Jump to content
  • Advertisement
Sign in to follow this  
StarikKalachnikov

Is LOD Worth it?

This topic is 4696 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi, If you make a level editor where you can create terrain (Warcraft3/Simcity style editing), can't you save a number of different files with different pre-calculated level of details?

Share this post


Link to post
Share on other sites
Advertisement
Of course you can :)

But i think what you are trying to say is.
When you have static terrain and a huge area is it smart to devide the terrain in smaller parts. And a part in multiple files which are different in detail but the same part of the area.

This way you can use the less detailed(smaller files) in the distance and the higher detailed files up-close.

And is that faster /same speed / slower then when generating parts with LOD algo's?
(guess that depends on the implementation but does anybody have any experience with the given approach ?)

Thanks for reading.

Share this post


Link to post
Share on other sites
Quote:
Original post by StarikKalachnikov
If you make a level editor where you can create terrain (Warcraft3/Simcity style editing), can't you save a number of different files with different pre-calculated level of details?


"Is LOD Worth it?"
Yes, and guess what? The method you described is a form of LOD.

Share this post


Link to post
Share on other sites
Quote:
Original post by haphazardlynamed
Quote:
Original post by StarikKalachnikov
If you make a level editor where you can create terrain (Warcraft3/Simcity style editing), can't you save a number of different files with different pre-calculated level of details?


"Is LOD Worth it?"
Yes, and guess what? The method you described is a form of LOD.


Wow, I didn't know about it, do you have any information about it?
What is the performance hit, since you keep loading the terrain from harddisk?
Can you give a comparison between the one described and various run-time algorithms?

Any links /suggestions would be very much appreciated!

Share this post


Link to post
Share on other sites
Depending on what kind of terrain (how detailed, how much LOD swapping occurs, etc) you may simply want to cache the different LODs in memory and swap them out at the appropriate time.

Share this post


Link to post
Share on other sites
LOD is an optimization technique. As with all optimization, it is impossible to know if an optimization is "worth it" without knowing the characteristics of where the it is being made.

The answer to your question is, it depends.

Share this post


Link to post
Share on other sites
The method you originally described is called discrete LOD. You have different models of the same object with more or less detail, and render the correct one based on the detail required during that frame.

The other method is called continuous LOD, where every frame (or every 2 or 4 frames) you decide what detail level you need and somehow generate an appropriate model to render. There are lots of ways to do this, depending on how your model is defined (i.e. irregular mesh vs regular mesh vs curved surfaces, etc).

With DLOD, you use extra memory to store your model at different detail levels, but very little processing time per frame. With CLOD you don't use a lot of extra memory, but you must calculate a new model in real-time, so your algorithm for generating your model needs to do it's work quickly.

Share this post


Link to post
Share on other sites
With a discrete LOD system if you enforce the rule that a lower LOD's vertices are a subset of next highest LOD's vertices then you can reorder the vertices by the lowest LOD they are used in (i.e. vertices used in the lowest LOD upwards come first, vertices only used in the highest LOD come last) and just create a unique index buffer for each LOD (rather than a unique VB as well). The VB for each LOD is simply a subset of the VB for the highest LOD (starting from vertex 0).

Of course this screws up any optimisation you may have done on the order of the vertices and in numerous cases enforcing the VB subset rule results in a worse quality LOD for a given number of vertices than could be achieved without this restriction. For terrain it usually works out OK though (except for the optimised ordering issue).

Just something to think about, I'm not particularly advocating doing this though,

Mike.
(If this makes no sense, I apologise! It's a bit late here.)

Share this post


Link to post
Share on other sites
Thank you all,

If there is no information even hard information is good info.
After my vacation i will look further into it all.

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!