POV-Ray : Newsgroups : povray.beta-test.binaries : Requested fisheye behavior Server Time
16 May 2024 03:45:08 EDT (-0400)
  Requested fisheye behavior (Message 1 to 7 of 7)  
From: Rune
Subject: Requested fisheye behavior
Date: 16 Sep 2001 10:05:11
Message: <3ba4b197@news.povray.org>
The attached image shows how I think the fisheye camera should work. Right
now it completely ignores the up and right vectors in the camera, but I
think the up and right vector should be used to "stretch" the image, just
like for the other camera types. I've compared the requested fisheye
behavior to that of the perspective camera.

Note that just like in the other camera types, specifying the angle should
of course adjust the horizontal angle while still maintaining the aspect
ratio. The black border should not be affected by the angle though, but only
be controlled with the up and right vectors.

Hope it's clear now.

Rune
--
3D images and anims, include files, tutorials and more:
Rune's World:    http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users:   http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk


Post a reply to this message


Attachments:
Download 'fisheye.jpg' (38 KB)

Preview of image 'fisheye.jpg'
fisheye.jpg


 

From: Nathan Kopp
Subject: Re: Requested fisheye behavior
Date: 18 Sep 2001 22:13:30
Message: <3ba7ff4a@news.povray.org>
"Rune" <run### [at] mobilixnetdk> wrote...
> The attached image shows how I think the fisheye camera should work. Right
> now it completely ignores the up and right vectors in the camera, but I
> think the up and right vector should be used to "stretch" the image, just
> like for the other camera types. I've compared the requested fisheye
> behavior to that of the perspective camera.

Does anybody have any comments about this?

An likely alternative would be to have the last two scenes:
    up y   right x*2
and
    up 0.5*y    right x
be equivilent.

That would better mimic the behavior of the perspective camera, where these
two cameras would be equivilent.  This provides better consistency across
camera types, though it does sacrifice some flexibility.

The desired effect (a zoomed-in view of the fisheye lens) could be achieved,
though, by using POV's partial-render capibilities.

Comments?

-Nathan


Post a reply to this message

From: Bob H 
Subject: Re: Requested fisheye behavior
Date: 19 Sep 2001 01:45:11
Message: <3ba830e7@news.povray.org>
"Nathan Kopp" <nat### [at] koppcom> wrote in message
news:3ba7ff4a@news.povray.org...
>
> "Rune" <run### [at] mobilixnetdk> wrote...
> > The attached image shows how I think the fisheye camera should work.
Right
> > now it completely ignores the up and right vectors in the camera, but I
> > think the up and right vector should be used to "stretch" the image,
just
> > like for the other camera types. I've compared the requested fisheye
> > behavior to that of the perspective camera.
>
> Does anybody have any comments about this?

I thought the ultra_wide_angle camera was the rectangular (or square, all
dependant on aspect used) version of the fisheye camera.  If so then no need
to change fisheye, if not then a fullview aspect changeable fisheye lens is
an okay idea far as I'm concerned.

> An likely alternative would be to have the last two scenes:
>     up y   right x*2
> and
>     up 0.5*y    right x
> be equivilent.
>
> That would better mimic the behavior of the perspective camera, where
these
> two cameras would be equivilent.  This provides better consistency across
> camera types, though it does sacrifice some flexibility.
>
> The desired effect (a zoomed-in view of the fisheye lens) could be
achieved,
> though, by using POV's partial-render capibilities.
>
> Comments?

See above, please.

Bob H.


Post a reply to this message

From: Bob H 
Subject: Re: Requested fisheye behavior (compared with ultra_wide_angle)
Date: 19 Sep 2001 02:26:13
Message: <3ba83a85@news.povray.org>
Figured I better try a comparison of the two since I hadn't done so for
years.
ultra_wide_angle camera renders very differently so I was wrong to assume
they were more similar.

// cmd: +w300 +h300

// distortions of fisheye lens compared with ultra wide angle
camera {
// fisheye // uncomment
 ultra_wide_angle // comment
  location -z*1 // any distance
  look_at 0 // try 5 here also to see interesting effect
  up y
  right x*image_width/image_height
  angle 360 //
}

#declare Checker=
pigment {checker color rgb 0 color rgb 10 scale 0.05}

#declare Hexagon=
pigment {hexagon color red 10 color green 10 color blue 10 scale 0.05 rotate
90*x}

box {-1,1
        pigment {gradient z
                pigment_map {
                [0 Checker]
                [1 Hexagon]
        } scale 2.001 translate -z}
        scale 10
}


Post a reply to this message


Attachments:
Download 'uwa_fisheye.jpg' (66 KB)

Preview of image 'uwa_fisheye.jpg'
uwa_fisheye.jpg


 

From: Rune
Subject: Re: Requested fisheye behavior
Date: 19 Sep 2001 11:06:03
Message: <3ba8b45b@news.povray.org>
"Nathan Kopp" wrote:
> An likely alternative would be to have the last two scenes:
>     up y   right x*2
> and
>     up 0.5*y    right x
> be equivilent.
>
> That would better mimic the behavior of the perspective camera,
> where these two cameras would be equivilent.

That is *not* true! Using those two variations with the perspective camera
produces different results. In fact the perspective images in the comparison
image I made were rendered doing just that and as you can see, they differ.

The reason of your misconception probably is that they will produce
equivalent results if the angle keyword is used after the up and right
vectors are specified. When the angle is specified the lengths of the up and
right vectors are adjusted internally to match the angle. But it is not
required to specify the angle, and if it isn't specified, the lengths of the
up and right vectors do make a difference!

I already mentioned the angle keyword in my original post:

"Note that just like in the other camera types, specifying the angle should
of course adjust the horizontal angle while still maintaining the aspect
ratio. The black border should not be affected by the angle though, but only
be controlled with the up and right vectors."

> This provides better consistency across camera types

On the contrary.

> though it does sacrifice some flexibility.

I still think my suggestion is the best one. It's consistent *and* flexible.

> The desired effect (a zoomed-in view of the fisheye lens) could
> be achieved, though, by using POV's partial-render capibilities.

An extremely clumsy solution, especially for animations.

I would appreciate if you would do reread my original message and then tell
me what's wrong with it.

Rune
--
3D images and anims, include files, tutorials and more:
Rune's World:    http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users:   http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk


Post a reply to this message

From: Rune
Subject: Re: Requested fisheye behavior
Date: 19 Sep 2001 11:12:03
Message: <3ba8b5c3@news.povray.org>
"Nathan Kopp" wrote:
> Comments?

I don't know if you're aware of it already, but Thorsten Froehlich replied
to it long ago in povray.beta-test:

> Thanks.  I agree with you that the shown output is the way it should work.
> However, I am not sure this will/can be changed in 3.5.  We will see...

Rune
--
3D images and anims, include files, tutorials and more:
Rune's World:    http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users:   http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk


Post a reply to this message

From: Nathan Kopp
Subject: Re: Requested fisheye behavior
Date: 19 Sep 2001 19:42:49
Message: <3ba92d79$1@news.povray.org>
"Rune" <run### [at] mobilixnetdk> wrote...
> I would appreciate if you would do reread my original message and then
tell
> me what's wrong with it.

Sorry man.  It was getting late and things were starting to get fuzzy.....

(You are correct that the "angle" keyword was the last line in the camera
and it was resizing the up and right vectors without me realizing it.)

-Nathan


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.