This topic is 2023 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

Recommended Posts

I want to use a matrix for my instance position and stuff instead of just a vector, but I don't see a semantic for anything more than float4. I'm using a second vertex buffer for my instance data.

Looking here, I see nothing for a matrix / float4x4.
http://msdn.microsoft.com/en-us/library/windows/desktop/bb509647

This, however, makes it look like I can do it, but I have to have 4 seperate parts for the input layout, then I can just simply use it as a float4x4 in the shader? Is that right?
http://www.gamedev.net/topic/608857-dx11-how-to-transfer-matrix-to-vs/

But what is WORLDVIEW? Why isn't it on the MSDN list?

Also, should I be striving to pad my structures out to a specific size?

Share on other sites

It should help if you can paste the code you have so far.

Btw, WORLDVIEW, in simplicity, is the frame of reference of the object you're trying to render. I am surprised why you rely on MSDN so much, there is Google too. ;-)

Share on other sites
Which is one of the problems, why doesn't MSDN have the complete list of these things? It doesn't even mention that WORLDVIEW exists. :/

Ok, before I just used a single vector for my instance position, but I want to be able to rotate, scale, etc, so I figured I'd just make use a matrix instead.

So before, I just did something like this:
layout[2].SemanticName		= "TEXCOORD";
layout[2].SemanticIndex		= 0;
layout[2].Format		= DXGI_FORMAT_R32G32B32A32_FLOAT;
layout[2].InputSlot		= 1;
layout[2].AlignedByteOffset	= 0;
layout[2].InputSlotClass	= D3D11_INPUT_PER_INSTANCE_DATA;
layout[2].InstanceDataStepRate	= 1;

...

VOut vs(float4 pos : POSITION, float4 col : COLOR, float4 ins : TEXCOORD0)
{
VOut output;

pos.x += ins.x;
pos.y += ins.y;
...

Ok, bam, works.

But I'm not sure what I'm doing wrong here trying to access my matrix. I changed my input layout to this:
layout[2].SemanticName		= "TEXCOORD";
layout[2].SemanticIndex		= 0;
layout[2].Format		= DXGI_FORMAT_R32G32B32A32_FLOAT;
layout[2].InputSlot		= 1;
layout[2].AlignedByteOffset	= 0;
layout[2].InputSlotClass	= D3D11_INPUT_PER_INSTANCE_DATA;
layout[2].InstanceDataStepRate	= 1;

layout[3].SemanticName		= "TEXCOORD";
layout[3].SemanticIndex		= 1;
layout[3].Format		= DXGI_FORMAT_R32G32B32A32_FLOAT;
layout[3].InputSlot		= 1;
layout[3].AlignedByteOffset	= D3D11_APPEND_ALIGNED_ELEMENT;
layout[3].InputSlotClass	= D3D11_INPUT_PER_INSTANCE_DATA;
layout[3].InstanceDataStepRate	= 1;

layout[4].SemanticName		= "TEXCOORD";
layout[4].SemanticIndex		= 2;
layout[4].Format		= DXGI_FORMAT_R32G32B32A32_FLOAT;
layout[4].InputSlot		= 1;
layout[4].AlignedByteOffset	= D3D11_APPEND_ALIGNED_ELEMENT;
layout[4].InputSlotClass	= D3D11_INPUT_PER_INSTANCE_DATA;
layout[4].InstanceDataStepRate	= 1;

layout[5].SemanticName		= "TEXCOORD";
layout[5].SemanticIndex		= 3;
layout[5].Format		= DXGI_FORMAT_R32G32B32A32_FLOAT;
layout[5].InputSlot		= 1;
layout[5].AlignedByteOffset	= D3D11_APPEND_ALIGNED_ELEMENT;
layout[5].InputSlotClass	= D3D11_INPUT_PER_INSTANCE_DATA;
layout[5].InstanceDataStepRate	= 1;

...

The post I linked makes it look like I can just do something like this:
VOut vs(float4 pos : POSITION, float4 col : COLOR, float4x4 ins : TEXCOORD)
{
VOut output;

pos = mul(pos, ins);
...


The shader seems to compile fine, but the runtime immediately complains that:
D3D11: ERROR: ID3D11Device::CreateInputLayout: The provided input signature expects to read an element with SemanticName/Index: 'TEXCOORD'/1, but the declaration doesn't provide a matching name. [ STATE_CREATION ERROR #163: CREATEINPUTLAYOUT_MISSINGELEMENT ]

On creation of the input layout. Edited by clashie

Share on other sites
Actually I'm retarded and solved that. I done goofed somewhere and I wasn't giving CreateInputLayout() the right number. I have pretty triangles spread about again.

Still curious as to why the MSDN list of semantics is apparently incomplete, and if I should be trying to pad stuff out to a certain size.

Share on other sites

Structure padding is always better, but if you don't want it then use the pragma directive to overcome memory alignment for each object..

From my experience with MSDN, it generally addresses only whatever is introduced as part of Microsoft's code.

Worldview is more of a graphics concept and they seem to assume that it is already known to programmers, by default.

Share on other sites
Seems kind of weird that they just leave things off their list, but whatever, I guess it doesn't really matter.

Now that things are working, I made it into a starfield. Pretty neat how I can just draw the whole thing in one call, and they can all be uniquely scaled, rotated, etc.
https://dl.dropbox.com/u/10565193/d/starfield.png

edit: Also solving things shortly after I post seems to be a trend of mine. I'm apparently incapable of working things out before I get annoyed and go to ask for help. Edited by clashie

Share on other sites
Nice, a "the first triangle"-starfield. Congrats.

Still curious as to why the MSDN list of semantics is apparently incomplete, and if I should be trying to pad stuff out to a certain size.

A complete list would be quite big,... since you can name them anyway you want (except for the system value semantics) as long as they match.

Share on other sites

just to be clear, MSDN does not specify WORLDVIEW because you can use any semantic name you want. WORLDVIEW is not a system value. you could have named it WV, or WORLDVIEWMATRIX if you wanted, as long as you have matching names in both the input layout and shader parameters

you could even use POS instead of POSITION, or TEXTURECOORDINATE instead of the common TEXCOORD if you wanted. system values (SV_) are values the non-programmable stages of the graphics pipeline such as the rasterizer stage can supply or take in. These are all documented on MSDN as you've seen.

I don't know why microsoft doesn't mention that you can change the semantic name as long as its not a system value, but i'm guessing its to keep things standard

Edited by iedoc

Share on other sites

The docs for semantics are really weird because they cover both DX9-style HLSL (SM2.x and SM3.0) as well as DX10/DX11-style HLSL (SM4.x and SM5.0). In DX9 there were no user-defined semantics, there was a fixed set of semantics. These are all of the uppercase semantics that you see in the first two tables on that documentation page (POSITION, TEXCOORD, COLOR, etc.). When DX10 came along they extended this page to include the SV_ system-value semantics as well, but that page alone doesn't really make it clear that all non-system-value semantics are completely arbitrary in DX10/DX11. The issue is a made a little bit worse by the fact that a lot of people still use the DX9 semantics in their DX10/DX11 shader code, even though they don't have to anymore.

1. 1
2. 2
Rutin
19
3. 3
JoeJ
16
4. 4
5. 5

• 35
• 23
• 13
• 13
• 17
• Forum Statistics

• Total Topics
631700
• Total Posts
3001808
×