"tcgetattr" <tg### [at] getattrinvalid> wrote in message 
news:4aeaf65b$2@news.povray.org...
> Chris B wrote:
>>
>>>> I've got a bunch of satellit pictures like this one
>>>> http://foo.buvette.org/trucs/nuages.jpeg and I want
>>>> use them to make an animation of the earth.
>>>
>>> I think that i need this sort of pictures :
>>> http://earthobservatory.nasa.gov/Features/BlueMarble/BlueMarble_monthlies.php
>> The cloud image you have only covers a small part of the planet whereas 
>> the BlueMarble images cover the whole planet, so, unless you have images 
>> from other multiple geostationary positions (or unless it's not
>
>   I'm a poor man, I only have one geostat satellite ;)

Ah! So you're Thierry rather than the girl in the gif :o(
Well, to state the obvious, you can only get a partial earth coverage with 
the images you've got.

>
>> Mapping the cloud cover image precisely to the projection of the 
>> BlueMarble images is likely to be tricky (if you need to do that).
>
>   You hit the real point here... I'm seeking for that mapping.
>
>   http://foo.buvette.org/trucs/geostatpics.tar.gz (6 Mo)

Ok. Well here's my idea:

At a geostationary distance (37,786km according to Wikipedia) the image 
should be fairly close to a linear projection, so you should be able to use 
the spherical camera trick to get something fairly close to what you want. 
Here's an example:

//camera {spherical location 0 look_at z}
camera {location <-1,2,-2> look_at 0}

sphere {0,1
  pigment {
    image_map {jpeg "nuages.jpeg" once}
    translate -y
    scale <1.61,0.7,1>
    translate -x+y
  }
  finish {ambient 1}
}

The second camera definition may be handy to understand quite what's going 
on. The image is mapped straight through the sphere like the letters passing 
through a stick of rock at the seaside (hope that analogy works for you). 
You can think of it as the light passing back from the satellite in almost 
parallel rays and staining the sphere. The spherical camera looking out from 
the centre of the sphere towards the 'far' side of the sphere draws what it 
sees around the 360 degree surface of the inside of the sphere.

The only bit that's valid for you is the middle half of the top half of the 
image, but I think this should be quite close to the projection used for the 
BlueMarble images. It won't be exact as the light entering the satellite 
doesn't really arrive in parallel rays, but at that sort of distance the 
distortion, although probably noticable, should be fairly small. You may be 
able to improve on the scaling of the original image which I got to by trial 
and error. It would obviously be much easier if the starting image was 
symmetrical and covered a full half of the globe.

Regards,
Chris B.

Post a reply to this message