Forums > Digital Art and Retouching > Selecting local highlights and local shadows

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

I feel rather stupid now...i know quite something about behind IHP, FS, illuminosity masks,...but somehow this keeps eluding me, as if it's like being on the tip of my tong...:

how to select local highlights and local shadows in order to create masks. Probably with a given radius

i guess if i have to put it different what is outside a certain median....

probably it's very easy, but even googling didn't help very much...

Herman
www.hermanvangestel.com

Jul 30 14 11:01 am Link

Retoucher

DevKevin

Posts: 120

Wittenberg, Wisconsin, US

Have you tried color range for highlights and shadows?

Jul 30 14 11:07 am Link

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

DevKevin wrote:
Have you tried color range for highlights and shadows?

that doesn't do local does it?...

my feeling says it probably has to do something with USM, HiRaLoAm...but then.....

Jul 30 14 11:09 am Link

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

ha got it...but only with creating 2 layars after HiRaLoAm, one to blend "lighten", other to blend "darken" and do a  "difference"...the result is  respectively 2 masks, 1 for the local lights, other for the local shadows..

But if somebody knows a smarter way?

Herman
www.hermanvangestel.com

Jul 30 14 11:15 am Link

Photographer

Giacomo Cirrincioni

Posts: 22232

Stamford, Connecticut, US

Jul 30 14 12:24 pm Link

Photographer

Tulack

Posts: 836

Albuquerque, New Mexico, US

Why do you think luminosity mask wouldn't work?

Jul 30 14 12:53 pm Link

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

Tulack wrote:
Why do you think luminosity mask wouldn't work?

it's not local... it takes only certain "tonalities" in the luminosity....and not the deviation of medians...hence local

Herman
www.hermanvangestel.com

Jul 30 14 01:12 pm Link

Photographer

Tulack

Posts: 836

Albuquerque, New Mexico, US

If you convert it to TSL would it work?

Jul 30 14 02:28 pm Link

Photographer

NothingIsRealButTheGirl

Posts: 35726

Los Angeles, California, US

I guess you want to blur it somewhat and subtract the blur from the original to get the differences.

I'm never sure what to do with negative numbers in Photoshop, though. In Nuke or Shake you subtract something and if parts of the image go negative it's no big deal. You can't directly view it, but the data is still fine.

This whole "50% gray is like zero" thing seems like a kludge to me.

Jul 30 14 02:58 pm Link

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

NothingIsRealButTheGirl wrote:
I guess you want to blur it somewhat and subtract the blur from the original to get the differences.

I'm never sure what to do with negative numbers in Photoshop, though. In Nuke or Shake you subtract something and if parts of the image go negative it's no big deal. You can't directly view it, but the data is still fine.

This whole "50% gray is like zero" thing seems like a kludge to me.

so far i got already..you do that through differences....once you have your 50% mask you can use levels to extract the mask....i guess it must be possible with "Apply Image" as well....must be faster

Herman
www.hermanvangestel.com

Jul 30 14 03:04 pm Link

Photographer

NothingIsRealButTheGirl

Posts: 35726

Los Angeles, California, US

As an experiment I looked at differences of several high pass layers, and I also looked at differences between an original image and one with Nik > Color Efex Pro > Pro Contrast applied, since that seems to do the local contrast thing already, so why not just capture the differences from it (assuming one has it installed)

Jul 30 14 03:09 pm Link

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

creating the masks is possible...just USM generates too weak local contrast detection....any idea?

I tried with median....not much luck there too...

Jul 30 14 04:39 pm Link

Photographer

NothingIsRealButTheGirl

Posts: 35726

Los Angeles, California, US

Herman van Gestel wrote:
creating the masks is possible...just USM generates too weak local contrast detection....any idea?

I tried with median....not much luck there too...

I guess I don't totally follow. If you have weak deviations from a gray base then can levels accentuate them to what you need?

Jul 30 14 04:46 pm Link

Photographer

Jakov Markovic

Posts: 1128

Belgrade, Central Serbia, Serbia

Select channels. Open channels palette and Command+click on a channel (even RGB one) to select the channel, then you can enhance the mask with curves or dodge and burn tools.

Jul 30 14 10:19 pm Link

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

Jakov Markovic wrote:
Select channels. Open channels palette and Command+click on a channel (even RGB one) to select the channel, then you can enhance the mask with curves or dodge and burn tools.

that doesn't do local contrast (with a radius of 60 pixels for example..), that's based on luminosity instead of the difference of luminosity...

Jul 31 14 02:25 am Link

Photographer

Herman van Gestel

Posts: 2266

Amsterdam, Noord-Holland, Netherlands

NothingIsRealButTheGirl wrote:

I guess I don't totally follow. If you have weak deviations from a gray base then can levels accentuate them to what you need?

you are right there...but with USM it takes also quite some of the "even"areas...

the solution to that is to work only withe a blurred image (pref surface blur) ...

...but we're getting there smile

Herman
www.hermanvangestel.com

Jul 31 14 02:26 am Link