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
  #101  
Old 11-02-2010, 02:39 PM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Back to reality

Quote:
Originally Posted by ShadowLight View Post
When you actually implement the two separate techniques you do get this result: (see attached pic)
as you see contrast -50 does not give you a 127/8 grey, in this case is more like 154/5.
Put that file up for us - let me show everyone something. Please.
Quote:
Originally Posted by ShadowLight View Post
Second, SkinSmooth actually uses "Fill" and there is no clipping whatsoever.
Makes no difference - as I said above, Fill and Opacity create the same result here, in both cases clipped.
Quote:
Originally Posted by ShadowLight View Post
BTW, Congrats on the upcoming delivery. You should get some sleep now when you can.
I said 'deployment', not 'delivery'.... it's not exactly a congratulatory moment, but I thank you for the intent just the same.
Reply With Quote top
  #102  
Old 11-02-2010, 03:47 PM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Rather than wait to (maybe) get the original file shown above, I've gone ahead and recreated it, down to the emulation of the lum "shift" discussed. The PSD below takes those layers, duplicates them, and then applies a HighPass filtration to each (the next step in anyone's IHP process). There is a 50% gray layer underneath to show you what each does and does not retain. You be the judge of which retains more detail. -50 Contrast on the left. -50 Fill on the right. Be sure to select each layer individually and look at its histogram (Ctrl or Cmd+Click on the layer) - it's brutal.

PSD

Please let me know if there are questions - preferably publicly vs. private - I like to share knowledge.

Sorry about the preview size - I don't have as large a monitor as ShadowLight does .
Attached Images
File Type: jpg HPDemoPt2.jpg (44.8 KB, 33 views)
Reply With Quote top
  #103  
Old 11-02-2010, 03:55 PM
mistermonday's Avatar
mistermonday mistermonday is offline
Moderator
Patron
 
Join Date: Dec 2005
Posts: 3,028
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by bakerser View Post
Rather than wait to (maybe) get the original file shown above, I've gone ahead and recreated it, down to the emulation of the lum "shift" discussed. The PSD below takes those layers, duplicates them, and then applies a HighPass filtration to each (the next step in anyone's IHP process). There is a 50% gray layer underneath to show you what each does and does not retain. You be the judge of which retains more detail. -50 Contrast on the left. -50 Fill on the right. Be sure to select each layer individually and look at its histogram (Ctrl or Cmd+Click on the layer) - it's brutal.
+1

Quote:
Please let me know if there are questions - preferably publicly vs. private - I like to share knowledge.
Yes, one of the main purposes of the forum is to learn and share knowledge. This thread has initiated some interesting questions and many members would like to learn the answers and achieve satisfactory closure.

Regards, Murray
Reply With Quote top
  #104  
Old 11-02-2010, 05:27 PM
ShadowLight ShadowLight is offline
Senior Member
 
Join Date: Oct 2009
Posts: 397
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by mistermonday View Post
+1



Yes, one of the main purposes of the forum is to learn and share knowledge. This thread has initiated some interesting questions and many members would like to learn the answers and achieve satisfactory closure.

Regards, Murray
-2

I don't mind sharing knowledge, however proliferation of misinformation is derogatory to the put effort.

- You have not only not tested what the application does, nor its functionality.
- You have not asked any questions, just made assumptions without even testing.
- You do not answer for the problems I showed your implementation creates.
- The information you put forward is not relevant to the functionality of the application.
- You do not offer a better solution.
- And do not offer credible tests.

I think this is the definition of "trolling".

If you test something, test what the application does, not something else.

I intend to answer to question of people with regard to the application in this thread, not to correct disillusioned conclusions based on erroneous assumptions.

btw... my mistake then... as you were doing the deployment a few weeks ago, so I "assumed" it was a real baby.
Reply With Quote top
  #105  
Old 11-02-2010, 05:54 PM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by ShadowLight View Post
- You have not only not tested what the application does, nor its functionality.
Please let me know how the pre-Smart Filter layer which your application uses is different than the one in the test file posted above. I wouldn't want to perform an invalid test.

And I have seen how the program operates, and wouldn't be making the efforts I have to correct this misunderstanding if I didn't think it could be an invaluable workflow tool for retouchers at all levels. But it can only hold that status with a good methodology.
Quote:
Originally Posted by ShadowLight View Post
- You have not asked any questions, just made assumptions without even testing.
I've tested constantly here as I did on MM. I can't help if you don't like the results, only to try to explain it to you as to make the tool better.
Quote:
Originally Posted by ShadowLight View Post
- You do not answer for the problems I showed your implementation creates.
Which problems were those? Please don't ignore the other prompts to answer this one alone, but I am curious whether you mean something other than the 1/32k level 'noise' (which I assure you no one can see).
Quote:
Originally Posted by ShadowLight View Post
- The information you put forward is not relevant to the functionality of the application.
How is that? If you want to answer questions, please explain to me how you're going to be able to be able to nullify data whose information you've lost to clipping. Please explain the histogram spikes at 0 and 255 in the example file while you're at it if you are sure that this is irrelevant.
Quote:
Originally Posted by ShadowLight View Post
- You do not offer a better solution.
I've offered no fewer than three options for how you might do this - all of which are very similar to what you're doing now, 2/3 are compatible with a nondestructive workflow, and all have no or negligible impact on file size.
Quote:
Originally Posted by ShadowLight View Post
- And do not offer credible tests.
While you've much ballyhoo'd my 'failure' to detect the difference in Brightness-Contrast versions (despite my never having used the Adjustment Layer version before doing so to try to help you devise a better implementation for your application), in fact I have some of the better testing methodologies around when I look into IQ. Unless you already knew that PS's quick resize algorithm interferes with sampling tool results?
Quote:
Originally Posted by ShadowLight View Post
If you test something, test what the application does, not something else.
As above, I quite enjoy what it could be capable of and if I didn't have a world of respect for your programming skill I wouldn't have made the effort to begin with.
Quote:
Originally Posted by ShadowLight View Post
btw... my mistake then... as you were doing the deployment a few weeks ago, so I "assumed" it was a real baby.
Watch the news a little more; I think you'll get the idea of what a deployment actually means.
Reply With Quote top
  #106  
Old 11-02-2010, 06:52 PM
ShadowLight ShadowLight is offline
Senior Member
 
Join Date: Oct 2009
Posts: 397
Re: Automatic Skin Smoothing.

and this is what I tried to avoid...

Quote:
Originally Posted by bakerser View Post
Which problems were those? Please don't ignore the other prompts to answer this one alone, but I am curious whether you mean something other than the 1/32k level 'noise' (which I assure you no one can see).
yes, the shift of the lum. values

the contrast - "smart object" layer has only the contrast -50 applied, noting before, nothing after... there isn't anything else "in it"
(this is what happens when you don't use an adjustment layer)

another problem with contrast -50 will be the dynamic content, if it were dynamically adjusted as direct application as soon as you change the background your adjustments on the sking, the one you have already done will change too

if it were as an adjustment layer it will give the proper corrections, however the structure and dynamics becomes a problem.

I'm attaching your example, however with the intent of proper use.
on the left is Contrast -50 (with adjustment layer)
on the right is Opacity 50%

both layers show the result after running a HP over them (as smart objects).
so the "worst" thing you may get using "Alpha control" is some detail showing back when you boost the HP towards (over) the max useful range for portrait corrections.

I'm showing it for the sake of people who wanted to see the difference, and as far as I see it, the difference will not be detectable in real-use cases, and requires contrast -50 to be used as an adjustment layer.

if that doesn't clear things up for you... I don't know what will.


Edit:
FYI: the max usable range SkinSmooth calculates for this test image is gb:1.5 / hp :5.3
Attached Images
File Type: jpg HPDemo.jpg (97.1 KB, 24 views)

Last edited by ShadowLight; 11-02-2010 at 07:10 PM.
Reply With Quote top
  #107  
Old 11-02-2010, 07:32 PM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Look, I'm trying to help here. I really really know WTF I'm talking about. Ask around. So I'm going to ask a few questions again and hope that this time I get more than one of them answered.
Quote:
Originally Posted by ShadowLight View Post
yes, the shift of the lum. values
Please explain to us (and feel free to use math) how that matters when the image will always be run through the HP filter. Don't show a picture of two different grays - tell us why those shades of gray matter at all to the result of the HP. Or better, show us a PSD with the results.
Quote:
Originally Posted by ShadowLight View Post
the contrast - "smart object" layer has only the contrast -50 applied, noting before, nothing after... there isn't anything else "in it"
(this is what happens when you don't use an adjustment layer)
OK, so my test was valid, then. I'm glad you're accepting at least some of my work at explaining this.
Quote:
Originally Posted by ShadowLight View Post
another problem with contrast -50 will be the dynamic content, if it were dynamically adjusted as direct application as soon as you change the background your adjustments on the sking, the one you have already done will change too
How about using Filter->Other->Custom with:

0 0 0 0 0
0 0 0 0 0
0 0 1 0 0
0 0 0 0 0
0 0 0 0 0

Scale 2 Offset 64 ?

...

and that's #4 btw.
Quote:
Originally Posted by ShadowLight View Post
both layers show the result after running a HP over them (as smart objects).
so the "worst" thing you may get using "Alpha control" is some detail showing back when you boost the HP towards (over) the max useful range for portrait corrections.
To quote you, reality isn't that simple unfortunately. Reality is a nasty equation relating total contrast over the local and all sizes in between, and it is completely impossible as an author to know what the user is going to present an automation system like this with. Ironically, the more contrasty the image starts (and therefore, the more likely it is to need a clip-proof algorithm), the more likely it is that the user is going to want to use something like this to clean the skin up. Which means a good bit of your base actually benefits from doing it right in the first place.
Quote:
Originally Posted by ShadowLight View Post
I'm showing it for the sake of people who wanted to see the difference, and as far as I see it, the difference will not be detectable in real-use cases, and requires contrast -50 to be used as an adjustment layer.
As a pointer, I think they might appreciate it more if you actually posted some of the PSDs at some point for them / us to evaluate independently. It's the awful truth of PS geeks that we know just how little you can trust a JPEG .
Reply With Quote top
  #108  
Old 11-02-2010, 07:57 PM
ShadowLight ShadowLight is offline
Senior Member
 
Join Date: Oct 2009
Posts: 397
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by bakerser View Post

OK, so my test was valid, then. I'm glad you're accepting at least some of my work at explaining this.
I have no idea on what world you live in...

actually I showed you exactly that your test design has failed. Just like applying Contrast -50 test you failed to notice the lum. shift for two years.... looks like you are not really good in this

tip: in the test I showed you, the conceptual differences are clear and that the result, unless you want to remove 90% of the image is the same with using Alpha, and Contrast -50 (as an adjustment layer)
(exactly the opposite faulty conclusion you made with making invalid test)

anyway... try to "get it" there's nothing more to be said.

just "stop trolling"...
Reply With Quote top
  #109  
Old 11-02-2010, 08:53 PM
Flashtones Flashtones is online now
Senior Member
 
Join Date: Apr 2009
Posts: 956
Re: Automatic Skin Smoothing.

I've been enjoying the discussion. Thanks guys.
Reply With Quote top
  #110  
Old 11-02-2010, 08:58 PM
bakerser bakerser is offline
Member
 
Join Date: Jan 2008
Posts: 91
Re: Automatic Skin Smoothing.

Quote:
Originally Posted by ShadowLight View Post
actually I showed you exactly that your test design has failed. Just like applying Contrast -50 test you failed to notice the lum. shift for two years.... looks like you are not really good in this
Well, it's actually been just over a year since we determined that the issue with HP is the contrast piece, but I get where you were going hyperbolicly.

More to the point, please let us know how and when you want to address the fact that it's not actually a luminance shift, as the mean value of the image is maintained through the course of the operation which you reject, while in fact it's the version you favor which imparts the only "shift" of the image mean.

Further let us know if / when you'll be demonstrating (with a PSD please!) how that "shift" makes a difference to the HP result on which the rest of your algorithm flow depends. I keep asking you to explain the math, and there are obviously others here who would like to see it as well, so please let us see it. As has been stated and repeated numerous times, we're all here to learn.
Quote:
Originally Posted by ShadowLight View Post
tip: in the test I showed you, the conceptual differences are clear and that the result, unless you want to remove 90% of the image is the same with using Alpha, and Contrast -50 (as an adjustment layer)
Well, the JPEG which you showed us appears to demonstrate a similarity of results, but you continue to avoid showing us a PSD. It makes one wonder, why?

And of course, if you'd like to respond to any of the other six questions which you've ignored above (including one honestly rather sneaky SO workaround), feel free. I welcome your responses and won't hold your answering things out of order against you.

Quote:
Originally Posted by Flashtones View Post
I've been enjoying the discussion. Thanks guys.
I find it to be like watching one of the movies from MST3K. I just wish there were popcorn.
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 08:29 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