Hi,

I've been rendering some rolling spheres across the screen, but I'm noticing
that the spheres start to warp at the edges of my view. I've been playing with
the right function and other camera parameters, but I can't seem to get it
right.

This artifact appears almost like a stretching so the balls on the outside are
no longer perfectly spherical.

Has anyone seen this issue before? I'm thinking its something simple I'm not
doing.

Here's my camera and global setup:
_________________________________________________________
global_settings {ambient_light rgb< 1.5,1.5, 1.5>
}

camera {
  location  <0, 0, -12>
  direction 1.5*z
  up y
  right  x*image_width/image_height
  look_at   <0.0,0.0,0.0>
}

Here's one of my spheres:
_________________________________________________________
 object { GolfBall
  scale 0.7
  rotate <0, 0, -clock*a*180>
  translate <-5.5, 2.9, 0>
  translate <1,0,0>
          }
________________________________________________________

Is this possibly related to the scale of my objects/spheres?

Post a reply to this message

RC wrote:
> Hi,
> 
> I've been rendering some rolling spheres across the screen, but I'm noticing
> that the spheres start to warp at the edges of my view.

It's a projection effect. To reduce it, increase the distance
to the camera while decreasing the viewing angle.

Post a reply to this message

Am 16.06.2010 00:06, schrieb RC:
> Hi,
>
> I've been rendering some rolling spheres across the screen, but I'm noticing
> that the spheres start to warp at the edges of my view. I've been playing with
> the right function and other camera parameters, but I can't seem to get it
> right.

Well, the stretching /is/ right. You can't map the view of a 3d world, 
as seen from a particular point, onto a 2d plane without distorting 
/something/.

You would see the same effects in the real world with a pinhole camera, 
which is the camera type after which POV-Ray's default camera is 
modeled. Wide-angle lenses for cameras are sometimes(?) designed to 
somewhat compensate for such effects, but at the cost of distorting the 
image in some other way, e.g. distorting straight lines (fisheye lenses 
are an extreme example of this).

That said, the ultra_wide_angle camera type might give you what you want.

To avoid any distortions at all, you'll need to move the camera as far 
away from the scene as possible and reduce the camera angle (in that 
case it's pretty irrelevant whether you choose the default camera or the 
ultra_wide_angle camera - or go for the orthographic camera right away.

Post a reply to this message

> Hi,
>
> I've been rendering some rolling spheres across the screen, but I'm noticing
> that the spheres start to warp at the edges of my view. I've been playing with
> the right function and other camera parameters, but I can't seem to get it
> right.
>
> This artifact appears almost like a stretching so the balls on the outside are
> no longer perfectly spherical.
>
> Has anyone seen this issue before? I'm thinking its something simple I'm not
> doing.
>

That's perfectly normal.
A sphere define a cone with it's apex at the camera location.
That cone intersects the view plane.
The intersection of a cone and a plane is an ellipse.

Thus, unless the sphere is at the very center of the image, giving a 
perpendicular intersection, you get a more or less exentric ellipse.



Alain

Post a reply to this message

Alain <aze### [at] qwertyorg> wrote:

> > Hi,
> >
> > I've been rendering some rolling spheres across the screen, but I'm noticing
> > that the spheres start to warp at the edges of my view. I've been playing with
> > the right function and other camera parameters, but I can't seem to get it
> > right.
> >
> > This artifact appears almost like a stretching so the balls on the outside are
> > no longer perfectly spherical.
> >
> > Has anyone seen this issue before? I'm thinking its something simple I'm not
> > doing.
> >
>
> That's perfectly normal.
> A sphere define a cone with it's apex at the camera location.
> That cone intersects the view plane.
> The intersection of a cone and a plane is an ellipse.
>
> Thus, unless the sphere is at the very center of the image, giving a
> perpendicular intersection, you get a more or less exentric ellipse.
>
>
>
> Alain

Hi Alain,

Thanks you are right. After some calculating, all I have to do is render the
image larger by a desired amount, then crop the frames. Its just soooo slow to
render now, but I'm willing to wait it out to have the images correct.

THANK YOU!

Post a reply to this message

Am 17.06.2010 18:23, schrieb RC:
> Thanks you are right. After some calculating, all I have to do is render the
> image larger by a desired amount, then crop the frames. Its just soooo slow to
> render now, but I'm willing to wait it out to have the images correct.

???

Rendering a larger section of your scene doesn't change anything about 
the stretching. What you'll need to do is move the camera backwards 
instead. While this of course gives you a larger section of your scene 
by default, too, you can compensate for this by using the "angle" 
keyword. No need to crop with an external tool - and most of all no need 
to wait for surplus parts of the screen to be rendered.

Post a reply to this message

.... evidence that if we don't need new camera syntax, we need new camera docs.
The docs don't lie to the person who wrote the code, but don't teach anything to
the newbie.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Rendering a larger section of your scene doesn't change anything about
> the stretching. What you'll need to do is move the camera backwards
> instead. While this of course gives you a larger section of your scene
> by default, too, you can compensate for this by using the "angle"
> keyword. No need to crop with an external tool - and most of all no need
> to wait for surplus parts of the screen to be rendered.

Isn't this essentially Hitchcock's famous 'Vertigo Effect' (dolly zoom)?  It
would be neat to try and model this in a POV animation...

Post a reply to this message