精品久久久久久,精品久久国产一区二区三区香蕉,亚洲精品无码久久千人斩

glAlphaFunc in OpenGL ES2.0

情绝��D��(fresmaster) — Mon, 19 Nov 2012 14:28:00 GMT

In OpenGL ES 2.0 glAlphaFunc isn't available, you have to implement it in a fragment shader. There isn't a lot of reference out there for this (not that I could find anyway) so I thought I'd write this up.

It's actually quite simple to implement a quick alpha test. I needed to be able to cookie cut out sprites and fonts so I simply needed to reject fragments where the alpha value was zero. Here are the guts of a shader to do this:

#ifdef GL_ES 
precision highp float;
#endif 
uniform sampler2D u_tex0; 
uniform bool u_alphatestenable; 
varying vec2 v_texCoord;
varying vec4 v_color;
void main(){ 
    //calculate the fragment color based on the texture and the vertex colour
    vec4 basecolor = texture2D( u_tex0, v_texCoord ) * v_color;

    //if testing is enabled, check the alpha component and discard if zero      
    if(u_alphatestenable){ 
        if(basecolor.a == 0.0){ 
            //throw this fragment away
            discard;
        }
    }
    gl_FragColor = basecolor;
}

You need to set up a uniform variable u_alphatestenable which enables the alpha test. If you want to support different test types ( less than, greater than etc) then you will need two more uniform variables: one for the test type and one for the value to test against.

int uni_alphatest_enable = glGetUniformLocation(mProgram, "u_alphatestenable"); 
bool alphateston = true; 
glUniform1i(uni_alphatest_enabl

note that you shouldn't call glGetUniformLocation every frame, it should be cached somewhere.
It's quite simple and while you may be thinking oh that is so slow it not that bad. It's faster then the fixed function pipeline which is doing tests for alpha, lights, blend modes etc. If you get paranoid then you can create multiple shaders that support different subsets of features. All you need to be careful of is the cost of calling glSetProgram (to switch shaders) which can be expensive and cause bubbles in the vertex pipeline in the hardware.

情绝��D��(fresmaster) 2012-11-19 22:28 发表评论

Gamma Correction

情绝��D��(fresmaster) — Thu, 15 Nov 2012 15:05:00 GMT

在计��机囑�Ş学领域时常听到gamma correction �Q�gamma correction 控制�?ji��n)图像整体的亮度�Q�reproduce colors也需要gamma correction的一些理论知识，gamma correction不仅仅是控制�?ji��n)图像的亮度�Q�而且�q�控制了(ji��n)RGB各个分量的比例，我们知道渲染器是�U�性的�Q�而显�C�器�q��U�性，其实�?sh��)子打在屏幕上从而��生亮点，�?sh��)子的运动受电(sh��)压控制�Q�这两者是指数关系的，所以��生的亮度也跟�?sh��)压成指数关�p�，而发送给昄��器的voltages范围�?~1�Q?/font>

对于我们输入的图像，如果直接昄��Q�那么就�?x��)篇暗，��?gu��)��已知�?sh��)压与显�C�Z��度的关系�Q�进行gamma correction �Q�其实就是对gamma曲线的修正。一般生产厂家不加说明，他们的伽码值大�U�等�?.5.

代码�Q?/p>

   gammaCorrection = 1 / gamma
   colour = GetPixelColour(x, y)
   newRed   = 255 * (Red(colour)   / 255) ^ gammaCorrection
   newGreen = 255 * (Green(colour) / 255) ^ gammaCorrection
   newBlue  = 255 * (Blue(colour)  / 255) ^ gammaCorrection
   PutPixelColour(x, y) = RGB(newRed, newGreen, newBlue)

知道monitor不是一个线性的�Q�那么我们在�q�行颜色加法�Ӟ��我们得到的颜色�ƈ不是真正的颜色值的相加�Q�比如gamma factor�?.2

red = add (r1, r2);

red= add (0.235,0.156);

对于一个线性设备，red = 0.391,对于未经修正的montior red=0.126;

因�ؓ(f��)有一个幂函数的运��：(x��)C_out = C_in^2.2

现在使用gamma correction :C_corrected= C_out^1.0/2.2

0.391^2.2 = 0.126

0.126^1.0/2.2 = 0.39

我们看到使用伽码校正以后我们能得到我们预想的颜色�?.39.

There are two ways to do gamma correction:

Using the renderer. The renderer (the graphics card or GPU) is a linear device. Modern renderers have the support of gamma correction via sRGB textures and framebuffer formats. See the following OpenGL extensions for more details: GL_ARB_framebuffer_sRGB and GL_EXT_texture_sRGB. With these extensions you can get gamma corrected values for free but gamma correction factor is set to 2.2. You can’t change it.
Using a software gamma correction. The gamma correction is applied to the final scene buffer thanks to apixel shader and you can set the gamma correction you want.

In OpenGL, using GL_ARB_framebuffer_sRGB is really simple: once your FBO is bound, just enable the sRGB space with