Davide Barranca | Notes on sharpening

Article Index
Davide Barranca | Notes on sharpening
2. Gaussian sharpening
3. Difference of Gaussians
4. Other blurring kernels
5. Combined use of different blurring kernels
6. Image decomposition
7. A sharpening equalizer
8. Bilateral and WLS Pyramids
9. Mixed Pyramids
10. (temporary) Conclusions
11. Links
12. Acronym list
13. Acknowledgements
14. Notes on the notes’ author
Comments
All Pages

3. Difference of Gaussians

Let’s shift a bit our point of view; what if, instead of subtracting a blurred version from the untouched original, the subtraction is between two differently blurred originals? Have a look to the graph before we investigate the why of such a move with GBs:

(Fig 3.1) Intensity transition and Difference of Gaussians.

(Fig 3.1) Intensity transition and Difference of Gaussians.

The concept behind the Difference of Gaussians (DoG) is quite simple: noise is usually high frequency spatial information, and it’s blown away in both blurred version, so won’t be boosted; on the other hand the two versions keep detail in different frequency ranges. So their subtraction is a way to enhance, precisely, that frequency window:

(Fig 3.2) The subtraction between two versions of the original blurred with different radii (r1</p> <p class=(Fig 3.2) The subtraction between two versions of the original blurred with different radii (r12).

(Fig 3.3) The original plus the difference gives the sharpened version: high frequency has not been affected by DoG (to have a gentler effect it could be suggested a lower opacity LL blend or an inverse S-shaped curve to reduce the contrast of the difference layer).

(Fig 3.3) The original plus the difference equals the sharpened version: high frequency has not been affected by DoG (to have a gentler effect it could be suggested a lower opacity LL blend or an inverse S-shaped curve to reduce the contrast of the difference layer).

(Fig 3.4) DoG with R1=4.0 and R2=20.0 and SS2, applied LL with an extra opacity lowering at 50%.

I’m recalling here the DoG because it represent a ïŹrst step off the traditional sharpening track: I can report that it’s usually applied using a blur ratio ranging from 4:1 to 5:1, while a ratio of 1.6 mimics the Laplacian of Gaussian (LoG: an operator that calculates the second derivative of signal intensity, so it’s good for instance in ïŹnding edges. I plan to add more on LoG here later on).

I still have to test it extensively: nevertheless an important feature that should be noted is its resemblance to HiRaLoAm (as Dan Margulis uses to call USM with High Radius Low Amount). But there’s a remarkable difference: namely, that edges are less or not sharpened at all, because they belong to the high frequency detail window that’s untouched by DoG (less or no difference between edges in GB1 and GB2, so less or no boosting at all).This could be a beneïŹt in workïŹ‚ows where different sharpening rounds are planned and an extra step is worth its time.



Comments
Add New Search
"Shadow/Highlights" clarification please
George Machen 2009-01-07 01:58:49

> 8. Bilateral and WLS Pyramids
>
> ...Shadow/Highlights with SB (instead of GB)
in its
> “engine” gives the same halo-free look by the way...

I don't have
CS4. Are you saying that its Shadow/Highlights dialog now features an option to
choose whether it uses Gaussian or Surface Blur? If not, what did you mean
exactly in this statement?
Thanks! (G-R-E-A-T write-up, Davide!)
Davide 2009-01-07 22:53:37

Hi George,
I'm glad you liked it!
No, unfortunately there's no GB/SB option in
CS4. But you can always custom build your Shadows/Highlights: for instance,
create a curve adjustment layer to lighten the picture, then add a luminosity
mask to it (or choose the channel that better suits the image needs and use it
as a mask). Now you can experiment using GB and SB for blurring the mask: the
first one gives you a standard Photoshop S/H feeling, while SB (as long as the
algorithm doesn't reverse) shows less or no halos.
Ciao,
Davide
Alexey Rybakov 2009-01-07 19:58:18

Thanks for nice article with informative illustrations and examples. I'll glad
to read more if you'll develop this farther.
Re: "Shadow/Highlights" clarification please
Stephen Marsh 2009-01-08 12:48:48

Thank you Davide, this is a great exploration, I am eagerly awaiting some
actions!

George, I presume that Davide is referring to manually simulating what
the Shadows/Highlights command does. Dan Margulis described a method for this
and one of the Applied Color Theory list members created an action. This action
is available to list members from the files section.

Margulis PSU
ON2004.atn
http://tech.groups.yahoo.com/group/colo rtheory/files/
How to
re-create the Shadow/Highlight effect of Photoshop CS as described by Dan
Margulis in the Oct/Nov issue of Photoshop User titled "Making Two Ends
Meet", pg. 44-47.


Regards,

Stephen
Marsh
http://members.ozemail.com.au/~binaryfx/
Bravissimo !!!
Andrea Baldini 2009-01-08 23:02:13

Devo dire che Ă© una roba incredibile !

Davvero superba la genialitĂ  del
progetto, non avevo mai visto nulla di simile!
Complimenti anche per le
illustrazioni realizzate molto bene.
Spero che l' articolo venga tradotto anche
in italiano.

Complimenti ancora,

Saluti

Andrea Baldini
good work
Jack 2009-02-10 23:13:44

I'd love to test the actions when you create those. Also it would be nice to
see samples on photographs as apposed to the painting your using here. It's
difficult to see the sharpening.
thank you!
Timo Kirves 2009-03-14 17:51:02

Dear Davide,

The whole series of your techniques open dramatically better
chances to
succesful fine art photographic prints with todays inkjet
printers.

I work as a photographer and instructor in Helsinki, Finland, working
on both own and colleaguesÂŽphotographic prints.

Thank you for shearing this
work of yours with the rest of community.

Best

Timo

Kirves
Davide 2009-03-14 22:48:29

Dear Timo, thank you very much for your comment, knowing that some of my
thoughts may help someone out there getting better results and enjoying his
pictures is a great reward for me.
I wish I had more time to explore other
interesting topics - I've something about local contrast in mind but I'd like to
dig a bit deeper before publishing.
Kind regards,

Davide
Thanks Davide
Miriam Calzada 2010-05-18 00:39:32

Hola, mi nombre es Miriam soy fotografo. Vivo y trabajo en Republica Dominicana.
He disfrutado mucho toda esta informacion que ofreces.
Trabajo por años con
Hasselblad y pongo mucho interes en cualquier forma de perfeccionar mi
trabajo.
Especialmente viniendo de un profesional como
tu.
Atentamente.
Miriam


miriamcalzada.com
Thank you Miriam!
Davide Barranca 2010-06-01 09:23:46

It's a great gift to read feedbacks like yours!
I've seen the images in your
website, and I'm sure the actual prints must be really gorgeous.
Thanks again ad
all the best,
Davide
Write comment
Name:
Email:
 
Title:
Please input the anti-spam code that you can read in the image.

3.26 Copyright (C) 2008 Compojoom.com / Copyright (C) 2007 Alain Georgette / Copyright (C) 2006 Frantisek Hliva. All rights reserved."