# DX11 D3D + GLM Depth Reconstruction Issues

## Recommended Posts

 I'm trying to port my engine to DirectX and I'm currently having issues with depth reconstruction. It works perfectly in OpenGL (even though I use a bit of an expensive method). Every part besides the depth reconstruction works so far. I use GLM because it's a good math library that has no need to install any dependencies or anything for the user. So basically I get my GLM matrices: struct DefferedUBO { glm::mat4 view; glm::mat4 invProj; glm::vec4 eyePos; glm::vec4 resolution; }; DefferedUBO deffUBOBuffer; // ... glm::mat4 projection = glm::perspective(engine.settings.fov, aspectRatio, 0.1f, 100.0f); // Get My Camera CTransform *transform = &engine.transformSystem.components[engine.entities[entityID].components[COMPONENT_TRANSFORM]]; // Get the View Matrix glm::mat4 view = glm::lookAt( transform->GetPosition(), transform->GetPosition() + transform->GetForward(), transform->GetUp() ); deffUBOBuffer.invProj = glm::inverse(projection); deffUBOBuffer.view = glm::inverse(view); if (engine.settings.graphicsLanguage == GRAPHICS_DIRECTX) { deffUBOBuffer.invProj = glm::transpose(deffUBOBuffer.invProj); deffUBOBuffer.view = glm::transpose(deffUBOBuffer.view); } // Abstracted so I can use OGL, DX, VK, or even Metal when I get around to it. deffUBO->UpdateUniformBuffer(&deffUBOBuffer); deffUBO->Bind());  Then in HLSL, I simply use the following: cbuffer MatrixInfoType { matrix invView; matrix invProj; float4 eyePos; float4 resolution; }; float4 ViewPosFromDepth(float depth, float2 TexCoord) { float z = depth; // * 2.0 - 1.0; float4 clipSpacePosition = float4(TexCoord * 2.0 - 1.0, z, 1.0); float4 viewSpacePosition = mul(invProj, clipSpacePosition); viewSpacePosition /= viewSpacePosition.w; return viewSpacePosition; } float3 WorldPosFromViewPos(float4 view) { float4 worldSpacePosition = mul(invView, view); return worldSpacePosition.xyz; } float3 WorldPosFromDepth(float depth, float2 TexCoord) { return WorldPosFromViewPos(ViewPosFromDepth(depth, TexCoord)); } // ... // Sample the hardware depth buffer. float depth = shaderTexture[3].Sample(SampleType[0], input.texCoord).r; float3 position = WorldPosFromDepth(depth, input.texCoord).rgb;  Here's the result: This just looks like random colors multiplied with the depth. Ironically when I remove transposing, I get something closer to the truth, but not quite: You're looking at Crytek Sponza. As you can see, the green area moves and rotates with the bottom of the camera. I have no idea at all why. The correct version, along with Albedo, Specular, and Normals.

##### Share on other sites

GL's NDC (post projection) Z coordinates range from -1 to 1, but D3D's range from 0 to 1.

glm::perspective will create a GL style projection matrix. You need to concatenate this with a matrix that scales z by 0.5 and translates by 0.5 to make it valid for D3D.

In normal rendering, the effect of this bug will be quite small - your near plane appearing about twice as far forward as you intended... But it will mess with depth reconstruction too.

Btw there should be no need to transpose your matrices on D3D - both GLSL and HLSL store 2D arrays in column-major element ordering.

##### Share on other sites

Thanks for the quick response. Firstly, GL is column major whereas directx is row major. I've already had to transpose for my first geometry stage and it works well.

Second, will I need to change my first stage to accommodate this change as well? Also can I just multiply it by glm::translate(0,0,0.5)xProjection

EDIT: I've switched to row major vertices in DirectX using the following:

#pragma pack_matrix( row_major )

I guess DirectX just uses row major by default. I'm still having the same issues though. I tried using the following in ViewPosFromDepth:

float z = depth * 0.5 + 0.5;

Edited by KarimIO
Update

##### Share on other sites
1 hour ago, KarimIO said:

Thanks for the quick response. Firstly, GL is column major whereas directx is row major. I've already had to transpose for my first geometry stage and it works well.

Second, will I need to change my first stage to accommodate this change as well? Also can I just multiply it by glm::translate(0,0,0.5)xProjection

1- That's old info that hasn't applied since the fixed function graphics days. D3D/GL don't pick conventions for you. You can use any conventions on either API.

GLSL and HLSL both use column-major array indexing by default (but can be told to use row-major indexing such as with that pragma). Both can work with column-vector maths or row-vector maths (i.e. whether you write mul(vec,mat) or mul(mat,vec))

IIRC, GLM uses column-major storage and column-vector maths, and DirectXMath / D3DX use row-major storage and row-vector math... Which ironically results in them storing the exact same pattern of 64 bytes in RAM, but requires opposite multiplication order by the programmer

If you use the same math library, you can use the same shader code and matrix data on both APIs with no need to transpose anything.

2- yeah whenever you produce a projection matrix you have to scale in z by 0.5 and then translate in z by 0.5. You need to do this for your vertex shaders so that the rasterization is correct. When fetching from the depth buffer, don't scale/offset the fetched value.

##### Share on other sites
1 hour ago, Hodgman said:

yeah whenever you produce a projection matrix you have to scale in z by 0.5 and then translate in z by 0.5. You need to do this for your vertex shaders so that the rasterization is correct. When fetching from the depth buffer, don't scale/offset the fetched value

I've tried this, but now it's far too zoomed in. Originally, it did look quite like my OpenGL results. Is there a reason for this?

projection *= glm::translate(glm::vec3(0.0f,0.0f,0.5f)) * glm::scale(glm::vec3(1.0f,1.0f,0.5f));

Edited by KarimIO
Formatting

##### Share on other sites

If GLM uses column-vector maths, your multiplication order might be backwards there. Try:

projection = glm::translate(glm::vec3(0.0f,0.0f,0.5f)) * glm::scale(glm::vec3(1.0f,1.0f,0.5f)) * projection;

##### Share on other sites

Okay thank you a lot, Hodgman I finally got it to work! But I do have a question, my main vertex.hlsl which takes the actual geometry and pushes it into the gbuffer requires row-major whereas the rest works fine using column major. Do you have any idea why that could be?

##### Share on other sites

@Hodgman Got any idea for the question above?

##### Share on other sites

Do you use the exact same VS math in your HLSL and GLSL versions?

Post some VS shader code and we'll have a look

##### Share on other sites

@Hodgman Sorry! Didn't see the response until now! Keep in mind as I used GLM, it's column major. Here's the code that works:

////////////////////////////////////////////////////////////////////////////////
// Filename: mainVert.vs
////////////////////////////////////////////////////////////////////////////////

/////////////
// GLOBALS //
/////////////
#pragma pack_matrix( row_major )

cbuffer MatrixBuffer
{
matrix worldMatrix;
matrix viewMatrix;
matrix projectionMatrix;
};

//////////////
// TYPEDEFS //
//////////////
struct VertexInputType {
float3 position : POSITION;
float3 normal : NORMAL;
float3 tangent : TANGENT;
float2 texCoord : TEXCOORD0;
};

struct PixelInputType {
float4 position : SV_POSITION;
float3 worldPosition : POSITION;
float3 normal : NORMAL;
float3 tangent : TANGENT;
float2 texCoord : TEXCOORD0;
};

////////////////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////////////////
PixelInputType main(VertexInputType input) {
float4 position;
PixelInputType output;

// Change the position vector to be 4 units for proper matrix calculations.
position = float4(input.position, 1.0f);

// Calculate the position of the vertex against the world, view, and projection matrices.
position = mul(position, worldMatrix);
output.worldPosition = position.xyz;
position = mul(position, viewMatrix);
output.position = mul(position, projectionMatrix);

output.normal = normalize(mul(float4(input.normal, 1.0), worldMatrix).xyz);
output.tangent = normalize(mul(float4(input.tangent, 1.0), worldMatrix).xyz);
output.texCoord = float2(input.texCoord.x, -input.texCoord.y);

return output;
}
////////////////////////////////////////////////////////////////////////////////
// Filename: pointLightFrag.ps
////////////////////////////////////////////////////////////////////////////////

#pragma pack_matrix( column_major )

#include "inc_transform.hlsl"
#include "inc_light.hlsl"

//////////////
// TYPEDEFS //
//////////////
struct PixelInputType {
float4 position : SV_POSITION;
float2 texCoord : TEXCOORD0;
float3 viewRay : POSITION;
};

SamplerState SampleType[4];

cbuffer MatrixInfoType {
matrix invView;
matrix invProj;
float4 eyePos;
float4 resolution;
};

cbuffer Light {
float3 lightPosition;
float3 lightColor;
float lightIntensity;
};

////////////////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////////////////
float4 main(PixelInputType input) : SV_TARGET {
float3 Position = WorldPosFromDepth(invProj, invView, depth, input.texCoord);
//return float4(position, 1.0);
/*float near = 0.1;
float far = 100;
float ProjectionA = far / (far - near);
float ProjectionB = (-far * near) / (far - near);
depth = ProjectionB / ((depth - ProjectionA));
float4 position = float4(input.viewRay * depth, 1.0);*/
// Convert to World Space:
// position = mul(invView, position);

float3 lightPow = lightColor * lightIntensity;
float3 outColor = LightPointCalc(Albedo.rgb, Position.xyz, Specular, Normal.xyz, lightPosition, lightAttenuationRadius, lightPow, eyePos.xyz); // hdrGammaTransform()
return float4(hdrGammaTransform(outColor), 1.0f);
}

## Create an account

Register a new account

• 10
• 16
• 9
• 13
• 41
• ### Similar Content

• By chiffre
Introduction:
In general my questions pertain to the differences between floating- and fixed-point data. Additionally I would like to understand when it can be advantageous to prefer fixed-point representation over floating-point representation in the context of vertex data and how the hardware deals with the different data-types. I believe I should be able to reduce the amount of data (bytes) necessary per vertex by choosing the most opportune representations for my vertex attributes. Thanks ahead of time if you, the reader, are considering the effort of reading this and helping me.
I found an old topic that shows this is possible in principal, but I am not sure I understand what the pitfalls are when using fixed-point representation and whether there are any hardware-based performance advantages/disadvantages.
(TLDR at bottom)
The Actual Post:
To my understanding HLSL/D3D11 offers not just the traditional floating point model in half-,single-, and double-precision, but also the fixed-point model in form of signed/unsigned normalized integers in 8-,10-,16-,24-, and 32-bit variants. Both models offer a finite sequence of "grid-points". The obvious difference between the two models is that the fixed-point model offers a constant spacing between values in the normalized range of [0,1] or [-1,1], while the floating point model allows for smaller "deltas" as you get closer to 0, and larger "deltas" the further you are away from 0.
To add some context, let me define a struct as an example:
struct VertexData { float[3] position; //3x32-bits float[2] texCoord; //2x32-bits float[3] normals; //3x32-bits } //Total of 32 bytes Every vertex gets a position, a coordinate on my texture, and a normal to do some light calculations. In this case we have 8x32=256bits per vertex. Since the texture coordinates lie in the interval [0,1] and the normal vector components are in the interval [-1,1] it would seem useful to use normalized representation as suggested in the topic linked at the top of the post. The texture coordinates might as well be represented in a fixed-point model, because it seems most useful to be able to sample the texture in a uniform manner, as the pixels don't get any "denser" as we get closer to 0. In other words the "delta" does not need to become any smaller as the texture coordinates approach (0,0). A similar argument can be made for the normal-vector, as a normal vector should be normalized anyway, and we want as many points as possible on the sphere around (0,0,0) with a radius of 1, and we don't care about precision around the origin. Even if we have large textures such as 4k by 4k (or the maximum allowed by D3D11, 16k by 16k) we only need as many grid-points on one axis, as there are pixels on one axis. An unsigned normalized 14 bit integer would be ideal, but because it is both unsupported and impractical, we will stick to an unsigned normalized 16 bit integer. The same type should take care of the normal vector coordinates, and might even be a bit overkill.
struct VertexData { float[3] position; //3x32-bits uint16_t[2] texCoord; //2x16bits uint16_t[3] normals; //3x16bits } //Total of 22 bytes Seems like a good start, and we might even be able to take it further, but before we pursue that path, here is my first question: can the GPU even work with the data in this format, or is all I have accomplished minimizing CPU-side RAM usage? Does the GPU have to convert the texture coordinates back to a floating-point model when I hand them over to the sampler in my pixel shader? I have looked up the data types for HLSL and I am not sure I even comprehend how to declare the vertex input type in HLSL. Would the following work?
struct VertexInputType { float3 pos; //this one is obvious unorm half2 tex; //half corresponds to a 16-bit float, so I assume this is wrong, but this the only 16-bit type I found on the linked MSDN site snorm half3 normal; //same as above } I assume this is possible somehow, as I have found input element formats such as: DXGI_FORMAT_R16G16B16A16_SNORM and DXGI_FORMAT_R16G16B16A16_UNORM (also available with a different number of components, as well as different component lengths). I might have to avoid 3-component vectors because there is no 3-component 16-bit input element format, but that is the least of my worries. The next question would be: what happens with my normals if I try to do lighting calculations with them in such a normalized-fixed-point format? Is there no issue as long as I take care not to mix floating- and fixed-point data? Or would that work as well? In general this gives rise to the question: how does the GPU handle fixed-point arithmetic? Is it the same as integer-arithmetic, and/or is it faster/slower than floating-point arithmetic?
Assuming that we still have a valid and useful VertexData format, how far could I take this while remaining on the sensible side of what could be called optimization? Theoretically I could use the an input element format such as DXGI_FORMAT_R10G10B10A2_UNORM to pack my normal coordinates into a 10-bit fixed-point format, and my verticies (in object space) might even be representable in a 16-bit unsigned normalized fixed-point format. That way I could end up with something like the following struct:
struct VertexData { uint16_t[3] pos; //3x16bits uint16_t[2] texCoord; //2x16bits uint32_t packedNormals; //10+10+10+2bits } //Total of 14 bytes Could I use a vertex structure like this without too much performance-loss on the GPU-side? If the GPU has to execute some sort of unpacking algorithm in the background I might as well let it be. In the end I have a functioning deferred renderer, but I would like to reduce the memory footprint of the huge amount of vertecies involved in rendering my landscape.
TLDR: I have a lot of vertices that I need to render and I want to reduce the RAM-usage without introducing crazy compression/decompression algorithms to the CPU or GPU. I am hoping to find a solution by involving fixed-point data-types, but I am not exactly sure how how that would work.
• By cozzie
Hi all,
I was wondering it it matters in which order you draw 2D and 3D items, looking at the BeginDraw/EndDraw calls on a D2D rendertarget.
The order in which you do the actual draw calls is clear, 3D first then 2D, means the 2D (DrawText in this case) is in front of the 3D scene.
The question is mainly about when to call the BeginDraw and EndDraw.
Note that I'm drawing D2D stuff through a DXGI surface linked to the 3D RT.
Option 1:
A - Begin frame, clear D3D RT
B - Draw 3D
C - BeginDraw D2D RT
D - Draw 2D
E - EndDraw D2D RT
F - Present
Option 2:
A - Begin frame, clear D3D RT + BeginDraw D2D RT
B - Draw 3D
C - Draw 2D
D - EndDraw D2D RT
E- Present
Would there be a difference (performance/issue?) in using option 2? (versus 1)
Any input is appreciated.

• Do you know any papers that cover custom data structures like lists or binary trees implemented in hlsl without CUDA that work perfectly fine no matter how many threads try to use them at any given time?
• By cozzie
Hi all,
Last week I noticed that when I run my test application(s) in Renderdoc, it crashes when it enable my code that uses D2D/DirectWrite. In Visual Studio no issues occur (debug or release), but when I run the same executable in Renderdoc, it crashes somehow (assert of D2D rendertarget or without any information). Before I spend hours on debugging/ figuring it out, does someone have experience with this symptom and/or know if Renderdoc has known issues with D2D? (if so, that would be bad news for debugging my application in the future );
I can also post some more information on what happens, code and which code commented out, eliminates the problems (when running in RenderDoc).
Any input is appreciated.

• Hi Guys,
I understand how to create input layouts etc... But I am wondering is it at all possible to derive an input layout from a shader and create the input layout directly from this? (Rather than manually specifying the input layout format?)