Normal Maps

G

gragland

Guest
If you were to look at a normal map at any other angle besides the intended angle wouldn't it look stretched out. Like if there was a wall with jutting rocks from it (but it was actually a normal map, i.e 1 polygon), i know that if the light source changed so would the shadows, but the normal map wouldn't change in appearance depending on the angle you are looking at it would it? I hope someone understand what im saying.
 
I get what you're saying, but I dont think anyone here cares about it.

The way I percieve it in my mind, i'm not exactly thrilled about it, but I think my vision might be flawed. I wont be satisfied until I get to be in game in a room with the 1 polygon wall normal mapped image sitting there in front of me for me to view at all angles, flush etc. But I certainly wont not like the game if it doesn't look that great for me, i'll just prefer it to be used for excellent small details rather than a humongous object i'll have glaring at me if it has imperfections.
 
No that thread doesn't answer my question. But thanks for the info.
 
The way you worded it, it sounds like you think normal maps act like sprites. They don't.

Normal maps are very effecient at tricking your eye from any angle. There have been plenty of tech demos that prove that. Don't worry about.
 
Okay, here's a simplification of my original question. If there is a normal map that appears to have a rectangle protruding from it. Would you be able to see the side of the rectangle if you looked at the normal map at an angle? If so then that means the normal map not only changes changes shadows depending on the light source, but also changes in appearance depedning on your angle to it, to appear 3d. And that would be good, but i doubt thats the case.
 
No. If you want a cube protruding directly from the side, you have to model that. However if you want some bumps, you can model them with a simplified model and add in normal mapping to make it look much more complex.
 
ahnteis is right, you won't see the sides of the cube. the so called "base texture" is what you are actually seeing, which stays the same. the normal mapping just places some dark spots onto the texture where appropriate.
 
I have seen first hand a tech demo, I may be able to find it on my hard drive if you want me too, but what I've seen is very VERY real looking, things don't jut out, but they definately have a believable depth, I couldn't believe what I was looking at was flat, even though I knew it was, trust me, normal maps are AMAZING!
 
There shouldn't be much problem.. I mean if you're looking flush down the side of the surface it may look a little hinky but otherwise it should look great.. Normal mapping really is particularly groovy.
 
Yep, and it's not used to make giant, complex 3D structures, I.E. you won't see an ant lion made out of a normal map ;) You will, however, see say, a rocky wall made out of a normal map. If you look at the downforce scoop on the ferrari illustration, it DOES indeed look 3D. On a wall, you couldn't get *behind the wall to see it's not a normal map, see what I'm getting at? At any rate it's a huge improvement over just a flat texture glued to something that looks like a rock. Suspension of disbelief my friend.
 
is this the same thing that there is/was a video of showing some Valve guy doing it. He made the 3D wall, and then made another polygon and applied what i think they called a "lightmap" to it so it looked like the original 3D wall.
 
I think in the extreme case showen in the HL2 videos with a huge wall normal maps like that will not allow you to get too close to see it. Its realy only ment for objects that are further away and you cant get close enough to them to see what you suggest will happen, and in practise, what will accutaly happen (which is the same thing ;))

So what i sugest that valve will do is use these normal maps to replicate 20000 polygon objects that appear in the distance that would slow down proformance with these 1 polygon normal maped objects. Obviously the same technique could be used for small objects that you wouldnt realy nottice and you could be very close too, like a rock on the ground.

P.S. sorry for spelling / grammer / general stupidness, ive just finished work. Please dont hurt me :bonce:
 
I doubt they would use one polygon surfaces to model 20000 surfaces. That was just a demonstration on how good the technology is. They use normal mapping to reduce poly counts but 1 poly is an extreme.

And normal maps are supposed to be used for finer details, you shouldnt for example try and make a whole house out of one.
 
Back
Top