How to Implement GPUImageMaskFilter using GPUImage

后端 未结 2 1323
难免孤独
难免孤独 2021-02-08 15:31

I need to Cut from the Full Image using the mask and created the masked Image.

\"Full +

相关标签:
2条回答
  • 2021-02-08 16:07

    The mask is the second target, as can been seen in the filter shader code (textureColor2).

    //Averages mask's the RGB values, and scales that value by the mask's alpha
    //
    //The dot product should take fewer cycles than doing an average normally
    //
    //Typical/ideal case, R,G, and B will be the same, and Alpha will be 1.0
    
     lowp float newAlpha = dot(textureColor2.rgb, vec3(.33333334, .33333334, .33333334)) * textureColor2.a;
    
     gl_FragColor = vec4(textureColor.xyz, newAlpha);
    

    Then you need to "invert" your mask : white heart on black background, as the filter uses the "weight" of the RGB pixel value to set the alpha value on the target image.

    So your code should be

    // Image first, Mask next
    [FullGpuImage addTarget:maskingFilter];
    [FullGpuImage processImage];
    
    [maskingFilter useNextFrameForImageCapture];
    
    [maskGpuImage addTarget:maskingFilter];
    [maskGpuImage processImage];    
    

    and your mask (ok I did an ugly quick test, use a proper image) like

    white heart on black background

    for the expected result.

    masked image

    0 讨论(0)
  • 2021-02-08 16:26

    You don't need a mask filter but an alpha mask blend.

    I implemented one here like this:

    // GPUImage shader strings
    NSString * const kNBUAlphaMaskShaderString = SHADER_STRING
    (
     varying highp vec2 textureCoordinate;
     varying highp vec2 textureCoordinate2;
    
     uniform sampler2D inputImageTexture;
     uniform sampler2D inputImageTexture2;
    
     void main()
     {
         lowp vec4 textureColor = texture2D(inputImageTexture, textureCoordinate);
         lowp vec4 textureColor2 = texture2D(inputImageTexture2, textureCoordinate2);
    
         gl_FragColor = vec4(textureColor.xyz, textureColor2.a);
     }
     );
    

    Simply keeps the colors from one image and the alpha channel of the second one.

    Then create a GPUImageTwoInputFilter:

    GPUImageTwoInputFilter * alphaMask = [[GPUImageTwoInputFilter alloc] initWithFragmentShaderFromString:kNBUAlphaMaskShaderString];
    

    I'm not sure if a similar blend filter has been added to GPUImage since.


    Just checked again to see if there's a built-in blend that does it, and there isn't. But the blend filter I used as inspiration is still there (GPUImageAlphaBlendFilter). It merges two images using the alpha mask to mix them. The filter mentioned above doesn't require a second "empty" image.

    0 讨论(0)
提交回复
热议问题