Brendan Ryan wrote:
>Apache wrote:
>> 7.4.5 Focal blur hexgrid constants
>> Hex_Blur1 = 7
>> Hex_Blur2 = 19
>> Hex_Blur3 = 37
>>
>> I have two questions about this:
>> 1. What are they used for?
>> 2. I'd like to know how I might calculate Hex_blur4, Hex_blur5 etc etc
>> 3. When using focal blur, does blur_samples 8 look worse blur_samples 7?
>>
>
>I don't know what they are, but I used my TI-86 to find that the points
>(1,7), (2,19) and (3,37) fit on the curve y = 3x^2 + 3x + 1 where x
>would be the x in Hex_Blurx and y would be the constants.
>According to the above equation, Hex_Blur4 would be 61, Hex_Blur5 would
>be 91 and Hex_Blur6 would be 127.
>I used the quadratic fit because it gave 7, 19 and 37 extactly.  The
>other fits like linear and logrithimic gave correlations of .99 or .97.
>
>Brendan
>
Um, a quadratic fit with only three points will ALWAYS fit those three
points exactly.  When your fit model has the same number of points as your
data set, the model will always fit exactly and yet will be basically
meaningless.  Or to be more blunt, the parabola found here is no more
meaningful than the line you'd get from fitting a set of two points.  It
might be the best you can do, given the limited data set, but there's no
reason to believe it reflects the underlying structure.

Post a reply to this message