killer modeling problem...

[meshpig]

Hi

Suck on this if you will, I have for the better part of today.


There's more to this than meets the eye.

I'm working on a conical sieve. Simple you say, just boolean the cone and be done with it.

Great, but the holes are actually drilled in a spiral fashion. There are also too many to boolean on an angle and if you drill flat then magnet or morph it back to a cone the holes are distended.

So, with a spiral/UV to weight/stretch using the w map as the falloff got the vertical proportions tapered so I can bevel the holes but they increase in number across the girth too as you go up the spiral so linear organization is irrelevant. ie. The holes get bigger as the circumference increases.

Before taper, no bevels;


62990

A conical sieve...
62991

Thoughts?

m

[cagey5]

With something like this I would always try and replicate the way it was actually made. And it was probably formed into the conical shape before machining, otherwise they'd have the same problems of elongation as you have. With that in mind I would use something like Art Sphere's clone+ and if there are too many polys to boolean it all in one go I'd break it down into manageable segments.

[jin choung]

how about generating the hole booleans with a radial clone with an offset? the one generally used to make a spiral staircase?

that ought to get you the spiral hole templates....

jin

[meshpig]

Thanks cagey... it certainly looks like some robot has poked holes in it whilst the steel is still hot. Clone + looks like it would do the trick but there's no mac version.

Jin, the radial clone doesn't have an offset but I think the spiral hole boolean rather than the spiral cone is the way to go. Just did a quick test and seems really straightforward.

Thanks!!

m

63032

[scratch33]

Copy your object to another layer, kill poly, create your boolean object to another layer, point clone your boolean object...

[TripD]

Oh wow! You sure know how to pick em! :) Are you planning to clean up those edges when done? :cursin:

[Hopper]

Copy your object to another layer, kill poly, create your boolean object to another layer, point clone your boolean object...
That's my take, but I was thinking of shrinking the cone, then using point clone with cylinders. Use some jitter to make the holes non-uniform, etc...

[meshpig]

OK, here's the "drill". I have tried every conceivable method today and am pretty sure there's another day's work figuring out how to diminish the number of polys logarithmically. ie. so the number of holes per row diminishes also


What I need is a script or the ability to write one because it's mathematically possible.

So, I tried weight mapping a long strip then stretching it with a falloff so the end polys were pre-distended and bending it but ran into problems with the 1023 point limit per polygon. Spline patching didn't like the dual spiral either as I couldn't get it to evenly patch.

Went for a boolean at the end of the day. The result has the makings of what I want but still has too many holes in the bottom rung as it were. I can see a lot of tweaking coming on...

I like days like these...


m:)
63085

63084

[cagey5]

Have you watched the Helix tutorial by Proton lately ;)

[meshpig]

That's my take, but I was thinking of shrinking the cone, then using point clone with cylinders. Use some jitter to make the holes non-uniform, etc...

I'll have to check point clone out because I've actually never used it:eek:

m

[cagey5]

Using that method to create a helix then rail cloning along it with a uniform spacing will give you evenly spaced cylinders ready for a boolean operation.

[colkai]

I'll have to check point clone out because I've actually never used it:eek:

m

Oh man, for stuff like this, it's a godsend.
I did something a bit back needing lots of correctly placed items and the copy / collapse polys / point clone makes it so quick. :)
One thing, make sure to unweld (Ctrl+u) first or you end up with just 1 point .. ask me how I found that one out. ;)

[meshpig]

Have you watched the Helix tutorial by Proton lately ;)

Ah, brilliant! Didn't think of rail clone. First thing tomorrow. Thanks!

m

[cagey5]

Another pic. Taken one stage further.

[meshpig]

Oh man, for stuff like this, it's a godsend.
I did something a bit back needing lots of correctly placed items and the copy / collapse polys / point clone makes it so quick. :)
One thing, make sure to unweld (Ctrl+u) first or you end up with just 1 point .. ask me how I found that one out. ;)

Col, I can imagine... do tell. Good going will check it out.

[meshpig]

Another pic. Taken one stage further.

OK cagey, you seem to have nailed it. Even more brilliant!!!

- again, first thing:thumbsup:

m

[meshpig]

... bit of an ordeal. Doh, Point clone plus for mac is already on board but no one tool does the job on it's own.

-The sheer number of holes means LOTS of manual tweaking. It's way too big @ 35 000 polys too.

m

63227

[cagey5]

Does this grill tutorial offer any hope?

http://www.ifw2.com/IFW2_Nodal/IFW2N_Grille_Tutorial.html

[Tobian]

That looks like a really sticky problem!

Not sure if it's too late, but I did think why not use true fillet to make the holes? Just select the points you want, set the segments to 2 and use the direction to backwards. And them maybe run a level 1 subdivide to make it quads?

I did a quick example (not the same type of spiral, but it was just a proof of concept hehe) selected on the corners, used truefilet on the holes with 2 segments, then connected up all the points, so they were all quads again. It's a lot lighter mesh. It has the advantage that all te holes will be the same size and there's no boolean involved :D

[meshpig]

Does this grill tutorial offer any hope?

http://www.ifw2.com/IFW2_Nodal/IFW2N_Grille_Tutorial.html

Thanks for the input C5, yes but I sort of need to nail it as a model.

m

[meshpig]

That looks like a really sticky problem!

Not sure if it's too late, but I did think why not use true fillet to make the holes? Just select the points you want, set the segments to 2 and use the direction to backwards. And them maybe run a level 1 subdivide to make it quads?

I did a quick example (not the same type of spiral, but it was just a proof of concept hehe) selected on the corners, used truefilet on the holes with 2 segments, then connected up all the points, so they were all quads again. It's a lot lighter mesh. It has the advantage that all te holes will be the same size and there's no boolean involved :D

No, not too late. Interesting, another tool I haven't done much with.

thanks.

m

[K-Dawg]

Hey tobian, nice idea. I had a similar thought but with rounder instead :)
Nice thing to try. I'll give it a shot too. See if I can get that done :D

Greetz

[Tobian]

Well with rounder, if you selected a bunch of points which were connected, it would connect them all up :D This just rounds the corner. It also works on open poly's too, making it great for working with stencils! :)

[K-Dawg]

Yeah sure, but I only used every other point on the same row and one row obove the same, just one point on the side.

This is what I cane up with after a lil cleaning up an tweaking. All subpatchable :)

Might not be the cleanest model, but was just more a try to see if it works :)

Greetz

[meshpig]

OK, I've been there in conical sieve land and done that...:hammer: this will do for the time being.

63274

[Tobian]

Well some of the rows look slightly out of sync, but it's a great model, well done :)

[meshpig]

Well some of the rows look slightly out of sync, but it's a great model, well done :)

Hey thanks Tobian, kinda mad because I also didn't want them to be in sync. Check the initial pic. No matter, I asked for it.


m:)

[K-Dawg]

Hey, as long as it comes out good :)

I like the surface you used. Nice going :thumbsup: