High!


to get the ASCII matrix derived from a GeoTIFF parsed properly... but 
now a new problem shows up: when I render the scene (code once again on 
p.b.s-f), I get the warning message "Camera vectors are not 
perpendicular. Making look_at the last statement may help."

I did the latter, but nothing changed. How can it be that "camera 
vectors are not perpendicular" - as I don't use up, right and direction, 
only location, look_at and sky?

The first picture is generated with 3DEM's 3D visualization module and 
(roughly) shows how the scene should look like.

The second picture is what I got instead...

See you in Khyberspace!

Yadgar

Now playing: Turn of the Century (Yes)

Post a reply to this message

Attachments:
Download 'sarehelmand-3dem.jpg' (39 KB) Download 'sarehelmand.png' (4 KB)

Preview of image 'sarehelmand-3dem.jpg'

Preview of image 'sarehelmand.png'

>High!
>

>to get the ASCII matrix derived from a GeoTIFF parsed properly... but 
>now a new problem shows up: when I render the scene (code once again on 
>p.b.s-f), I get the warning message "Camera vectors are not 
>perpendicular. Making look_at the last statement may help."
>
>I did the latter, but nothing changed. How can it be that "camera 
>vectors are not perpendicular" - as I don't use up, right and direction, 
>only location, look_at and sky?
>
>The first picture is generated with 3DEM's 3D visualization module and 
>(roughly) shows how the scene should look like.
>
>The second picture is what I got instead...


One of your trace() rays doesn't intersect Earthslice. This leaves you
with "lo" being in the same direction as the sky vector, which produces
the error that you are seeing.

In fact, it's unlikely that your second trace() would intersect a
roughly spherical object, unless cNormVect and lNormVect are very
similar.

I suggest firstly:
  #declare la=trace(EarthSlice, lNormVect, -lNormVect);

Secondly, never assume that a trace() successfully intersects a surface,
get it to return the normal and check that it is not zero. Not just for
this scene, but as a general principle.


-- 
Mike Williams
Gentleman of Leisure

Post a reply to this message

High!

Mike Williams wrote:

 >One of your trace() rays doesn't intersect Earthslice. This leaves you
 >with "lo" being in the same direction as the sky vector, which produces
 >the error that you are seeing.
 >
 >In fact, it's unlikely that your second trace() would intersect a
 >roughly spherical object, unless cNormVect and lNormVect are very
 >similar.
 >
 >I suggest firstly:
 >#declare la=trace(EarthSlice, lNormVect, -lNormVect);
 >

Ahhh, yes... of course the "direction vector" for the look_at ray can't 
be the same as for the location ray!

I fixed that, and got a reasonable view of the landscape (attached 
here)... but then another problem showed up: narrow gaps between (most) 
triangles! Should I smooth them - or is there another way to get rid of 
that?

See you in Khyberspace!

Yadgar

Now playing: Yellowstone Park (Tangerine Dream)

Post a reply to this message

Attachments:
Download '2005-04-20_sar-e_helmand_take_3_(yadgar).jpg' (42 KB)

Preview of image '2005-04-20_sar-e_helmand_take_3_(yadgar).jpg'