My first instinkt is lots of blur. But it gives totally wrong result if 90% is just black pixels and only 10% is actual height of something . Besides blur makes same kind of halos you see in hipass usually . So I wonder is there any proper math to calculate a height image middle point ?
Hello @kirkr5689!
The Summed Area Table node may be useful in this scenario: it returns the sum of all pixel values in the rectangle from the top left corner up to a pixel.
With this in mind, we can do the following:
1 is done by sampling the pixel on the bottom right corner of the Summed Area Table node output.
2 is done by rounding all pixel values in the source image so that computing their sum effectively returns the number of non-black pixels, then doing the same as in 1
The result looks like this :
You can see it in action in the attached video. Changing the size of the shape does not affect the result, since we ignore pixels with a value of zero.
This returns the average height value. If you want the height middle point, then a simple Distance node may do the trick to eliminate pixels with a value of zero, then using the Min Max node as suggested by @davescm.
I hope this addresses your request!
Best regards.
Already have an account? Login
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.
AdChoices