• Advertisement
Sign in to follow this  

DX11 Lighting problem in deferred rendering

This topic is 1282 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 !

I have a lighting problem, the scene is lit and a part or it is unlit, I don't understand why. If I amplify light intensity for example, lit parts of the scene are more lit but the part which is in shadow, is better lighted. There is no shadow mapping implemented in this program, and I even not manage to guess what kind of problem it could be.

This is deferred rendering, so in the image below, the lighting is applied on a texture pixel by pixel, which is applied on a quad.

It doesn't seem to be the number of lights (just 1 present in the scene), I tried with more lights and it is the same problem.

Somebody whould have an idea to understand it ? If you have more questions about this, don't hesitate to ask me !

 

[attachment=22703:bugged DX11.png]

Share this post


Link to post
Share on other sites
Advertisement

Hi theScore,

 

can we see how you reconstruct the position, normal, etc...? ( I think this is the fault )

 

in what space are they stored in? View, model, perspective?

 

So to help you further, it would be awesome if we could see this:

  • How you store the GBUFFER
  • How you use/extract from the GBUFFER
  • How you calculate your lighting with the extracted values from the GBUFFER

EDIT: Try to just render the reconstructed normals, positions to see if it looks right. Or use a modern graphics debugger to check the render targets.

 

Cheers!

-MIGI0027

Edited by Migi0027

Share this post


Link to post
Share on other sites

Hi !

Positions and normals are stored in 2 vectors of 4 floats of 16 bit for each one.

Threre are stored in view space.

 

Normals are stored like this :

 

[attachment=22704:normals texture.png]

 

And positions :

 

[attachment=22705:positions texture.png]

 

 

Share this post


Link to post
Share on other sites

Another problem may be here (vertex shader from Gbuffer pass) :

        matrix <float, 4, 4> projView = mul(proj, view) ;
	matrix <float, 4, 4> projViewWorld = mul(projView, world) ;
	

	output.RTposition = mul( input.position, view );
 	output.color = input.color ;
	output.RTnormal = mul( input.normal, view);


 	output.texCoord = input.texCoord ;

	output.finalPos= mul( projViewWorld, input.position );

The world matrix is an identity matrix from CPU code, it is for that, that I only multiply positions and normals by the view matrix.

But I am doing : 

 

positions * viewMatrix 

normals * viewMatrix

 

And :

ProjViewWorldMatrix * positions ; //so, not in the same order...

If I change the order in the shader it doen't work (black screen)

Share this post


Link to post
Share on other sites

Structs in VS :

struct vsIn
{
	half4 position : POSITION ;
 	half4 normal   : NORMAL ;
	half4 texCoord : TEXCOORD0 ; 
	half4 color    : COLOR0 ;
};

struct vsOut
{
	half4 finalPos :SV_POSITION ;
	half4 color : COLOR0 ;
	half4 RTposition : POSITION1 ;
 	half4 RTnormal   : NORMAL   ;
	half4 texCoord : TEXCOORD0 ;

};

And structs in pixel shader :

struct psIn
{	
	half4 finalPos		: SV_POSITION ;
 	half4 color			: COLOR0 ;
	half4 RTposition	: POSITION1 ;
	half4 RTnormal		: NORMAL ;
	half4 texCoord		: TEXCOORD0 ;

};

Share this post


Link to post
Share on other sites

How are the matrices structued? Row Major? Column Major?

 

Just to try, transpose the view matrix thats used on the normals and positions, and do as you did before: viewMatrix * position

Edited by Migi0027

Share this post


Link to post
Share on other sites

I use XMMatrixLookAtLH,it is in row major.

I did that and I have the same result than in my first post : (I am not sure that it is what you expected ?)

	matrix <float, 4, 4> projView = mul(proj, view) ;
	matrix <float, 4, 4> projViewWorld = mul(projView, world) ;
	
	matrix <float, 4, 4> transposedView = transpose(view);

	output.RTposition = mul( transposedView, input.position );
 	output.color = input.color ;
	output.RTnormal = mul( transposedView, input.normal);


 	output.texCoord = input.texCoord ;

	output.finalPos= mul( projViewWorld, input.position );
    
	return output ;

Share this post


Link to post
Share on other sites

Yeah that's what I meant. I've never used XMMatrix, though I should transition to it.

 

Forget the transposing. When you multiply the view matrix with the position, remember that its a point, so you'd need to write mul(view, float4(input.position.xyz, 1)); But for the normal, leave the w component 0, or instead multiply the normals by the 3x3 view matrix.

Share this post


Link to post
Share on other sites

Finally, I did that and it works :

	output.RTposition = mul( input.position, transposedView );
 	output.color = input.color ;
	output.RTnormal = mul( input.normal, view );

And I don't understand why positions can't be multipied by a row major matrix... Do you know it ?

Share this post


Link to post
Share on other sites

Awesome!

 

Im afraid my knowledge of matrices is very limited, maybe someone with a better overview of the concept can help you. I wouldnt want to pass wrong information to you.

 

-MIGI0027

Share this post


Link to post
Share on other sites

Without going into theory/details: when it comes to matrices A * B != B * A. If your matrix is different (row vs column) then multiplication order is reversed, you can also reverse it by transposing all matrices.

Share this post


Link to post
Share on other sites

Ok but here I perform the multiplication in the same order, with the same matrix but transposed once and the other time not transposed, so I don't understand why it works here and with the same processing not ?

Share this post


Link to post
Share on other sites
// these two are equivalent
output.RTposition = mul( input.position, transposedView ); // transposing reverses order
output.RTposition = mul( view, input.position ); // judging by your projViewWorld matrix this is correct order

// if those lines are correct, then this one is wrong order
output.RTnormal = mul( input.normal, view );

Share this post


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

  • Advertisement
  • Advertisement
  • Popular Tags

  • Advertisement
  • Popular Now

  • Similar Content

    • By AxeGuywithanAxe
      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.
    • By evelyn4you
      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); }  
    • By mister345
      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
    • By mister345
      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.
       
       
       

    • By mister345
      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