If I want to blend a real photo (from scanner) with a pov image it is
difficult to adjust the camera position so that 3D viewpoints and angles
match exactly.

Are there any tricks for doing this quicly?

For example, if I make (and scan) a photograph from my garden, and then
want to position a virtual piece of art in it, can I relate pov-camera
settings to the lenses in my photocamera?

Post a reply to this message

You might be able to pull this off with the code at
http://www.spiritone.com/~english/cyclopedia/background.html where I talk
about using an image as a background. If you want to have foreground and
background information from the same photograph, make a copy of the image
and using an alpha channel, make the background transparent. Make a second
box with the method I described with a smaller distance setting, so the
foreground is closer to the camera than the background.

( I used this code in my animation with a close distance and then pumped up
the filter value of the image so it fades to show the action. This happens
far too quickly in my animation)

Josh

"A. Jans-Beken" wrote:

> If I want to blend a real photo (from scanner) with a pov image it is
> difficult to adjust the camera position so that 3D viewpoints and angles
> match exactly.
>
> Are there any tricks for doing this quicly?
>
> For example, if I make (and scan) a photograph from my garden, and then
> want to position a virtual piece of art in it, can I relate pov-camera
> settings to the lenses in my photocamera?

--
Josh English -- Lexiphanic Lethomaniac
eng### [at] spiritonecom
The POV-Ray Cyclopedia http://www.spiritone.com/~english/cyclopedia/

Post a reply to this message

I would assume that if you took measurements from the real-life subject of
your picture and then composed a render using measurements from the
real-life world to create the objects you wished to add to your picture, you
could have things line up very well throughout.  You then need to place your
imagemapped photo in the right spot in the rendered world.

On another note:
I'm working on a render where I'm creating a space using real
measurements... a perspective so to speak, however, to get the image objects
to look like the photograph, I had to double the width between the objects.
Height-wise and depth-wise, everything else matches up.  I've not found any
solution for this phenomenae yet.  The question to POV'ers is is there a
phenomenae when photographing that would make objects close to the
photographer appear wider than objects further away?  In my render, I have
two parralel walls that when placed the correct distance appart do not match
the same two objects in a photograph of a real space, both in angle as well
as breath in the image.


A. Jans-Beken <jan### [at] wxsnl> wrote in message
news:3A6### [at] wxsnl...
> If I want to blend a real photo (from scanner) with a pov image it is
> difficult to adjust the camera position so that 3D viewpoints and angles
> match exactly.
>
> Are there any tricks for doing this quicly?
>
> For example, if I make (and scan) a photograph from my garden, and then
> want to position a virtual piece of art in it, can I relate pov-camera
> settings to the lenses in my photocamera?
>

Post a reply to this message

in <3a665426@news.povray.org> Robert J Becraft wrote:

>In my render, I have
>two parralel walls that when placed the correct distance appart do not
>match the same two objects in a photograph of a real space, both in
>angle as well as breath in the image.

One difference with POV an photography is that in POV the angle is the 
horizontal angle, where in photography the angle of a lens is the angle 
measured over the diagonal of the image.


Ingo

-- 
Photography: http://members.home.nl/ingoogni/
Pov-Ray    : http://members.home.nl/seed7/

Post a reply to this message

Basically you need 5 reference points to match the CG and real
coordinates. Of course the FOV of camera must be known also.
A Google search with "robotics vision photogrammetry reference points"
should produce some relevant matches.



______________________________________________________________________
Kari Kivisalo                                  http://www.kivisalo.net

Post a reply to this message

The reason that close objects don't match CG objects is the physical
dimension of the camera lens which is not a pin hole like the the CG
"lens". When objects are closer than, let's say, 10*lens_diameter,
the difference becomes more apparent. There was an article in IEEE
CG&A where real camera lens geometry was modelled thus providing a
perfect match with a corresponding photograph.

______________________________________________________________________
Kari Kivisalo                                  http://www.kivisalo.net

Post a reply to this message