Still in love with the elegance of user_defined cameras. Here's the code for a
"inverse panorama", also known as an "object panorama".

// Object panorama camera
// (x,y) range from -0.5 to 0.5 each
#declare ZMIN = -0.45;            // Vertical range of camera
#declare ZMAX = 0.45;             // Determined by actual height range of the
object
#declare RANGE = 10;              // Distance of camera, wide FOV to
orthographic
#declare PANANGLE = radians(90);  // Pan panorama
camera {
   user_defined
   location {
      function { RANGE * cos(x*pi*2 + PANANGLE) }
      function { RANGE * sin(x*pi*2 + PANANGLE) }
      function { ZMIN + (y+0.5) * (ZMAX-ZMIN) }
   }
   direction {
      function { -cos(x*pi*2 + PANANGLE) }
      function { -sin(x*pi*2 + PANANGLE) }
      function { 0 }
   }
}

Instead of a slit camera at a single position rotating about a vertical axis,
one moves a slit camera on a circle around an object where the camera is looking
towards the center of the object. A bit weird, like all multi-point projections.
Some examples attached of some Chinese puppets.

Post a reply to this message

Attachments:
Download 'example.jpg' (3316 KB)

Preview of image 'example.jpg'

On 27/12/2025 12:04, Paul Bourke wrote:
> Still in love with the elegance of user_defined cameras. Here's the code for a
> "inverse panorama", also known as an "object panorama".
> ...
It looks interesting and unusual, how large are models of puppets ?
-- 
YB

Post a reply to this message

yesbird wrote:
> On 27/12/2025 12:04, Paul Bourke wrote:
> > Still in love with the elegance of user_defined cameras. Here's the code for a
> > "inverse panorama", also known as an "object panorama".
> > ...
> It looks interesting and unusual, how large are models of puppets ?
> --

Slit camera's are nice, one can do strange things with them. I the past I
simulated one in POV-Ray by using a 1-3 pixel wide image and animate that around
a subject. Then stitch the images together.

for some inspiration:
https://www.flong.com/archive/texts/lists/slit_scan/index.html

Slit camera's are (also) used for photo finishes.

A fist attempt withe the user defined camera:
https://news.povray.org/povray.binaries.images/thread/%3CXnsA9DA8E84160EEseed7@news.povray.org%3E/?ttop=444438&toff=700


ingo

Post a reply to this message

"ingo" <nomail@nomail> wrote:
> In the past I simulated one in POV-Ray by using a 1-3 pixel wide image and
> animate that around a subject. Then stitch the images together.

found it at
https://news.povray.org/povray.binaries.images/thread/%3C389290B0.58D1D188%40email.com%3E/

ingo

Post a reply to this message

"ingo" <nomail@nomail> wrote:
> In the past I simulated one in POV-Ray by using a 1-3 pixel wide image and
> animate that around a subject. Then stitch the images together.

found it at
https://news.povray.org/povray.binaries.images/thread/%3C389290B0.58D1D188%40email.com%3E/

ingo

Post a reply to this message

yesbird wrote:
> On 27/12/2025 12:04, Paul Bourke wrote:
> > Still in love with the elegance of user_defined cameras. Here's the code for a
> > "inverse panorama", also known as an "object panorama".
> > ...
> It looks interesting and unusual, how large are models of puppets ?

Physical size or data size?

Physically between 0.5m to 0.8m.

For this process I render them at 6 million triangles, but that is mostly for
the rigging and editing processes. Then can be safely reduced to about 1 million
triangles. Currently a single 64Kx64K texture map.

Textured
   https://vimeo.com/1150255426
And just geometry
   https://vimeo.com/1150255327

Post a reply to this message

"ingo" <nomail@nomail> wrote:
> "ingo" <nomail@nomail> wrote:
> > In the past I simulated one in POV-Ray by using a 1-3 pixel wide image and
> > animate that around a subject. Then stitch the images together.
>
> found it at
>
https://news.povray.org/povray.binaries.images/thread/%3C389290B0.58D1D188%40email.com%3E/
>
> ingo

That's also how I did it in the past.

Post a reply to this message

On 01/01/2026 13:08, Paul Bourke wrote:
> Physically between 0.5m to 0.8m.
> 
> For this process I render them at 6 million triangles, but that is mostly for
> the rigging and editing processes. Then can be safely reduced to about 1 million
> triangles. Currently a single 64Kx64K texture map.
> 
> Textured
>     https://vimeo.com/1150255426
> And just geometry
>     https://vimeo.com/1150255327
> 

Thanks so much for this info. Now I am trying to optimize 'Metashape'
process, taking into account 6048 x 4032 resolution of my new camera.

I would like to find a balance between model quality and users
devices performance to extend potential community. Few experiments
show that texture size is a serious limit, so I am producing few
model's versions with 4K, 8K, (16k in future).

Now looking for optimal number of triangles ...
-- 
YB

Post a reply to this message