Hi guys -

A curious one for you here, I'm running into problems using bump maps when
I'm using a uv_mapping.

The problem is that the bump-map normals seem to be dependent upon the
direction of the u/v vectors - for example, a triangle with UV coordinates
<1,0> <0,0> <0,1> will give different results than a triangle with UV
coordinates <-1,0> <0,0> <0,1> even though the bump map is symmetrical in
the U and V dimensions, and the triangle vertices are exactly the same and
in the
same order.

It almost seems like the way that the bump map affects the surface normal is
dependent upon the U and V 'direction' of the bump map.  Note that the
physical triangle
normal is not affected by the UV coordinates - just the normal that is
generated
by the bump-map.

See the attached POV file - note that in the U-axis (along the wall) the
bump-map behaves one way, and in the
V-axis (up and down) it behaves in another.  Set the declaration of
"u_factor" to 1 to see the desired result where
the bump-map behaves appropriately.

Is this a bug?  I would expect U/V coordinates to simply define source
points on a bump-map or
texture.  Why does it affect the normal?

John Pallett

Post a reply to this message

Attachments:
Download 'bump_inversion.pov.txt' (2 KB) Download 'bump_pyramids.jpg' (3 KB)

Preview of image 'bump_pyramids.jpg'

Sorry - please change the u_factor declaration to -1 to see the bug.  I
accidentally sent out the file where things were set to behave correctly.

I'm attaching a copy of the POV file I originally meant to send.

JP

"John Pallett" <joh### [at] chiefarchitectcom> wrote in message
news:3d1cd210@news.povray.org...
> Hi guys -
>
> A curious one for you here, I'm running into problems using bump maps when
> I'm using a uv_mapping.
>
> The problem is that the bump-map normals seem to be dependent upon the
> direction of the u/v vectors - for example, a triangle with UV coordinates
> <1,0> <0,0> <0,1> will give different results than a triangle with UV
> coordinates <-1,0> <0,0> <0,1> even though the bump map is symmetrical in
> the U and V dimensions, and the triangle vertices are exactly the same and
> in the
> same order.
>
> It almost seems like the way that the bump map affects the surface normal
is
> dependent upon the U and V 'direction' of the bump map.  Note that the
> physical triangle
> normal is not affected by the UV coordinates - just the normal that is
> generated
> by the bump-map.
>
> See the attached POV file - note that in the U-axis (along the wall) the
> bump-map behaves one way, and in the
> V-axis (up and down) it behaves in another.  Set the declaration of
> "u_factor" to 1 to see the desired result where
> the bump-map behaves appropriately.
>
> Is this a bug?  I would expect U/V coordinates to simply define source
> points on a bump-map or
> texture.  Why does it affect the normal?
>
> John Pallett
>
>
>

Post a reply to this message

Attachments:
Download 'bump_inversion.pov.txt' (2 KB) Download 'bump_pyramids.jpg' (3 KB)

Preview of image 'bump_pyramids.jpg'

-Hail

  There seems to be something odd here indeed, I dont think the direction of
the image map should affect the outcome value, Pentium 233, W2k




            Vampyrium

Post a reply to this message