Jump to content

  • Log In with Google      Sign In   
  • Create Account

Vertex Shader Waves - Recalculating Smooth Normals


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
1 reply to this topic

#1 OMH   Members   -  Reputation: 155

Like
1Likes
Like

Posted 16 February 2014 - 05:14 AM

Hello everybody,

 

currently I'm working on a ocean shader with some vertex shader math to simulate the waves. For a more convincing effect I'd like to recalculate the surface normals after the vertex displacement has happenend, which is where I'm struggling at the moment. It's worth mentioning what I'm working with the UDK, which means there are several restrictions for the material editor that I won't be able to bypass.

 

 

So far I've come across 2 different approaches.

 

One is to use the derivative of the wave height function for recalculating the normals. Whilst this is looking pretty good it's adding lots of pixel shader instructions, depending on the wave simultion complexity. For instance:

http://www.antongerdelan.net/opengl/vertex_displacement.html

http://software.intel.com/en-us/articles/real-time-deep-ocean-simulation-on-multi-threaded-architectures

 

The other one uses partial derivates of the WorldPosition. This is has a fixed impact on instruction count, which is nice, but as expected the resulting normals are flat shaded.

 

 

Since an ocean wave algorithm might get pretty complex I'd rather go with approach #2. So I'm wondering if there are any techniques to calculate smooth normals in the pixel shader with a fixed instruction count?


Edited by OMH, 21 February 2014 - 11:21 AM.


Sponsor:

#2 alkisbkn   Members   -  Reputation: 440

Like
0Likes
Like

Posted 17 February 2014 - 03:47 AM

The normal can be calculated as the first derivative of the wave equation. I don't have code at the moment as I am at work, however I can post something for you later on.

In the meantime, you can do some research on calculating the first derivative, it shouldn't be too hard :)






Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS