• Advertisement

# DX11 DX11 Instancing - Only render certain instances

This topic is 1763 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 guys!

So, imagine you have a scene with 1000 trees and your rendering it with instancing, quite a lot, and you wan't to implement a lod system. Now your lod system is actually done, but now how do you change the buffers for that tree (the one too far away).

Illustration (real):

Now as you see I have quite a lot of trees there, but now I only wan't some of those trees from my instanced buffer to be rendered. As I understand you have to update the buffer, but wouldn't that be quite heavy, or do I have to split up the instanced buffer into chunks, and update those? Right below I'm gonna show you how I handle my rendering.

Rendering:

UberBuffer._instance = true;
UberShader.SendObjectBuffer(ObjectBuffer, devcon);
UberShader.SendUberBuffer(UberBuffer, devcon);
InstancedMesh[i]->SendInstancedData(ObjectBuffer, Camera, Bufferphase, devcon); // doesn't really matter for now, but if you wan't it, say it :)

bool changed = false;

for (int lod = -1; lod < InstancedMesh[i]->Lods.GetCount(); lod++) // we start from -1 as some meshes don't have lods, so we actually trick it!
{
if (changed)
{
InstancedMesh[i]->SendBuffers(devcon, dev, Camera, ObjectBuffer, Bufferphase);
changed = false;
}

if (lod > -1) // is this a lod?
{
for (int instance = 0; instance < InstancedMesh[i]->instanceCount; instance++)
{
if (sqrtf(D3DXVec3Dot(&*Camera->GetPostion(), &InstancedMesh[i]->instances[instance].pos)) >= InstancedMesh[i]->Lods.GetComputedADistance(lod)) // distance...
{
InstancedMesh[i]->Lods.GetComputedLod(lod)->SendVIBuffers(devcon, dev, Bufferphase); // new vertex buffer
changed = true;

// DO SOME STUFF HERE
}
}
}

UberShader.SendConstantBuffers(devcon);
UberShader.Render(devcon, Bufferphase, true, InstancedMesh[i]->instanceCount); // Render Instanced, one line of code: devcon->DrawInstanced(Bufferphase.IndexCount, instanceCount, 0, 0);

for (int inSM = 0; inSM < (InstancedMesh[i]->SubMeshes.size()); inSM++) // Sub meshes, doesn't really matter so much for now, I'll take care of those.
{
InstancedMesh[i]->SendSubBuffers(devcon, dev, Camera, ObjectBuffer, Bufferphase, inSM);
UberShader.Render(devcon, Bufferphase, true, InstancedMesh[i]->instanceCount);
}
} 

Now for each Lod i check all the instances for the distance, and if the distance is larger, then apply the different vertex buffer. But somehow I only need to apply this vertex buffer to the instances affected by it!

Now how is this possible?

#### Share this post

##### Share on other sites
Advertisement

I would try to just use two different vertex buffers, and you can stick with a single instance buffer.  The trick is to reshuffle your instance data to have all of the high LOD instances in a continuous block, while the lower LOD instances should be in a separate block (you could just as easily implement this with two instance buffers I guess...).  This takes two draw calls, but that isn't a big deal.

This would require modifying the data in the buffers, but it would only be necessary once in a while when the view is significantly moved.  If you store your instances in an intelligent order (like in a quadtree fashion) then moving the data around should be minimally intrusive and will let you quickly get through the update.

#### Share this post

##### Share on other sites

Alright got it working, thanks for pointing the way!

#### Share this post

##### Share on other sites

• Advertisement
• Advertisement
• ### Popular Now

• 11
• 9
• 16
• 16
• 10
• Advertisement
• ### Similar Content

• I wanted to see how others are currently handling descriptor heap updates and management.
I've read a few articles and there tends to be three major strategies :
1 ) You split up descriptor heaps per shader stage ( i.e one for vertex shader , pixel , hull, etc)
2) You have one descriptor heap for an entire pipeline
3) You split up descriptor heaps for update each update frequency (i.e EResourceSet_PerInstance , EResourceSet_PerPass , EResourceSet_PerMaterial, etc)
The benefits of the first two approaches is that it makes it easier to port current code, and descriptor / resource descriptor management and updating tends to be easier to manage, but it seems to be not as efficient.
The benefits of the third approach seems to be that it's the most efficient because you only manage and update objects when they change.

• hi,
until now i use typical vertexshader approach for skinning with a Constantbuffer containing the transform matrix for the bones and an the vertexbuffer containing bone index and bone weight.
Now i have implemented realtime environment  probe cubemaping so i have to render my scene from many point of views and the time for skinning takes too long because it is recalculated for every side of the cubemap.
For Info i am working on Win7 an therefore use one Shadermodel 5.0 not 5.x that have more options, or is there a way to use 5.x in Win 7
My Graphic Card is Directx 12 compatible NVidia GTX 960
the member turanszkij has posted a good for me understandable compute shader. ( for Info: in his engine he uses an optimized version of it )
https://turanszkij.wordpress.com/2017/09/09/skinning-in-compute-shader/
Now my questions
is it possible to feed the compute shader with my orignial vertexbuffer or do i have to copy it in several ByteAdressBuffers as implemented in the following code ?
the same question is about the constant buffer of the matrixes
my more urgent question is how do i feed my normal pipeline with the result of the compute Shader which are 2 RWByteAddressBuffers that contain position an normal
for example i could use 2 vertexbuffer bindings
1 containing only the uv coordinates
2.containing position and normal
How do i copy from the RWByteAddressBuffers to the vertexbuffer ?

(Code from turanszkij )
Here is my shader implementation for skinning a mesh in a compute shader:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 struct Bone { float4x4 pose; }; StructuredBuffer<Bone> boneBuffer;   ByteAddressBuffer vertexBuffer_POS; // T-Pose pos ByteAddressBuffer vertexBuffer_NOR; // T-Pose normal ByteAddressBuffer vertexBuffer_WEI; // bone weights ByteAddressBuffer vertexBuffer_BON; // bone indices   RWByteAddressBuffer streamoutBuffer_POS; // skinned pos RWByteAddressBuffer streamoutBuffer_NOR; // skinned normal RWByteAddressBuffer streamoutBuffer_PRE; // previous frame skinned pos   inline void Skinning(inout float4 pos, inout float4 nor, in float4 inBon, in float4 inWei) {  float4 p = 0, pp = 0;  float3 n = 0;  float4x4 m;  float3x3 m3;  float weisum = 0;   // force loop to reduce register pressure  // though this way we can not interleave TEX - ALU operations  [loop]  for (uint i = 0; ((i &lt; 4) &amp;&amp; (weisum&lt;1.0f)); ++i)  {  m = boneBuffer[(uint)inBon].pose;  m3 = (float3x3)m;   p += mul(float4(pos.xyz, 1), m)*inWei;  n += mul(nor.xyz, m3)*inWei;   weisum += inWei;  }   bool w = any(inWei);  pos.xyz = w ? p.xyz : pos.xyz;  nor.xyz = w ? n : nor.xyz; }   [numthreads(1024, 1, 1)] void main( uint3 DTid : SV_DispatchThreadID ) {  const uint fetchAddress = DTid.x * 16; // stride is 16 bytes for each vertex buffer now...   uint4 pos_u = vertexBuffer_POS.Load4(fetchAddress);  uint4 nor_u = vertexBuffer_NOR.Load4(fetchAddress);  uint4 wei_u = vertexBuffer_WEI.Load4(fetchAddress);  uint4 bon_u = vertexBuffer_BON.Load4(fetchAddress);   float4 pos = asfloat(pos_u);  float4 nor = asfloat(nor_u);  float4 wei = asfloat(wei_u);  float4 bon = asfloat(bon_u);   Skinning(pos, nor, bon, wei);   pos_u = asuint(pos);  nor_u = asuint(nor);   // copy prev frame current pos to current frame prev pos streamoutBuffer_PRE.Store4(fetchAddress, streamoutBuffer_POS.Load4(fetchAddress)); // write out skinned props:  streamoutBuffer_POS.Store4(fetchAddress, pos_u);  streamoutBuffer_NOR.Store4(fetchAddress, nor_u); }

• Hi, can someone please explain why this is giving an assertion EyePosition!=0 exception?

_lightBufferVS->viewMatrix = DirectX::XMMatrixLookAtLH(XMLoadFloat3(&_lightBufferVS->position), XMLoadFloat3(&_lookAt), XMLoadFloat3(&up));
It looks like DirectX doesnt want the 2nd parameter to be a zero vector in the assertion, but I passed in a zero vector with this exact same code in another program and it ran just fine. (Here is the version of the code that worked - note XMLoadFloat3(&m_lookAt) parameter value is (0,0,0) at runtime - I debugged it - but it throws no exceptions.
m_viewMatrix = DirectX::XMMatrixLookAtLH(XMLoadFloat3(&m_position), XMLoadFloat3(&m_lookAt), XMLoadFloat3(&up)); Here is the repo for the broken code (See LightClass) https://github.com/mister51213/DirectX11Engine/blob/master/DirectX11Engine/LightClass.cpp
and here is the repo with the alternative version of the code that is working with a value of (0,0,0) for the second parameter.
https://github.com/mister51213/DX11Port_SoftShadows/blob/master/Engine/lightclass.cpp

• Hi, can somebody please tell me in clear simple steps how to debug and step through an hlsl shader file?
I already did Debug > Start Graphics Debugging > then captured some frames from Visual Studio and
double clicked on the frame to open it, but no idea where to go from there.

I've been searching for hours and there's no information on this, not even on the Microsoft Website!
They say "open the  Graphics Pixel History window" but there is no such window!
Then they say, in the "Pipeline Stages choose Start Debugging"  but the Start Debugging option is nowhere to be found in the whole interface.
Also, how do I even open the hlsl file that I want to set a break point in from inside the Graphics Debugger?

All I want to do is set a break point in a specific hlsl file, step thru it, and see the data, but this is so unbelievably complicated
and Microsoft's instructions are horrible! Somebody please, please help.

• I finally ported Rastertek's tutorial # 42 on soft shadows and blur shading. This tutorial has a ton of really useful effects and there's no working version anywhere online.
Unfortunately it just draws a black screen. Not sure what's causing it. I'm guessing the camera or ortho matrix transforms are wrong, light directions, or maybe texture resources not being properly initialized.  I didnt change any of the variables though, only upgraded all types and functions DirectX3DVector3 to XMFLOAT3, and used DirectXTK for texture loading. If anyone is willing to take a look at what might be causing the black screen, maybe something pops out to you, let me know, thanks.
https://github.com/mister51213/DX11Port_SoftShadows

Also, for reference, here's tutorial #40 which has normal shadows but no blur, which I also ported, and it works perfectly.
https://github.com/mister51213/DX11Port_ShadowMapping

• Advertisement