RetouchPRO

Go Back   RetouchPRO > Technique > Photo Retouching
Register Blogs FAQ Site Nav Search Today's Posts Mark Forums Read


Photo Retouching "Improving" photos, post-production, correction, etc.

Frequency separation math

Reply
 
Thread Tools
  #21  
Old 09-21-2014, 05:57 AM
nemrob nemrob is offline
Junior Member
 
Join Date: Sep 2014
Posts: 14
Re: Frequency separation math

Quote:
Originally Posted by byRo View Post
Well actually...
We were already talking about frequency separation RIGHT HERE at RetouchPRO quite a long time before this!
Yes, I knew it, but I didn't find the math behind the separation, that's why I started this thread. I know it is of limited intrest (it works if you don't know anything about math), so I wanted to create a topic apart from the aforementioned one.

Thank you very much, byRo, for the action.
Reply With Quote top
  #22  
Old 09-25-2014, 12:58 AM
yagovic yagovic is offline
Junior Member
 
Join Date: Nov 2010
Posts: 4
Re: Frequency separation math

thank you. this thread really help me
Reply With Quote top
  #23  
Old 09-25-2014, 06:05 AM
nemrob nemrob is offline
Junior Member
 
Join Date: Sep 2014
Posts: 14
Re: Frequency separation math

I tried to make a 3 layer split, but it's just not working. I make the 1st split: I have a Low layer with 30 px Gaussian and a Middle layer after apply image (Subtract, 128, 2). Then I copy the Middle, rename to High. Blur the Middle to 10 px and Apply image to the High. Then I set Mid and High to Linear light.
The result is not the original.
Where am I scr.ing it up? I tried many ways, but it's just not working.
Reply With Quote top
  #24  
Old 09-25-2014, 08:15 AM
skoobey skoobey is offline
Senior Member
 
Join Date: Jan 2012
Posts: 1,376
Re: Frequency separation math

Why not read what I wrote about splitting it into 3?

There is NO MIDDLE.

There is low and the high. In order to split further you DO NOT deal with the high.

So, first split.

Duplicate the layer two times. One is the low, other is high, merged file in the bottom.

Blur the low. Go to high and apply from the low via Substract, just like you do.

Now, for an additional frequency you duplicate the low.

Blur the original low, one on the bottom, no matter the diameter of the blur it can be larger or smaller than the first one.

Now, go to the middle layer (one you made by copying the low) and go to apply image, apply from the original low.

Now you have to highs and one low.

That's the way you do it. As the low is always non transparent, and highs are transparent in nature due to the blending modes. You can do as many highs as you can, but there can only be one low.
Reply With Quote top
  #25  
Old 09-25-2014, 10:12 AM
nemrob nemrob is offline
Junior Member
 
Join Date: Sep 2014
Posts: 14
Re: Frequency separation math

Quote:
Originally Posted by skoobey View Post
Why not read what I wrote about splitting it into 3?
Don't be so touchy. What makes you think I didn't read? I didn't understand, that's another thing

Besides you wrote: "Just repeat the process on the top, or the bottom layer." So I've gone upwards to the high.

This way I got it, but the result is different than the original, a bit more blurred than the original. Not too big, just like with the HP filter. Is it normal, or it should be exactly the same?
ByRo put up earlier an action for tri-split, but it has the same blurriness. So I guess it won't be super precise on 3 layers.

Thank you for your help.
Reply With Quote top
  #26  
Old 09-25-2014, 10:51 AM
skoobey skoobey is offline
Senior Member
 
Join Date: Jan 2012
Posts: 1,376
Re: Frequency separation math

Looks the same to me.

About my previous explanation... I agree, it sucked. Well, you see, there are other ways to make the split, including splitting the color and luminosity and joining them to the particular channels. Now, this means that your split cloning and healing only works on luminosity, color is unaffected, and you can change the radius whenever you want, as it's attached to a smart object. Anyway, it's a really inefficient way to work. And you can always just make two splits, and crop the results of the less used ones to save on the file size.

And no split is perfectly accurate, if you set the spilt to difference blend mode and then make a threshold adjustment layer above, you'll see that there is at least two levels of differentiation.

It's always what you gain, and what you loose. You loose realism with retouching, but you gain purity in terms of idea to reality. The benefits of a split outweigh it's downfalls.

What do you need this double split for anyway?

Last edited by skoobey; 09-25-2014 at 10:56 AM.
Reply With Quote top
  #27  
Old 09-25-2014, 11:04 AM
nemrob nemrob is offline
Junior Member
 
Join Date: Sep 2014
Posts: 14
Re: Frequency separation math

OK, thank you.

No special need, I'm retouching mostly as hobby and I want to know my tools better. I heard and read from retouchers that 3 levels of split is the best, but I rarely used it. I had an action for it, but I lost it, so I wanted to make a new one for myself.
Also as I wrote earlier, if I just do as others do (the split) it feels like having someone else's clothes on, so I decided to dug deeper. And it's really interesting.
Reply With Quote top
  #28  
Old 09-26-2014, 01:38 AM
klev klev is offline
Senior Member
 
Join Date: Jun 2010
Posts: 1,109
Re: Frequency separation math

Quote:
Originally Posted by nemrob View Post
I heard and read from retouchers that 3 levels of split is the best, but I rarely used it.
Make sure you trust the quality of advice. The uses that are typically suggested on here mean a lot of file bloat, and I have demonstrated the weakness of the gaussian blur filter with its inverted haloing. That doesn't change between 8 and 16 bits. 16 bit operations are just less likely to incur banding. When people complain about issues painting on something such as a low frequency layer, the issue is that they often change the gradient, or rather the average rate of change between neighboring pixels within a given radius. If it becomes too steep, you end up with banding regardless of your bit depth.
Reply With Quote top
  #29  
Old 09-26-2014, 05:13 AM
skoobey skoobey is offline
Senior Member
 
Join Date: Jan 2012
Posts: 1,376
Re: Frequency separation math

Quote:
Originally Posted by klev View Post
Make sure you trust the quality of advice. The uses that are typically suggested on here mean a lot of file bloat, and I have demonstrated the weakness of the gaussian blur filter with its inverted haloing. That doesn't change between 8 and 16 bits. 16 bit operations are just less likely to incur banding. When people complain about issues painting on something such as a low frequency layer, the issue is that they often change the gradient, or rather the average rate of change between neighboring pixels within a given radius. If it becomes too steep, you end up with banding regardless of your bit depth.
Absolutely. It's about knowing what you want the image to look like before you start working on it; knowing where it can go, and where it can't.

But it's typical hype. Liek in music industry, when masses started calling every sound autotune, by the name of an iphone app. lol
Reply With Quote top
  #30  
Old 10-09-2014, 11:14 AM
Chaikovsky007 Chaikovsky007 is offline
Junior Member
 
Join Date: Jan 2014
Location: London
Posts: 7
Re: Frequency separation math

Great info here!
I am gonna print this whole thing and make wallpaper in my room
Reply With Quote top
Reply

  RetouchPRO > Technique > Photo Retouching


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Frequency separation: Really non-destructive? baryon Photo Retouching 27 10-09-2013 05:19 PM
question about frequency separation method Caravaggio Photo Retouching 12 07-17-2013 12:21 AM
Skin retouching better than frequency separation gwlaw99 Photo Retouching 6 12-16-2012 03:34 AM
"asymmetric" frequency separation? drode Photo Retouching 12 12-25-2011 02:55 AM
Extended frequency separation fraiseap Photo Retouching 16 12-31-2010 01:33 PM


All times are GMT -6. The time now is 12:27 AM.


Powered by vBulletin® Version 3.8.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
SEO by vBSEO ©2011, Crawlability, Inc.
Copyright © 2016 Doug Nelson. All Rights Reserved