Reproduce Unsharp Mask Using Gaussian Blur

Question

Hello,
I know the effect of Unsharp Mask can be achieved using Gaussian Blur and Blending Mode only.
Given a layer 'L' (The original) its USM(L, Amount - 100, Radius - 10, Threshold - 0) is given by:
GBL = GB(L, Radius - 10) - Gaussian Blur.
Difference  = L - BGL.
USM = L + Difference.
I made it using 'Image Apply' and Gaussian Blur yet the result is different.
Does Photoshop uses different method?
How can the accurate result be achieved?

Answer 1

Hi Chris,
All I did was as follwoing:
1. Created black image in 32 Bit Mode (RGB Mode so numbers will be the same in MATLAB).
2. Applied GB with Radius 3.
3. Saved as 32 bit TIF.
4. Loaded at MATLAB using imread.
5. Extracted the first channel (They aer all the same).
The result is up.
Something went wrong with the numbers.
Anyhow, how come the result of my classic Gaussian Kernel are so different from Photoshop's?
Here are the example:
This is the GB Kernel of Photoshop for Radius = Sigma = 2:
Here is the GB Kernel of a classic GB using STD of 2 and support of size [11, 11] (Just like photoshop):
And here is the difference:
As you can see, it is very noticeable difference.
The blur using the classic kernel is much harsher.
Moreover, Using the classic kernel trying to replicate USM (   USM(O) = O + 2(O - B)   )yields something completely different.
What am I missing here?
Thank You.
P.S.
Well, I'm just playing for fun, so I don't have "Run Time" limitations :-).