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.

Automatic Skin Smoothing.

Reply
 
Thread Tools
  #111  
Old 11-02-2010, 09:35 PM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

So, for those playing at home, I've put up another PSD (linked at the bottom again) to dispel some of this hoopla about "shifted luminosity". What follows is a description of the layers you'll find in the PSD and then a discussion of their import.
  1. The bottom layer is the same image I've been using - a headshot of the lovely Bianca and serves only as your reference.
  2. The second and third layers (counting up) are the "gold standard" of Gaussian Blur + Apply Image HF, grouped together for easy identification. The second layer is a copy of the Original, Gaussian Blur'd at a radius of 50 pixels, while the third layer is the high-frequency component derived in a way I'm sure anyone reading this already knows .
  3. The fourth layer is a Smart Object. It has not had its opacity adjusted, an adjustment layer added, etc. It is 'pure'. To it I've applied two Smart Filters - the first, a variation on the Custom filter which I suggested to ShadowLight above; the second is a High Pass filter run @ 50px.
  4. The fifth layer is almost exactly like the fourth, except that instead of applying an offset of 0 in the Custom filter dialog, I've applied an offset of 125, nearly (but not quite) clipping the brightest highlight. And of course, it has the same 50px HP filter applied to it as well.

So what does that mean? Take a look through the PSD. Alternately switch the blend mode of one and then the other of the Smart Objects to Difference to see that the HP result is identical to the Apply Image version. Pretty cool, isn't it? It turns out that I haven't been full of it all along - that extraction of the upper frequency range from an image doesn't care about the starting mean value of that image - it's the difference of those values across space which defines the result. Whichever Brightness / Contrast implementation you use, it will have no impact on your ability to do a proper frequency separation / IHP / etc.

Edit: Don't forget that you can turn off the visibility of each Smart Filter independently, allowing you to enable or disable the HP filter and see that the images being fed to it are indeed very different.

...

Also feel free to mess with the Smart Object itself, realizing that this means of contrast-halving allows total freedom of image manipulation. It has no logistical ties whatsoever to restrict it (much like other solutions which have been offered up and as quickly rejected - this is just the one I'm showing off right now because it's new and shiny ).

PSD

For those in a big hurry, the image below shows the starting image, the two contrast-halved variations of different 'brightness'es, their mean values, and the HP result for each. While you can't see that there's a 0/32k difference just from the JPEG, it should serve as a quick reference for those not able to open a PSD at the moment.
Attached Images
File Type: jpg HPDemoPt3.jpg (87.7 KB, 19 views)

Last edited by bakerser; 11-02-2010 at 10:05 PM. Reason: Added Smart Filter reminder.
Reply With Quote top
  #112  
Old 11-03-2010, 03:32 AM
ShadowLight ShadowLight is offline
Senior Member
 
Join Date: Oct 2009
Posts: 397
Re: Automatic Skin Smoothing.

again... you are not testing the technique...
No matter how you spin it, when changing the background affects the portrait corrections it is not usable implementation.
(which is what happens with the Contrast -50 technique).

again... when you construct your test wrong you end up with the wrong conclusion.


The bottom line is that the "problem" with the alpha implementation is not relevant at all, since it occurs in values that are not usable anyway. (and when you limit the corrections as the implementation requires and let high-frequency in, even in those unusable values the negative effects are nullified)

again... for the example you use, the maximum usable HP value is less than 6.

I suggest you spend more time modelling your tests better, maybe this will help you avoid the wrong conclusions you get to.


Murray, I have no idea how "this" so far can be seen as "Educational" to anyone (if we don't count Sean)... but so be it... class dismissed

Edit: I added a comparison of the two techniques.
- using the calculated maximum usable range
- mask unchecked
- second row is "Contrast-boost for visibility".
You figure which technique is which.
Attached Images
File Type: jpg Compared(alpha-contrast).jpg (97.2 KB, 15 views)
File Type: jpg Compared2(alpha-contrast).jpg (99.2 KB, 11 views)

Last edited by ShadowLight; 11-03-2010 at 04:38 AM.
Reply With Quote top
  #113  
Old 11-03-2010, 05:10 AM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by ShadowLight View Post
again... you are not testing the technique...
Please, write out the PS steps which are necessary to duplicate "the technique" so that I may post a PSD which you might actually download before commenting on next time.
Quote:
Originally Posted by ShadowLight View Post
No matter how you spin it, when changing the background affects the portrait corrections it is not usable implementation.
(which is what happens with the Contrast -50 technique).
And yet the PSD above demonstrates this to be entirely untrue. Of course, it's equally obvious to anyone who has a clue about the math involved, but I'd hoped you would have at least benefited from the visual reference.
Quote:
Originally Posted by ShadowLight View Post
The bottom line is that the "problem" with the alpha implementation is not relevant at all, since it occurs in values that are not usable anyway. (and when you limit the corrections as the implementation requires and let high-frequency in, even in those unusable values the negative effects are nullified)
Don't be confused by the values used in my demonstration. As I explained to you earlier, it's a matter of total contrast over distance - this problem happens at all radii, not just ones which I selected for the tests images I've posted - and will happen to a greater and lesser degree in all images.

Further, as explained (and ignored by you) earlier in this conversation, you can't remove detail which wasn't retained (and eliminating retained detail is what IHP does). You can't get back detail which you've allowed to clip. Please go back and read my explanation of why high-contrast shooters are both more likely to need a utility like this as well to need it to actually work right.

"letting HF in" doesn't fix anything - all it does is to try to hide the problem, like naively blurring an image with clipped highlights in hopes of hiding the fact that there's no detail. That doesn't make for a good result.
Quote:
Originally Posted by ShadowLight View Post
again... for the example you use, the maximum usable HP value is less than 6.
You've gotten very good at throwing arbitrary requirements in here. Just the same, please read above - the problems with your methodology can and do occur at all radii. It's a design flaw, not a leveraged capability.
Quote:
Originally Posted by ShadowLight View Post
I suggest you spend more time modelling your tests better, maybe this will help you avoid the wrong conclusions you get to.
And I'll suggest that you actually bother to download the demonstration PSDs before digging yourself in any deeper.
Quote:
Originally Posted by ShadowLight View Post
Edit: I added a comparison of the two techniques.
- using the calculated maximum usable range
- mask unchecked
- second row is "Contrast-boost for visibility".
You figure which technique is which.
How is that supposed to be helpful to anyone when they can't actually analyze what you've done?
Reply With Quote top
  #114  
Old 11-03-2010, 05:47 AM
ShadowLight ShadowLight is offline
Senior Member
 
Join Date: Oct 2009
Posts: 397
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by bakerser View Post
How is that supposed to be helpful to anyone when they can't actually analyze what you've done?
how about noticing that nigher has the "clipping" you talk about...

if you can't tell the difference it should be enough for everyone

thanks
Reply With Quote top
  #115  
Old 11-03-2010, 06:05 AM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by ShadowLight View Post
how about noticing that nigher has the "clipping" you talk about...

if you can't tell the difference it should be enough for everyone

thanks
First of all, I'm not special, and everyone should have the chance to check what each of us is asserting [referring to your emphasis on myself].

But more importantly, no one can check it while it's just a JPEG - we know neither whether you actually used valid methodology nor whether the results are as you claim.

You're of course free to continue answering as few prompts as you like and to continue offering no meaningful (or veritable) evidence of your position, but I'm being honest when I say that I have a world of respect for your coding ability and hate to see you keep digging yourself into this hole of misinformation. You little app could be one of the great successes of internet PS geekery... but you'll have to accept that this user base is one which is capable of providing valuable feedback to you on more than just the UI. We know what we're talking about.
Reply With Quote top
  #116  
Old 11-03-2010, 06:47 AM
rider_get6 rider_get6 is offline
Junior Member
 
Join Date: Nov 2009
Location: indonesia
Posts: 1
Re: Automatic Skin Smoothing.

does it work on CS3??
Reply With Quote top
  #117  
Old 11-03-2010, 06:52 AM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by rider_get6 View Post
does it work on CS3??
Unfortunately it can't - support for Flash panels wasn't introduced until CS4, do you have to stick with more manual approaches (though actions & scripts remain an option).
Reply With Quote top
  #118  
Old 11-03-2010, 07:07 AM
ShadowLight ShadowLight is offline
Senior Member
 
Join Date: Oct 2009
Posts: 397
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by bakerser View Post
But more importantly, no one can check it while it's just a JPEG - we know neither whether you actually used valid methodology nor whether the results are as you claim.
The plugin is free, download it, and test it yourself.
Set a similar setup for gb/hp with Contrast -50 approach and compare.

can't be simpler than that,
and you don't have to take my word for it if you don't want to

rider_get6, he is correct.
as far as I know CS3 does not support that kind of interfaces.
Reply With Quote top
  #119  
Old 11-03-2010, 07:56 AM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by ShadowLight View Post
The plugin is free, download it, and test it yourself.
Set a similar setup for gb/hp with Contrast -50 approach and compare.
See, but I have. And I've been sharing my test PSDs so that everyone can verify my findings. You keep rejecting these and offering only JPEGs in response, giving us nothing but bravado to go on. No explanation. No evidence. Just claims alongside a very selective pragma for what you'll even deem worthy of response.

Out of morbid curiosity, did you even try the Custom filter option?
Reply With Quote top
  #120  
Old 11-03-2010, 08:06 AM
ShadowLight ShadowLight is offline
Senior Member
 
Join Date: Oct 2009
Posts: 397
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by bakerser View Post
See, but I have. And I've been sharing my test PSDs so that everyone can verify my findings. You keep rejecting these and offering only JPEGs in response, giving us nothing but bravado to go on. No explanation. No evidence. Just claims alongside a very selective pragma for what you'll even deem worthy of response.

Out of morbid curiosity, did you even try the Custom filter option?
the only problem in your "test" is that you didn't make the comparison of what the application does.

I didn't lie about the results, but couldn't attach the PSD directly.
How hard is it to press "Create" and move the sliders to "blotch" and "variable"?

do it and make a contrast -50 (or whatever you like to compare it with), set the gb/hp as the one in the SkinSmooth layer and compare.

I showed you the result, if you think I'm making it up, the easiest way is to press "Create" in the menu of SkinSmooth and see for yourself.


I didn't test the custom filter (it's hard to use it in dynamic implementation), but if it works as well as contrast -50 (as adj. layer) it should be viable alternative.
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
Glamour retouch challenge here!!!! superkoax Photo Retouching 332 02-08-2011 09:52 PM
Automatic Skin Tone Retouching - Seeking feedback Pixarra Photo Retouching 15 11-09-2010 01:38 AM
Skin Smoothing Technique (Again) Novi Photo Retouching 10 10-16-2008 04:25 PM
need help doing pageant eyelashes and skin smoothing w/ photoshop cs sandylavallie Photo Retouching 3 05-06-2005 08:53 AM
skin smoothing action john_opitz Photo Retouching 10 08-21-2002 10:48 PM


All times are GMT -6. The time now is 11:38 PM.


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