Hi guys!

This may seem extremely simple, yet it isn't for me, and therefore I'm asking for your help on this topic. I'd like to convert coordinates from 0 to 1 that start at the top left corner, and end in the bottom right corner. Unfortunately I can't just insert these values into the NDC coordinates in my vertex shader, as that's not how they work, description from msdn: (http://msdn.microsoft.com/en-us/library/windows/desktop/cc308049(v=vs.85).aspx)

The reason for this is because I'm starting on my 2d simple gui, nothing fancy. So I simply have a structure called Rectangle, defining x0, y0, x1, y1, where 0,0 is the top left cornor, and 1,1 is the bottom right, but I'd need to convert these into the NDC space coordinates, how?

Maybe I should know how to do this, maybe not, either way, how could this be done? Any input would be appreciated.

Thank you, as usual.

-MIGI0027

x0 = x0 * 2.0f - 1.0f;
x1 = x1 * 2.0f - 1.0f;
y0 = y0 * -2.0f + 1.0f;
y1 = y1 * -2.0f + 1.0f;

Thank you.

When using the coordinates: 0, 0, .5f, .5f, and transforming them using your method, the result isn't as expected.

I would have expected some thing going from the top left to the middle of the screen. My implementation:

// Creating the vertices
Surface->SetArea(&CDK::Rectangle::ToScreen(CDK::Rectangle(0, 0, .5f, .5f));

// ::ToScreen, I do realize that this can be optimized, but that's not the goal for now.
CDK::Rectangle& CDK::Rectangle::ToScreen(Rectangle& a)
{
	a.x0 = a.x0 * 2.0f - 1.0f;
	a.x1 = a.x1 * 2.0f - 1.0f;
	a.y0 = a.y0 * 2.0f - 1.0f;
	a.y1 = a.y1 * 2.0f - 1.0f;

	return a;
}

Edit:

...
Rectangle(float _x0, float _y0, float _x1, float _y1)
{
	x0 = _x0;
	x1 = _x1;
	y0 = _y0;
	y1 = _y1;
}

OUT = IN;

CDK::Rectangle& CDK::Rectangle::ToScreen(Rectangle& a)
{
	a.x0 = a.x0 * 2.0f - 1.0f;
	a.x1 = a.x1 * 2.0f - 1.0f;
	a.y0 = a.y0 * -2.0f + 1.0f;
	a.y1 = a.y1 * -2.0f + 1.0f;

	return a;
}

Sorry for the confusion here, my vertex shader isn't anything fancy as it's user customization. (Notice the #REGION things, they are replaced by the engine).

Maybe I should, but no, there's no projection matrix involved here. I'm simply sending a structure containing x, y, tx, ty, (t -> texture).

So to convert those coordinates I'd have to use the ortho matrix? (D3DXMatrixOrthoLH)

#REGION_INPUTS
SamplerState ss;

struct VS_Output
{  
	float4 Pos : SV_POSITION;
	float2 Tex : TEXCOORD0;
};
 
VS_Output VShader(float2 position : POSITION, float2 tex : TEXCOORD)
{
	VS_Output Output;
	Output.Tex = tex; // Just pass it
	Output.Pos = float4(position.xy, 0, 1);

#REGION_VERTEX
	
	return Output;
}

float4 PShader(VS_Output input) : SV_TARGET
{
	float4 color;

#REGION_PIXEL

	return color;
}

PS. The post you sent me isn't read yet, doing it after this post is written.

And thank you.

I provided [EDIT] tags in both posts to show the solution to your problem.

L. Spiro

Thank you, it seems to work.

-MIGI0027