Can someone Explain Tessellation?

[Veltrix]

I know what tessellation is on a basic level and in terms of what it does in graphics but I am looking for anyone who can explain to me in an somewhat understandable manner behind of how it actually works in programming engines if anyone knows. I am specifically looking for any calculations that might be associate with tessellating and displacement mapping the base mesh of a model. I didn't know where to put this topic but I choose the most appropriate one I can find. I've been doing as much research as I can but I don't understand or I can't find the specific thing I am looking for. Please any information would be helpful! I am not a programmer or have any basic understanding of it so please consider that! 

 

Thanks to the LTT community! 

[Hecknologist]

Basically you open paint and draw a shape outline then try to jam as many smaller random shapes into it and if they overlap you get the eraser tool and disfigure it a bit. Before you know it you'll have a load of crap because you listened to someone who has no idea what they're talking about

Nah i'm just kidding, hopefully someone is more knowledgeable them me...

[Flojer0]

I don't know any of the math but I know that in principle what it does is procedurally add triangles to a mesh to increase the detail. Sort of like a LOD 2.0. And depending on the desired effect you can control how the mesh is modified so that you get more than a just a smoother model.

 

That's all I know on it. But if you haven't seen it yet here is a tech demo by Nvidia showing some of what you can do with tessellation.

 

http://www.youtube.com/watch?v=1c_PVtMIz-A

[joelthezombie15]

Cry engine 3 has some great tessellation.

 

http://www.youtube.com/watch?feature=player_detailpage&v=-TtgW20IEm0#t=43

[_ASSASSIN_]

makes textures look 3D instead of being flat

[legodude]

this is what it looks like, how it works in magic, I someone can explain it better than me

 

[WanderingFool]

Tessellation for DX puts two concepts together.  Actual tessellation and displacement maps....in this post I will use the term points instead of the correct term vertices, this is mainly because I have found people can visual the word point instead of vertices better.

 

Actual tessellation for models doesn't do much.  It just adds extra points on the model (Tessellation doesn't move around those points...so tessellation in itself doesn't have much use, unless you manipulate the points after).

 

Displacement maps are basically a form of a texture.  Instead of being a type of texture you can see though, the "texture" contains height information.  The problem with displacement maps though is they can only manipulate the object if there are enough points...e.g. If you have just a square...if you make up that square using 4 points (the corners) and put on a displacement map you would get a square.  If you built a square with 5 points (4 corners and one in the center) and your displacement map had a high point in the center then you would create a pyramid.

 

Both technologies have been around of a while.  Displacement maps had the downside that you needed a very high polygon count surface to begin with (which just balloons the size of models)...and even if you do create a high poly surface, it would just be better to make the model high detailed.

Tessellation had the problem of creating a high resolution polygon in memory, but there wouldn't be any manipulation of it.

 

Tessellation for DX combines both together...tessellation creates a higher polygon object (as much as your video card can handle) and the displacement would manipulate the newly added points to add better detail to an object.

 

 

So in principle to add it to a game, the developers just need to create a displacement map and tell DX that the object is meant for tessellation....anyways hope this helps, I took a few liberties with it actually works but the general concept is there.

[Veltrix]

Tessellation for DX puts two concepts together.  Actual tessellation and displacement maps....in this post I will use the term points instead of the correct term vertices, this is mainly because I have found people can visual the word point instead of vertices better.

 

Actual tessellation for models doesn't do much.  It just adds extra points on the model (Tessellation doesn't move around those points...so tessellation in itself doesn't have much use, unless you manipulate the points after).

 

Displacement maps are basically a form of a texture.  Instead of being a type of texture you can see though, the "texture" contains height information.  The problem with displacement maps though is they can only manipulate the object if there are enough points...e.g. If you have just a square...if you make up that square using 4 points (the corners) and put on a displacement map you would get a square.  If you built a square with 5 points (4 corners and one in the center) and your displacement map had a high point in the center then you would create a pyramid.

 

Both technologies have been around of a while.  Displacement maps had the downside that you needed a very high polygon count surface to begin with (which just balloons the size of models)...and even if you do create a high poly surface, it would just be better to make the model high detailed.

Tessellation had the problem of creating a high resolution polygon in memory, but there wouldn't be any manipulation of it.

 

Tessellation for DX combines both together...tessellation creates a higher polygon object (as much as your video card can handle) and the displacement would manipulate the newly added points to add better detail to an object.

 

 

So in principle to add it to a game, the developers just need to create a displacement map and tell DX that the object is meant for tessellation....anyways hope this helps, I took a few liberties with it actually works but the general concept is there.

Wow thanks you summed it up well and I understood it and thank you so much for helping me out!