[LOD and you!]

Started by
3 comments, last by aDasTRa 21 years, 11 months ago
I have been thinking about a kinda cheap trick to have models with various levels of detail, and am just looking for some input. The idea is actually two or three, all kinda connected... I was thinking of using a well established format such as .md2 as the basis of this system. I am not too concerned with dynamic LOD either, and the number of animations for each model will be very small (walking and simple gesturing basically). So my first impression was to use a sort of 2-d array setup, where one dimension represents the current frame of model animation, and the other dimension being the LOD. Then i would have one model for every LOD, all with the same animations in the same order. Then i would just use some LOD indicator to index the right model, and then the frame number to draw the right model. That was my first idea...I know it has the drawback of using a fair bit of memory, but that is acceptable for this project. The idea has evolved slightly though, but i am not sure if it is still ''legal'', mostly because i am not too familiar with the .md2 format. What I was thinking was, since the number of different animations will be small, is i could simple put all of the LOD models into one file. For example, lets assume a .md2 file supports 100 frames of animation (i have no idea if this is the case, but let us assume). frames 0-19 will be a walking animation at the highest LOD, 20-39 the exact same animation with a lower poly count model, 40-59 lower still, etc... This would have the benefit of using less memory and might be a little easier to work with too...but i dont know if it''ll work... Any input or suggestions would be appreciated, as well as any good references on the .md2 (or more applicable) format. thanks <(o)>
<(o)>
Advertisement
A good LOD algorithm uses the same animations of the highest resolution model for the other LOD levels (without the artist having to supply them).

Try a Google search on continuous LOD.




I rest my case. Meaning; I lost my luggage.
-------homepage - email
that is true, countach, but i am looking for a simpler system than some kind of continous lod. i want something fast and uncomplicated.

but ill still give google a shot...

any one else?


<(o)>
<(o)>
Go with skeletal animation, and use the same skeleton across all your LOD models. Then LOD changes won't affect your animation at all.

[edited by - cheesegrater on May 10, 2002 2:08:18 PM]
If you look into the .pk3 of Quake3 you''ll notice that most models come in three different versions model.md3 model_1.md3 model_2.md3. Each on is a different LOD level. They contains exacly the same animation so LOD is jsut a matter of switching from one model to the other. You just have to becarefull to switch your model at the right time to avoid popping effects.

Emmanuel

This topic is closed to new replies.

Advertisement