Mailing Lists: Apple Mailing Lists
Image of Mac OS face in stamp
 
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Simple ROI/DOD question, and AffineTransform Question.

Hello. I have a simple GLSL alphablending shader I am attempting to port to CIKernel for kicks/learning experience. The following is the original GLSL (fragment shader only cause its all that is interesting);

// define our rectangular texture samplers
uniform sampler2D tex0;
uniform sampler2D tex1;
uniform sampler2D tex2;

// define our varying texture coordinates
varying vec2 texcoord0;
varying vec2 texcoord1;
varying vec2 texcoord2;

void main(void)
{
		
	vec4 input0 = texture2D(tex0, texcoord0);
	vec4 input1 = texture2D(tex1, texcoord1);	
	vec4 mask = texture2D(tex2, texcoord2);

	gl_FragColor = mix(input0, input1, mask.a);

}

simple enough. The CIKernel code I have is as equally as simple:

kernel vec4 v001AlphaBlend(sampler image1, sampler image2, sampler mask)
{
	vec4 image1 = sample(image1, samplerCoord(image1));
	vec4 image2 = sample(image2, samplerCoord(image2));
	vec4 mask = sample(mask, samplerCoord(mask));
		
	return mix(image1, image2, mask.a);	
}

Of course, this works if all the input image sizes are equal. It does not mirror the GLSL output if they differ. Now I understand that Core Image does lazy evaluation and uses the DOD/ROI to optimize the pixels it actually needs. My attempt after reading some posts to edit the filter function results in this non functional code: (it renders the same as before, nesting smaller images/masks within the larger image.

function v001ROIHandler(samplerIndex, dstRect, info)
{
	return info;
}

// set ROI handler
v001AlphaBlend.ROIHandler = v001ROIHandler;

function __image main(__image image1, __image image2, __image mask)
{
// input image sizes - get largest width/height.
var maxwidth = (image1.extent.width >= image2.extent.width) ? image1.extent.width : image2.extent.width;
maxwidth = (maxwidth > mask.extent.width) ? maxwidth : mask.extent.width;

var maxheight = (image1.extent.height >= image2.extent.height) ? image1.extent.height : image2.extent.height;
maxheight = (maxheight > mask.extent.height) ? maxheight : mask.extent.height;

// output image dimensions
var dodRect = new Vec(0.0, 0.0, maxwidth, maxheight);

return v001AlphaBlend.apply(dodRect, dodRect, image1, image2, mask);
}

I fail. Any pointers?

Also, is there an example of using AffineTransform in lieu of mat4 rotation/translation matrices withing a CIKernel? Can I use an affine transform within the kernel itself, or must I break out operations within the 'filter function' and apply transforms 'between' sampling operations? I have some other more interesting shaders which use mat4 matrices which I cannot port straightforwardly.

Curious, and eager to get my GLSL shaders working in CI.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list      (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/email@hidden

This email sent to email@hidden


Visit the Apple Store online or at retail locations.
1-800-MY-APPLE

Contact Apple | Terms of Use | Privacy Policy

Copyright © 2007 Apple Inc. All rights reserved.