Vince_B
![]() |
As I started to work with Filter Forge, the bump maps made sense, but the normal maps did not. At first, since I'm new to FF I thought it was me, but don't anymore. Assuming that they are supposed to be tangent space normal maps, those generated by FF (as part of a surface texture) are not at all like those generated from the bump map using the NVIDIA routines or others.
As a specific example, I made a texture for a twine binding, like you might use to lash a point to a wooden spear. First the bump map: ![]() Then the normal map generated by Filter Forge: ![]() Then a normal map created from the bump map using NVIDIA's routines (as used in their Texture Tools 2 and PhotoShop plugin): ![]() Finally, a normal map created from the bump map using ShadeMapPro: ![]() It seems to me that the FF generated normal map does not correspond to the bump map, but those created with NVIDIA or ShaderMapPro do. Although I can always use those programs to create my normal maps, I'd like to know if there is an issue with the FF routine. Looking at the normal maps listed with several of the filters on the FF site, it seems there is. Aren't the NVIDIA texture tool routines available free to anyone, including commercial users? Maybe I'm wrong about that. I'd be glad to hear other's experience with the FF generated normal maps. |
|
Posted: February 15, 2011 1:28 am | ||
Sphinx.
![]() |
Do you have the filter for that test?
I think it might be because filter forge calculates normal maps at higher samplerate than those other programs (since they use the rasterized bumpmap and not the true sampled approach as in FF). |
|
Posted: February 15, 2011 1:46 am | ||
Sphinx.
![]() |
I just generated a normal map from your bump map image in FF - it looks the same as the two non FF samples your provided (with surface height set ~2). So it is as I suspected because the bump map is already rasterized.
You can probably simulate the effect by putting a blur (radius 0.0000000001) right before the result height input .. |
|
Posted: February 15, 2011 7:35 am | ||
Vince_B
![]() |
Sphinx, I tried blurring the height input slightly and it worked as you suggested. So, using the default sampling, the generated normal map does not contain the information relating to the high spatial frequency contours (the individual threads in this example), but blurring the input slightly fixes it--weird.
Thank you for your time and insight into this. |
|
Posted: February 16, 2011 12:45 am | ||
Sphinx.
![]() |
If it affects the texture look in a bad way, you could render the normal from the bump map in a new simple filter like the attached.
The filter is very simple: image component -> result height. Preset is set to render normal map only and surface height is 1. bump2normal.ffxml |
|
Posted: February 17, 2011 1:23 am |
Filter Forge has a thriving, vibrant, knowledgeable user community. Feel free to join us and have fun!
33,712 Registered Users
+19 new in 30 days!
153,534 Posts
+31 new in 30 days!
15,348 Topics
+72 new in year!
68 unregistered users.