Hello guys!

As you know, Directx has a feature called 'Instancing', which is quite wonderful actually. But how to use it properly is an entirely different chapter. And that's why I'm here.

Imagine Crysis or some other game where high level of vegetation is needed, as grass, and you need to render that. What I would do, and what I actually do:

When creating the scene:

When Creating the scene:

• Add all instances of grass
• Split all instances into different patches
• Create Instance Buffers

When Rendering the scene:

• Do some range testing and frustum culling for all patches, and all valid are added to a list.
• Render each individual element in the list

Now that method works fine for me when using static non destroyable or creatable objects, but imagine if one object has to be created? Do I recalculate the respective buffer for the bullets location.

Also in the Crysis Sandbox, or CryEngine 3 SDK, when placing/painting vegetation over the terrain, they appear lightning fast! If you have time, how does produceal vegetation work in a nutshell?

But what if we are simulating a space ship, IN SPACE! Where the space is almost unlimited, and then the player shoots, I can't have buffers for an unlimited space! (I could create buffers dynamically for where I am, but I'm not sure if that's efficient).

So, now, I ask you, what, if it exists, is the most efficient way of handeling instances of an object?

Thank You, again...

With the bullets, i mean this:

(Just an example)

Imagine thousands of spaceships, shooting lasers (or something) all the time, I would imagine that I would instance all bullets, wouldn't I? In that case I don't really see how I would split up each bullet up into a chunk, as they could go anywhere...

But Thanks!

In general you take a single geometry object (i.e. a vertex buffer and possibly an index buffer) and replicate it as needed. Where and how many to replicate is determined by the instance buffer. Here you will put the world matrix (for example) of each bullet in the world. In your shader you will multiply your model's vertices with the particular instance's world matrix to determine where in the world that particular instance is. It's like you were using the geometry you wanted to instance as a stamp, then the instance buffer would contain the locations where you need to stamp :) (and other properties, like the colour for example)

If the number of instances changes, one approach is to recompute all of it and re-bind it. If your geometry is mostly static, then another approach is to a create a "large enough" buffer. You can bind it even though it could not be full. When the need arises, you simply add more entries to the buffer, without having to recreate the other ones. This also sets some sort of limit for your rendering engine as it's like saying that you can render up to X instances.

In the specific case of spaceships though, I think you'd need to update them each frame. BUT! If you go for the hard science approach, then you don't need to worry about any of this at all, beam weapons are supposed to travel at the speed of light so .. problem solved!

note that if your vegetation doesn't change often, you can use static VB's and IB's. if it changes every frame or you recalculate every frame, then you'll want to go dynamic. so you can probably use static for vegetation, and dynamic for bullets. you'll have a VB and IB for your bullet or grass clump. then you'll have an instance buffer of world transforms for each instance you want to draw. the whole idea is to reduce the number of times you bind the instance buffer to a minimum. think of it this way: binding is bad. different binds have different costs: textures are highest (?). then VB's, then IB's, then constant buffers. apparently the hit for constant buffers isn't that bad. can't recall if binding dynamic vs static buffers is more costly (i think so). all this info is available in the docs.