Large cylinders

Crucial, the original poster asked "Is there any low-prim workaround for any kind of cylinder that is wider than 10m?"

I think my solution produces precisely what that question asked for.

In what way does it not?

On another tangent, didn't Cadroe Murphy make a scripted device that would allow such circles to be created by entering a few variables?

Suzanne, I think the original poster has already answered your question. Here, take a look:


The 2 things he specifically said he didn't want were flat panels and a lot of prims. Your circles, while well made, are the exact opposite of that. You've used boxes so the sides are flat, and you've used almost twice the amount of prims that the arc method requires. The 100M circle I outlined in my tutorial is only 40 prims, and it is perfectly round. The circles in the picture you posted appear to be around 70 prims and they're quite obviously faceted, not round.

As for the people who keep insisting that the arc method is flawed because arcs taken from elipses are not the same as arcs taken from circles, take a geometry class already. Every elipse in the universe is created by combining arcs from 4 different circles. That's what an elipse is, by definition. If you don't feel like opening a text book, then just see the attached animation for proof. Symetrical elipses are composed of arcs taken from 2 large circles and 2 small ones. Oblong elipses (eggs), which are impossible in SL as single prims, are made of arcs taken from 4 circles of varying sizes, usually one small, 2 large, and one medium.

The full answer to the question of whether or not an arc from an elipse is the same as an arc from a circle is that it depends on what part of the elipse you're talking about. As long as the section in question does not cross the border between circular sections, then the arc is part of a circle. If it does cross, then it's part of either a spiral or a parabola, depending on how much of the arc you include. Obviously the section I said to use (0.35 to 0.65, if you recall) only includes one circular section. For SL elipses, it's not until you get within about .3 from the end that you end up crossing the big circle/little circle border.

If after reading all this, and seeing the attached animation, you still don't believe that my circles are true circles, then my final answer is you just don't get it. For that I'm sorry, but ultimately if you choose to believe your preconceptions over the math and over the visuals then that's your perogative.

Anyway, I'll be dropping copies of large circles on the profiles of everyone who has asked for one next time I'm inworld.

EDIT: I've dropped a folder called "Large Circles" on everyone's profile in this thread except for Pete. For some reason there was no listing for Pete Moonlight in Find People. I can only assume Pete has cancelled his account. Pete, if you're still here, IM me inworld if you still want a copy. Also, I realize I've given copies even to people who didn't ask for them. This was meant in a spirit of generosity and helpfulness, but if you don't want them, feel free to delete them. I'd encourage you to hang on to them though. They are mighty useful. If anyone has questions, let me know.

The circle in my picture is about 240 meters in diameter, so it makes sense that it would have more facets than one only a hundred meters across. I may not have made it with the least number of prims, I didn't make it with that goal in mind. I made it because I saw the Rizal sim sitting flat and empty in the 1.6.12 preview grid and said "Hot Damn! Time for a giant circle!".
It' fun riding on the rezzers for big objects and watching the oddball view.

The "four circular arc" construction method displayed does not produce a true ellipse, it's a method of drawing a pseudo-ellipse when the inaccuracy doesn't matter. Such techniques were used by draftsmen before tools like CAD programs were available.

The statement that I've "used almost twice the amount of prims that the arc method requires" is simply false.

The "seamless" version included in the large cylinders folder we received uses an extra prim at each junction to hide the missing wedge on the interior, nearly doubling the prim count.

Well I read this entire post and found it quite amusing, in more ways than I expected. Somehow, there's an image developing in my head, one of little screaming kiddies in the sandbox throwing shapes at each other...

Well I think all presented methods here are simply wonderful and admirable. But COME ON FOLKS! If you think you need to burst into flames over boxes and arcs and circles and prims, I suggest you look for a 3rd life, cuz your Second Life seems to suck....

...just my opinion, nothing else. I just think we should try to see the positive side in each other. We're supposed to be adults, OK?

I wouldn't say it is a flawed method - using parts of circles, or even squashed circles (which are ellipses) to make approximations to large ellipses is I agree, the best way to do it within SL. But being a pedantic mathematician, I have to take issue with this statement that ellipses are created by taking arcs from 4 circles. Here is a proof that this cannot be true:

Definition of an ellipse, from http://mathworld.wolfram.com/Ellipse.html (11):

x^2/a^2 + y^2/b^2 = 1, where a and b are the semimajor and semiminor axes (if you like, the largest and smallest radii). It can be written in parametric form as:
(a*cos(t), b*sin(t)).

Also from that webpage, we see in (57), that the curvature of an ellipse at a point given by t in the above parametrised form, is:

K(t) = ab/(b^2 cos^2 (t) + a^2 sin^2 (t))^(3/2)

Curvature measures how much the curve is turning as it passes through a point. In particular, the curvature of a circle (taken from http://mathworld.wolfram.com/Circle.html (19)) is:

K(t) = 1/a, where a is the radius of the circle.

The curvature of a circle is constant (makes sense right? A circle is always turning at the same rate), whereas from the equation above, the curvature of an ellipse is not constant (because it depends on t), although it is a continuous function.

A curve made from 4 circles will have curvature that jumps discontinuously as you move along the curve from one circle to a different circle of different radius, whereas a true ellipse has curvature that changes but always continuously. Therefore the two curves cannot be the same.

Like I said, some people just don't get it. Your choice, your belief, your loss.

To those who do understand and who have thanked me for the help, you're welcome.

The end.

Not believing things that aren't true is not a loss. It's not true that ellipses are made from circular arcs. It not true that shapes made by using non-circular elliptical arcs are circles. It's not true that using huge regular polygons to approximate circles takes more prims than using non-circular elliptical arcs.

It is true that one can make a huge approximate circle using even less prims than either of the above mentioned methods by taking a cube and setting the Y or X axis top shear to .5. and the Z size to .01.

This yields a shape that is about 10 by 14 by .01. These can be positioned in a circle. This is probably the way to approximate a giant circle using the least prims possible, although it may not be the best looking circle one has even seen.

SL can't make huge circles out of a single prim.

Two good methods here for simulating such a shape. As with all simulations, each has its own advantages and flaws.

If you want surfaces that LOOK curved even when close up, Chosen's method is right for you.

If you want an automated system and don't mind flat sections, Cadroe's shape tool is a great option.

I use Xylor's ring maker and Cadroe's shape tool myself, but I'll have to try Chosen's method next time, just to see the difference.

(Though I'll probably automate it, 'cause that's just how I operate. )

You can make some pretty big uteruses with these things. Particularly with XyObject, I once saw a Xylorian womb that I was rebirthed out of.

Tiger, you left out my method, which doesn't use Cadroe's or Xylor's code at all, and makes the pieces of the polygon fit perfectly on both the outside and the inside with no over or underlap.

The ShapeMaker I am pretty sure doesn't make the pieces match together perfectly because is it is made to make more complex shapes than cylinders.

I am fairly sure the RingMaker doesn't make the plates fit perfectly with no overlap either.

--- edit --- according to Cadroe's post immediately following, both the ShapeMaker and the RingMaker do make both the inside and outside of the circle fit without gap or overlap -----------------

Hey - I wanted to clear up any confusion about the shapes ShapeMaker makes, but first let me just say I use it because it's easy and quick, and my priority is usually convenience rather than economy with prims. For all I know there are better ways to do it, but I'm honestly too lazy to find out. I didn't really look at the details of the other methods described here. Mime is money! (well ShapeMaker is free so in this case mime is mime I guess)

So anyway, ShapeMaker makes rings without seams or overlapping on the inside and outside (1st picture). With spheres and tori, I didn't see a way to have a perfect fit both inside and outside using this method with the box prims, so I made an arbitrary decision to have the gaps on the outside (2nd picture). There are no seams on the inside (3rd picture). BTW, RingMaker makes rings the same way as ShapeMaker, but I think the interface is a little more straightforward for making rings so I make it available in case anyone prefers it.

P.S. How the heck did Suzanne get an inline image in here?

Yeh if you pick the right amount of sides, with the right radius and the right other settings, you can get some decent results with Cadroe's tools. Obviously you're going to get holes if you try to use 8 sides on a 30m radius ring So there is a bit of experimenting that needs to be done. May be Cadroe could post some optimal sizes/sides for us?

Anyway, I love the RingMaker but I am constantly forced to use the ShapeMaker as well because the RingMaker doesn't do thickness or width.. or one of them.. anyway.. there's a couple of features in the ShapeMaker that RingMaker needs to make it a total ring making solution. As I sit here and think about it, I think it is the width of the blocks actually.. because as you decrease the radius, the ring-blocks change width and you end up using more rings than you should need to fill a very large cylindrical floor.

Once you get to a radius of 10, you can just select each block, set the top size to 0,0 and drag the tops in to the centre but for all the other rings you have to put up with an ever decreasing width, which ends up adding at least one or more rings to the total shape.

Would love to also see a semi-circle as well. Currently when I try to do this, it ends up with one of the pie-slices rotated past the 180 degree line so you don't get a true semi-circle. I'd like to see this feature added to the tools because semi-circles and quarter-circles have a lot more uses in general building than whole circles do, in my opinion.

Cadroe, I don't know why the images sometimes display inline and sometimes as links. I just did them as attachments using the manage attachments feature. It would be nice to know the trick of getting the images to do what you want, and to be the exact size you want to.

Since the RingMaker and ShapeMaker produce regular polygons with no gaps and no overlaps I suppose I will cease to mention my polygon maker in connection with making them.

I don't know if you can sit on the ShapeMaker or RingMaker while they do their stuff, or on the rezzor objects they might use, but it you can, it might be fun.

I believe that it does that when you attach only 1 image.
let's test it.
edit: seems to work.

yummm, science!

I just did a test with two images added as attachment, and they showed as links. I used the manage attachments feature to delete one of those, and the remaining one showed as an image. This is a forum with the IMG code turned off. It might be worth checking in the Gallery forum where the IMG code is turned on. That may not make any difference at all, but that's what one does experiments for, right?

One could certainly produce a script that would position non-circular elliptical arcs to produce an approximation to a right circular cylinder that would look smoother from the outside that one made from box prims with topsize adjusted, though it couldn't fit perfectly on both the outside and inside with no overlap. You might be able to get it to fit on the inside, overlap, and look just fine from the outside.

Actually, i managed to get Jeska to turn the [img ] tag on.
/invalid_link.html

with all the, ummmm, discussion on how to make large cylinders going on, I would like to discuss when to make large cylinders - or actually when not to.

the second post in this forum mentioned make a large tree, for example. remember: trees are not regular. if you craft one by hand using a few 10m cylinders that almost, but not quite fit together; have almost, but not quite the same eccenticity, and are almost, but not quite aligned - then once you drop the textures on, it's gonna look a lot more like a tree than a 30m perfect cylinder ever could.

there are definitely times you want perfection, but many, many time you should avoid it.

Suzanne,

I'm not sure why you would stop mentioning your own polygon maker. Does it have a name? Things have to have easily remembered names in SL. Someone should automate this ellipse method outlined by Chosen (someone else). I've seen a number of geodesic spheres in-world;are those done by Xylor's tools? I guessed so, since they're made of triangles.

I have a functioning prototype of a bezier curve tool, but it needs work. You position the handles and it reshapes the curve just like in a 2d app like Photoshop. If you combine a spline from this with something like RingMaker you get... a lathe operator This is basically how I created this saucer, but I had to do it manually. It'd be neat to see what people made with a tool like that (in addition to the obligatory giant champagne glass swimming pool).


I've been thinking about a generic component for rezzing and positioning blocks, instead of the task specific ones I keep adapting now. The BlockManager would be a script that took commands by link message. It would have commands for rezzing, freezing and deleting a set of blocks, and for changing any attributes of the rezzed blocks by referencing the block's index into the set. Each rezzed block would have a SmartBlock script that could adjust any attribute of the prim (position, totation, texture). This would just be a stand-alone, generalized version of the system I use now. Maybe if I made something like this available, it would encourage people to create more automated building tools. Basically all they would have to come up with would be the algorithms. For instance, it wasn't too much trouble for me to create these shapes once I learned about functions for parametric curves because I already had the code for manipulating sets of prims:


http://www.spinmass.com/2life/pics2/Epitrochoid2.jpg

EDIT - The IMG tag doesn't seem to be showing the second pic so I just added a link to it.

Gaz - Unfortunately I don't think I've used RingMaker since I wrote ShapeMaker, so it's a bit hazy. What I remember being better about RingMaker is that the radius you give it is the distance to the outside edge of the ring, versus ShapeMaker where it's the center of the ring and you have to do some calculation to get the total ring diamater you want. I think that's lame, but it sounds like maybe RingMaker creates its own problems. I don't think I'll be making any more changes to RingMaker, but hopefully I'll find time to improve ShapeMaker. That would include a paramater to rotate the whole ring by any amount, which would solve your problem with semicircles not lining up the way you expect. I had to link and rotate a set of columns by hand last week myself because ShapeMaker doesn't have this ability yet, so I know what you're talking about.

I used to use an old drafting program with a digitizing tablet, and it used a type of spline that worked just perfectly with a digitizing tablet.

I believe the spine was a cubic spline, maybe it was called a cubic B-spline. Unlike a Bezier curve, there aren't any handles off the curve, the curve passed through all the points that you make tracing the drawing on the tablet, and you just make more frequent points in areas where the curve behaves unexpectedly or has abrupt changes.

The math for this type of cube is a bit tougher than the Bezier curve, I believe, I suspect the relative simplicity of the Bezier curve's math is part of the reason for it's popularity.

At any rate, perhaps someone can produce such a spline generator for SL.

If the Ringmaker makes regular polygons that meet perfectly with no overlap then it probably uses the same trick mine does to make the shapes because there is no other way in SL to do it, and has some sort of sensible control approach. I have avoided looking at the code for Ringmaker and Shapemaker because if I read the good part of the code then I will never be able to get the joy of writing something similar on my own own.