On Jan 20, 2011, at 3:42 PM, Christopher Wright wrote:
>> I have a coupla questions... First, why does it only display a center cut
>> rectangle of the full image. Why when I put it in a Render in Image node
>> does it sort of work, but then comes off its wheels and starts displaying
>> random noise? (It does show the full image when I feed the Render In Image
>> node into a Billboard).
>
> It only displays the center because the GLSL Grid's size is too big -- it
> should be 2 wide at most, and it looks to be 3 and a fraction. The Width and
> Height inputs control this, so you can see which patches control this (Image
> Dimensions, I think)
But isn't that the sort of thing that Image Dimensions is supposed to be for?
Even when I set it to 1 or 2 manually, the center cut behavior does not change.
>
> Render In Image probably needs a clear patch. If that doesn't help, please
> post a sample composition that shows the problem you're encountering.
Well, when I add a clear patch, if I enable it, it goes black, if I then
disable it, things are fine. Until I change something like the input image,
etc... then I need to manually clear it to black, then disable it.
>> Last but not least... How does it work? I've looked at the code and it seems
>> so simple. The Fragment code just feeds all pixels out to the Vertex code
>> and it does some tiny magic, and then boom, there it is.
>
> GPUs have been built to interpolate values between corners (Normals, Colors,
> texture coordinates) -- they do this extraordinarily well. This shader takes
> advantage of that -- it's basically "free" (no math required on your part)
> because of the way GPUs work.
I at least know that much, thanks. It was the reason for my question to start
with. I figured the GPU was all about interpolating triangle vertex values and
should be able to do this in its sleep as opposed to the math I was needing to
perform in my CI Kernel prototype.
> Also, the fragment doesn't go to the vertex - it's the other way around:
> Vertex Shaders operate on the vertices supplied (possibly changing them, and
> setting up data for the fragment shader), and then the fragment shader runs
> for each pixel defined by the (possibly changed) vertices. There are
> typically many many more fragments ("pixels", but not always) than there are
> vertices, so vertex-shader-supplied data is interpolated across the fragments
> by the GPU. You happened to want data interpolated in that manner, which
> made it pretty simple. :)
Okay, but how does this Vertex:
varying vec4 pixelColor;
uniform sampler2DRect texture;
void main()
{
//Transform vertex by modelview and projection matrices
gl_Position = gl_ModelViewProjectionMatrix * gl_Vertex;
//Forward current color and texture coordinates after applying texture
matrix
gl_FrontColor = gl_Color;
gl_TexCoord[0] = gl_TextureMatrix[0] * gl_MultiTexCoord0;
pixelColor = texture2DRect(texture, gl_TexCoord[0].xy);
}
and this Fragment:
varying vec4 pixelColor;
void main()
{
//Multiply color by texture
gl_FragColor = gl_Color * pixelColor;
}
Specify to interpolate the pixel values found at the vertices of the triangles?
I understand that GLSL Grid is determining the size of the triangles. I just
don't get how it's only sampling the corner points and providing an
interpolation between them as the values for the intermediate pixels.
I really appreciate everyone's time on this. Thank you...
Patrick
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
