Parameters and pixel math

[AlbertoLutzFL]

Following up on the discussion started by Johnny1143, I have not been successful in finding Registax documentation on the pixel formulas that correspond to the various layer functions. I understand that the layers implement each a form of unsharp masking. For the simple case of only one layer with no denoise correction and Gaussian filter, if the starting image is named Io, it would seem that the processed image I1 should be given by:
I1=Io+[Io-Io*B]xS
where * means convolution, B is a gaussian blur kernel, x means product and S is a strength multiplier.
If the above is right, the first questions I have are:
a) Which formula relates the Sharpen parameter (which goes from 0.01 to 2.0) to the value of sigma (in pixels) of the Gaussian kernel?
b) Which formula relates the slider number choice, (which goes from -10 to +100) to the strength value S?

There are follow up questions on denoise and how additional layers interact quantitatively but if someone can clarify the above first it will be much appreciated. Thanks for any help.

[CorBer]

Hi Alberto,

The formula is nearly right (see below). The routines calculate SIGMA by multiplying SHARPEN with 2.35, but be aware that I calculate the convolution efficient using two passes (horizontal/vertical).
S is the value of the Slider.

The formula is
Original is I0
Result1(R1)=I0B1 //this is the blurred version of the original
I1=[Io-R1] //stored the difference (unsharp layer1)
R2=R1B2 //blur again
I2=[I1-R2] //store the difference
R3=R2*B3 //blur again
I2=[I1-R2] //store the difference
The SUM of I1+I2+I3 + R3 equals IO.

When you sharpen the image this is I1S1+ I2S2 + I3*S3 +R3.

And ... its always interesting to hear why people want to know the defaults of this calculation.

cheers
Cor

[AlbertoLutzFL]

Cor, thanks much for the prompt reply. I will get back soon with the rest if you would kindly look into it. On why wanting to know, (as others I suppose) am trying to go a bit beyond an "experimental" approach to using this great tool, by knowing more how to anticipate response to changes - and of course just curiosity! Best, Alberto.

[CorBer]

Hi Alberto,

I have explained this to many people in the past ... you cannot anticipate the response up ahead. So the effect of a given blur-filter is greatly depending on the recording, the stacking etc. Thats why there is no "perfect setting" for the sharpening effect you get to see from a stacked image. For your setup you might find a good setting that often works as a starting "guesstimate" but tweaking by hand is always going to give more chances for better detail.

cheers
Cor