"Thomas de Groot" <tDOTdegroot@interDOTnlANOTHERDOTnet> wrote:
> "clipka" <nomail@nomail> schreef in bericht
> news:web.49d3a106f4dc8d89f708085d0@news.povray.org...
>
> As an aside, I don't like the fact that *they* ask for personal details
> outside a secure environment. I am rather wary of that kind of things.
>
> Thomas

I have no problem giving them personal information.  It just happens to be made
up, but it satisfies their curiosity.  I use HDRShop for all my HDR map
purposes: viewing, converting map types and creating my own light probes.  It
is very easy to use and is a low impact install (in fact it doesn't need to
install and write to the registry and such).



As an aside (or perhaps back on the original topic).  I prefer to use strictly
lat/long type maps.  Easier to look at unmapped.  Also the similarity between
mirror ball and angular maps can cause confusion as pointed out already.

Mirror balls are just that, an image of a mirrored ball, as you go outwards, it
reflects according to the angle of incidence, but this causes the data to get
squished as you get to the edges and can end up looking odd.  Mirrorball maps
are very difficult to utilize in POV, they basically need to be converted to a
usuable format.

Angular maps (sometimes called light probe) represent a linear interpolation of
the environment relative to the angle from the viewing vector, which more
evenly distributes the data. (e.g. center is 0 degrees from viewing angle, 1/4
distance from the center is 90deg, 1/2=180deg, 3/4=270deg, edge=360deg).  This
is much easier to map than mirrorball, but may require some thought on how to
map it porperly (megaPOV actually supports a map_type 7 which is this type).

Cubic environment or vertical cross is basically 6 parts that get mapped to the
6 sides of a cube.  With a little work, these could be mapped to a sphere as
well.  This type shows the least distortion in its unmapped format, but
basically requires each of the six parts to be mapped individually rather than
with a single function.

Lat/Long are basically a standard square to sphere mapping, x axis is longitude
(0 to 360 deg), y axis is latitude (-90 to 90 deg).  The image gets stretched
as you progress towards the top or bottom, but this is not a problem as it
essentially includes more resolution than necessary and typically doesn't
affect the quality, rather than squishing and getting less.  I find these are
the easiest to map and manipulate, and also the easiest to understand when
looking at the unmapped version.  My HDR lightdome macro uses strictly this
type for this reason (also it is much easier to run through the various
distribution algorithms within the macro in this format)

-tgq

Post a reply to this message