As a rule, should texturemaps have high dynamic ranges?
When I've mapped pics to geomtery, I just take my source tif and ra_tiff -r
to a pic.
Since it started as an 8 bits/channel tiff, the pic would not have any HDR
info.
Is that the best way to create/use maps?
I would think taking bracketed photos and creating a HDR texturemap would
include more ambient characteristics then the surface itself, and thus not
be viable for mapping.
Just curious
Rob F
Hi Rob,
You raise an interesting question. Since a modifier generally affects the material reflectance or transmittance, values in a pattern above 1/reflectance (or 1/transmittance) are not physically valid. However, they may be useful when representing an object that includes luminous elements, such as a panel display, or one that contains diffuse and specular materials, which you are not modeling separately. In other words, the surface may appear to "create light" either due to specular reflections, or due to actual emitting elements. The stickler in me says you need to model these as separate materials, but as a practical matter, it is usually easier to "cheat" and do it with an HDR pattern. The results won't be exactly right, but they'll be better than truncating the highlights and pasting the pattern on a diffuse surface.
As for the lower end of the dynamic range, real objects don't get much below 1% reflectance, so you don't really need HDR to represent the deep blacks. The standard image range will do. Of course, you must convert this as you say to HDR format for Radiance, but doing so with ra_tiff -r from a 24-bit RGB file should work fine.
True HDR patterns are perfectly appropriate for modifying glow sources, such as the sky. Jessi Stumpfel et al. have a paper on HDR capture of the sun annd sky, which will be presented at the Afrigraph conference this November. For more info, check Paul Debevec's website:
http://www.debevec.org/Publications/
I hope this is of use.
-Greg
From: "Fitzsimmons, Rob" <Rob.Fitzsimmons@Summit.Fiserv.com>
Date: July 22, 2004 12:25:12 PM PDTAs a rule, should texturemaps have high dynamic ranges?
When I've mapped pics to geomtery, I just take my source tif and ra_tiff -r
to a pic.
Since it started as an 8 bits/channel tiff, the pic would not have any HDR
info.
Is that the best way to create/use maps?
I would think taking bracketed photos and creating a HDR texturemap would
include more ambient characteristics then the surface itself, and thus not
be viable for mapping.Just curious
Rob F