PDA

View Full Version : Barber's Pole Spiral



pixeltek
08-18-2005, 12:58 PM
I probably can figure this out, but maybe there's a smart way to do this. How would I apply a color spiral band on a cylinder e.g. barber's pole?

Karl
http://www.cosmic-pearl.com/

evenflcw
08-18-2005, 03:28 PM
You could just UV map it, then skew the UVs (using Move tool with linear falloff) so the colored lines meet up at the UV seem.

Silkrooster
08-18-2005, 05:25 PM
Actually you did the opposite of a flag pole I did. I created the image map with the stripes at an angle, apporox. 45 deg. Then applied this to the uv map.
Same results just reversed.
Silk

skanky
08-18-2005, 06:20 PM
No need for an image map. Just make the base pole white then put a procedural checkerboard texture on it. Scale it something like 50m, 0.2m, 50m, then rotate the texture on bank to the angle you want.

Man, didn't Alan Chan have something ages back in a lightwave book that involved rotating the pole, then morphing it back into position?

evenflcw
08-18-2005, 06:33 PM
Yes, both approaches will work and they'll look pretty simular, but they won't be the same. By distorting the UVs instead of the image, the lines will always render clean nomatter how close you get, because they are as clean as they can get. Do it the other way, and you have to make sure the images resolution is high enough for the diagonal lines not to look jagged/pixelated. Obviously, it's not much of an issue, since the image won't take much space anyhow and the item probably won't render big enough for it ever to become an issue. But from an academic point of view...

EDIT: My first thought was also to use the checkered procedural, but after fiddling with it for a second without success I realised it could never work. Atleast I was not successfull.

The morphing variant I think in essence would be about the same as using UVs. But instead of distorting the projection, you distort the mesh. So we have come full circle here. Disorting the mesh (after projection), the UVs or image :)

skanky
08-18-2005, 06:58 PM
My first thought was also to use the checkered procedural, but after fiddling with it for a second without success I realised it could never work. Atleast I was not successfull.

Hmmm, maybe I'm missing the bleeding obvious here. OK, so it's not a spiral, but on a pole you can't see that without walking around the pole. What did I miss? Surface setting image and dodgy render attached...

I agree with you on the distorted UV's with image mapping - as long as the image is created pixel perfect, as in, no anti aliasing.

Cheers

evenflcw
08-18-2005, 08:43 PM
You didn't miss anything. You are right that the procedural would work ok for a still. Your screencap proves that. But for animation it wouldn't be enough in any situation, as these poles are suppose to spin continously, no?

pixeltek
08-18-2005, 09:24 PM
Thanks all. Great ideas, and yes, the poles are supposed to spin.

I guess it's not all that easy...mumble..mumble. Have to think about it some more. Two interlocking flat springs,...one red one white, offset 180 degrees?

"I'm trying to think, but nothing happens!" Curley - The Three Stooges.

Somebody surely solved this thing through some sneaky texturing method.

Karl

pixeltek
08-18-2005, 09:30 PM
Wait a minute. The first proposed method, if offset just right (the width of the stripes), will create a spiral, won't it? Duh! ;D

Karl

John Fornasar
08-18-2005, 09:42 PM
I use this jpg cylinder mapped on the pole, rotate the pole 360 degrees over 60 frames (or whatever speed you want)

C:\Data\LightWave\MainFire\Images\BarberPole.JPG

edit: I'm not sure what's going on with the attachment, I'm getting a message that POST Requests from foreign servers are not allowed..

if you want the .jpg and my pole model, (stripes rotated to a diagonal) send your email to me and I'll forward it

evenflcw
08-18-2005, 09:58 PM
Ya, that's the idea, to "offset just right"! You could do it by eye or simple math. I had 5 pairs of red-white stripes so I offset the UVs by (100/5) 20% on V.

You could get by with a 2 pixel image aswell I guess if you distort your UVs enough (By default the UV image is forever repeating in U and V so you just need to stretch it in V to introduce more stripes). But this sounds abit extreme even for me.

pixeltek
08-18-2005, 10:09 PM
Yes, thanks, evenflcw. I'll play with that.

John, interesting. Any jaggies from that method?
BTW, hehe, you're in Brooklyn, that explains it. Actually, maybe you were being leap-frogged around to some server in India.

Karl

skanky
08-18-2005, 10:22 PM
Ahh yes, rotation. I didn't think of that. So that's what I was missing out on! I'm going to have a go at doing it with rotation procedurally, but off the top of my head, I don't hold out much hope...

John Fornasar
08-19-2005, 07:19 AM
I use this .jpg, cylinder mapped on the pole

edit: sorry about the size, it's been a few years since I used it in the background of the scene...

also, the jpg is the red, white and blue striped version - in Brooklyn, there were a lot of these (including the barber shop up the street that I looked at when I was doing it)
here's the pole .lwo - rotate layer 2 360 degrees over 60 seconds (or whatever speed you want)

GregMalick
08-19-2005, 10:42 AM
Hey Skanky,

You are almost there.
Just can put an Envelope on the Y-Position of that procedural.
It makes it look like it's turning.

skanky
08-19-2005, 05:01 PM
Hey Greg, almost. Moving the texture on Y has to be to combined with targeting the camera - otherwise it looks all wrong when the object is rotated 90 to camera. And, being very retentive, it won't look quite right when the camera moves during an animation. I'd stick with an image....

GregMalick
08-19-2005, 09:09 PM
oooh yeah.... :foreheads

My avi did not have any camera movement - I was just testing and that simple shot looked OK.

pixeltek
08-20-2005, 08:26 AM
Thanks again, all, for your excellent, thoughtful, and helpful replies. John and evenflcw, that is essentially the identical result. Problem solved.

Karl