Remapping question

3 messages Options
Embed this post
Permalink
robertsladeuk

Remapping question

Reply Threaded More More options
Print post
Permalink
I have a Pano2VR project with input of six cube faces of 2000x2000
pixels each.

I want to create a transformation to an angular map "little planet"
(circular).

How do I calculate the maximum resolution possible given the size of the
cube faces?

Many thanks,
Robert

--
Robert Slade, Photographer
Manor Photography 07890 564889
http://www.manor-photography.com

Erik Krause

Re: Remapping question

Reply Threaded More More options
Print post
Permalink
Robert Slade wrote:

> I have a Pano2VR project with input of six cube faces of 2000x2000
> pixels each.
>
> I want to create a transformation to an angular map "little planet"
> (circular).
>
> How do I calculate the maximum resolution possible given the size of the
> cube faces?

Just an estimate: 2000px cube face width has the same resolution as
6282px equirectangular width. You can do "little planet" with a 360°
fisheye mapping which would be the same width as the equirect. It very
much looks as if Pano2VR uses this mapping.

However, the better mapping for "little planet" is stereographic, which
stretches the outer regions. PTGui f.e. suggest 300° for "little planet"
with about 14000px for a 6000px wide input equirect...

Perhaps you find some more info following the links on
http://wiki.panotools.org/Stereographic_Projection.

best regards
--
Erik Krause
http://www.erik-krause.de
Erik Krause

Re: Remapping question

Reply Threaded More More options
Print post
Permalink
Erik Krause wrote:

> However, the better mapping for "little planet" is stereographic, which
> stretches the outer regions. PTGui f.e. suggest 300° for "little planet"
> with about 14000px for a 6000px wide input equirect...

The formula is quite simple:

2 * equirect width * tan(FoV/4) / pi

The result is the stereographic image width and "FoV" is the angle of
the desired field of view of the stereographic image.

However, Pano2VR most likely doesn't use this mapping but the fisheye
one, where the output image width is the same as the equirect image width.

best regards
--
Erik Krause
http://www.erik-krause.de