Texture to normal map?
I was wondering if there was such a program online free if possible. That will allow you to input a normal image say an image of a wall texture, and generates a normal map for you. I just was wondering so I could perhaps work with bump mapping things. Is there such a tool could someone please link me?
I don't know the exact name/links, but I'm sure both the ATI and nVidia SDK's have normal map tools. I think theres a Photoshop plugin too.
I'll have a look around.
Edit: Found this: nVidia Texture Tools and ATI Tools
I'll have a look around.
Edit: Found this: nVidia Texture Tools and ATI Tools
Doing a simple greyscale conversion can generate a simple height-map which you can then use to create a normal map (the aforementioned NV tools are good for this).
However, going from a colour->normal map is in itself not going to be accurate. It can generate okay-ish/passable results, but at the same time they can be very easy to spot as being blatantly wrong.
Example:
Photo of a brick wall, convert to greyscale makes the shadows black and the highlighted areas white. Thus you can end up with "lop sided" bricks where 2 edges are "high" and the opposite 2 edges are "low" - rather than the whole brick face being "high" and the surrounding morter/cement being "low". You inherit the original photo/texture lighting into your data set - whereas your runtime effect is actually re-generating its own lighting...
I'd imagine there are more complex algorithms that can do a better job than a trivial greyscale conversion, but the image analysis that you can do on a simple colour photo/image is fundamentally limited ...
hth
Jack
However, going from a colour->normal map is in itself not going to be accurate. It can generate okay-ish/passable results, but at the same time they can be very easy to spot as being blatantly wrong.
Example:
Photo of a brick wall, convert to greyscale makes the shadows black and the highlighted areas white. Thus you can end up with "lop sided" bricks where 2 edges are "high" and the opposite 2 edges are "low" - rather than the whole brick face being "high" and the surrounding morter/cement being "low". You inherit the original photo/texture lighting into your data set - whereas your runtime effect is actually re-generating its own lighting...
I'd imagine there are more complex algorithms that can do a better job than a trivial greyscale conversion, but the image analysis that you can do on a simple colour photo/image is fundamentally limited ...
hth
Jack
I find that using the "high-pass" filter in photoshop can sometimes even out the problem areas, and then using the "paint daubs" filter a bit... this chnages the map a bit but creates more a hieghtmap.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement