flip attatched?

That's probably too easy... or not high enough rez or something... It was just a thought.

It helps if, before you insert extra CPs, you cut and paste a perfectly circular spline and then lock it off so that you can use it as a template. Anyway, writing a tutorial would be pretty complicated, so I'm just going to make the model file available here. It has the Porcelain material embedded:

BasketballSphere.zip

I don't think the three-spline intersection will be a problem so long as the curvature and bias are correct. The problem is, I don't know how to get the right curvature in the first place. Basically, this is a projection of the surface of a 4x4x4 cube onto a sphere, so there should be a way to calculate the coordinates of every point, but I'm not sure how.

(P.S. In case you missed it because of the page break, I made the "basketball" sphere model available in the last post of the previous page.)

What I said was....



what I should have said was DISPLACEMENT map rather than distortion map...

Knowing what little I do about mapping... why wouldn't that work? for most displacement mapping... from my understanding, the grid mesh must be fine... lots of patches... But if it were not so fine, and the contrast of the map was great...wouldn't that force the grid to conform to the map???

If not, there may be an algorithm or some other math function (who knows?) which would give you the coordinates for the cps and the splines of a soccer ball type grid...

If nothing else, somebody could scan a soccer ball at the intersections of the seams and have a perfect sphere, no?

soccer ball old idea

http://www.hash.com/forums/index.php?showt...13512&hl=soccer

Great model Julian... It distorts wonderfully. I threw a quick little mov together.
Smile

John, We can always count on you...!

unfortunately, that's too accurate a model of a socker ball!

Julian's example (on the right) gets me closer to the "perfect" sphere... I think... One that is infinitely squishable, and may not need porcelain...

"The Holeless Grail" as it were..

I'll work on it... there's gotta be a way... "no holes at the poles" that's my motto! And it's all Da Woz's fault!

A soccer ball type of tessellation isn't going to make a smooth sphere because the pentagons and hexagons will end up being broken up into triangles. Yves ran into this problem back in the version 8 days:

http://www.ypoart.com/Downloads/SoccerBall.htm

I have it!

In this thread on the SDK forum, Emilio Le Roux has been working on a plugin called the Spherize wizard. This is the perfect tool for making the volleyball sphere.

Here's the model.

I started off with a cube [A] modeled so that all of its splines were continuous except at the corners, and with all of the CPs smoothed instead of peaked. (The spacing of the grid isn't completely even -- I get better results with Spherize if I take the four center squares on each face and scale them down 85%.)

[B] shows the result after running the Spherize plugin.

As can be seen in [C], the three-spline intersections do show a little bit of artifacting at the edges, even when Porcelain is used. I'll try applying SimCloth to this model later.

Hi:

Is there a way to use negitive bone cutter?

If so you should be able to male a sphere and a cube and have the sphere cut the cube corners off to make a sphere.

I have tried a couple quick test without success. But I'll try again later.

-Larry B

Bah! Well It certainly looks like you found it Julian...

I notice the artifacts (because you point them out) and look forward to seeing the result with cloth.

Now, does cloth "prove" it to be a true sphere? No distortion I presume, if a true sphere. I'm not using cloth yet... Heck, I am just making spheres and such...

I do have an idea that may work anyway... Gonna try it.

Boolean cutters are just a rendering trick; they don't produce real geometry. When the scene is rendered, the cutter and the volume where it intersects the rest of the model are made to look invisible. Since the initial problem is trying to make a sphere that can be distorted without creasing, you're now left with having to make a spherical Boolean cutter that can be distorted without creasing, so there's no improvement.

Trying to compute the real geometry of the intersection of two spline shapes is just too big of a programming challenge for the Hash team to take on. It would have to involve some fundamental math research to calculate the shape of the intersection's outline, and to connect the splines of the inside of the cutter shape and the outside of the model to it. It may require some artificially intelligent interface design so that the software can guess what the intersection is supposed to look like. Maybe some enterprising computer science grad student can write a PhD thesis about it.