Hi all,

I made this for a competition (to make a visual representation of a radio
station) a while ago. Since I didn't make it to the finalist stage
virtually no-one ever got to see it, so I thought I'd show it off here.

It's a ~16MB XVid file:

http://www.guidetomp3players.com/final_compressed.avi

The music was provided by a cool band called 'The Levitators'.

It started out from a little routine I had that could break any CSG object
up into little blocks and re-arrage them into another object:

http://www.guidetomp3players.com/torus.avi

What do you think?

Post a reply to this message

"scam" <sca### [at] mailusydeduau> wrote:
> Hi all,
....
> It started out from a little routine I had that could break any CSG object
> up into little blocks and re-arrage them into another object:
>
> http://www.guidetomp3players.com/torus.avi
>
> What do you think?

I think it's a nifty effect. (Nicely used in the radio spot as well.) I
*think* I have an idea of how it works, but if you regard the code as
something you would be willing to post, it would be appreciated. (If not,
that would be understandable too - no hard feelings.) It might yield some
interesting results to apply the same technique to the elements comprising
two blob objects. Thanks for the look.

Regards,
Mike C.

Post a reply to this message

"Mike the Elder" <zer### [at] wyanorg> wrote:
> I think it's a nifty effect. (Nicely used in the radio spot as well.) I
> *think* I have an idea of how it works, but if you regard the code as
> something you would be willing to post, it would be appreciated. (If not,
> that would be understandable too - no hard feelings.) It might yield some
> interesting results to apply the same technique to the elements comprising
> two blob objects. Thanks for the look.
>
> Regards,
> Mike C.

Thanks Mike. I'm happy to share the code, however in its current form it is
unintelligible to everyone but myself. I'll clean it up and post it once
the working week is done. Also I may have oversold the algorithm a bit, it
is very simple and not subtle or clever at all.

In short though, here is how the effect works:

1. Divide the beginning object up into blocks: Set a grid resolution then
use the 'inside' function to determine if the points of your grid lie
inside or outside the object. The internal points are of course the block
positions.
2. Make a list of the position vectors of each block: listA
3. Divide the final object up into a list of blocks: listB
4. Sort both lists so that the ones furthest from the origin appear first
(not very subtle!)
5. Write both listA and listB to a file so you don't have to discretize the
beginning and end objects at each step of the animation.

And here's the trick:

6. Translate the blocks from object A with position vector listA[ii] to the
position x*(listA[ii] + listB[ii]) for all ii, where x >= 1
7. Translate the block at x*(listA[ii] + listB[ii]) to position listB[ii]
for all ii

That's the basic idea. In effect it is the coarsest possible way to morph
between two objects, but it still looks kind of cool.

Some of the problems with this approach are: in step 4 no distinction is
made between internal blocks (those which have no face visible form the
outside) and external blocks (when the two objects have different numbers
of blocks the external ones should have a higher priority), blocks will
frequently collide and pass straight through each other, and the objects
need to be roughly the same volume.

Post a reply to this message

it looks pretty neat when the visible sides of the cubes change to show the
different logos but the morphing is too fast for my taste. one can not
actually see anything but only *guess* that it's morphing.

Post a reply to this message

"scam" <sca### [at] mailusydeduau> wrote in message 
news:web.4642d1471e4e942a5897f1740@news.povray.org...

> What do you think?

 To be fair, that is just BRILLIANT! Good work.

   ~Steve~

Post a reply to this message

"def.gsus-" <defgsus <bei> web <punkt> de> wrote:
> it looks pretty neat when the visible sides of the cubes change to show the
> different logos but the morphing is too fast for my taste. one can not
> actually see anything but only *guess* that it's morphing.

Fair enough. I probably spent too long looking at it and forgot how someone
would see it the first time they watched it.

Here's an earlier test I did (without music) that shows off the effect much
more clearly:

http://www.guidetomp3players.com/testpartial3.02.avi

Post a reply to this message

> Here's an earlier test I did (without music) that shows off the effect much
> more clearly
>
i could hear some music, though :) but some people maybe wouldnt consider
that as music...

the swapping logos look even more brilliant in this vid--really cool effect,
but still the morph doesnt work for me. maybe because it's no certain shape,
as far as i can determine--just (haha) exploding blocks--and one can see the
overlapping blocks now. thats a hard one for pure povray to overcome. i
stumbled over it a few times.

ofcourse you could check for every edge/corner of one block if it's inside
another block, having about a billion statements for every frame and it
wouldnt still be clear what to do IF they overlap..mhh..

did you want to get an explosion of the (logo showing) blocks? in that case
you wouldnt need a morph but just translate them with some function.

however, it's very inspiring

Post a reply to this message

ohh, just watched your torus.avi--missed it the last time.
this one looks really cool, and not just because of the use of area lights
;)

so i think it might have to do with the shape--if the shapes are easyly
recognizeable, the whole thing looks more convincing (for me). even in the
state between torus and sphere, all points are arranged sphere-like, which
as all povers know is some 'magical' solid primitive.

Post a reply to this message