[Radiance-general] Re: texpict...

[Greg Ward]

> From: "Lars O. Grobe" <grobe@gmx.net>
> I know that there has been a thread about bumpmaps not long ago. But I 
> found
> that Simon Crone has implemented a  texpict modifier for an old version 
> of
> radiance. Is there anybody out there  working with this and current 
> radiance
> releases? And why has this not been added  to the official radiance code
> branch?
>
> I need some kind of bump-mapping, as I have a large model of a church 
> with
> some more detailed elements. There are reliefs on parts of the building 
> which
> exist in  their geometry, but I don't want to model the relief ;-) BTW, 
> this
> would make the model too large, too.
>
> So I will have to use colorpict for now... but I don't really like 
> this... We
> have all those nice bumpmap features in our modeling software, and now 
> I have
> to tell the others that I work with mapped images... ;-)

What's wrong with the "texdata" primitive?  You'll have to convert your 
bump maps to ASCII data files.  The in-memory size is the same as if the 
input were from a picture.  The reason I didn't use pictures to describe 
the surface normal perturbation map is because they don't support 
negative values.  What form is your input?  If they're height fields, 
then you'll have to derive surface normal perturbations from these.  I 
never liked or supported the use of height fields for bump maps, because 
there is no definitive mapping from a height field to a surface normal 
map, which is what you really want for this type of texture.

>
> BTW, any hints regarding black/white-images as colorpict-maps?

Just use the colorpict map with "grey grey grey" as the modifier 
functions.  If you don't like using 3x as much memory as single-channel 
data, then you'll have to convert your data to ASCII again and use the 
brightdata primitive.

-Greg