Saturation Masks Log Out | Topics | Search
Moderators | Register | Edit Profile

Digital Light & Color Message Boards » Picture Window Support » Archive through July 14, 2004 » Saturation Masks « Previous Next »

Author Message
Top of pagePrevious messageNext messageBottom of page Link to this message

Ray Milewski
New member
Username: Rmilewsk

Post Number: 83
Registered: 01-2004
Posted on Monday, June 21, 2004 - 09:49 pm:   Edit Post Delete Post

I just saw this article on luminous landscape. It looks like a great technique. Does anyone know if this is possible in PWP?


http://www.luminous-landscape.com/tutorials/restore-clipped.shtml
Top of pagePrevious messageNext messageBottom of page Link to this message

Bob Walker
New member
Username: Rwalkernm

Post Number: 1
Registered: 06-2004
Posted on Monday, June 21, 2004 - 10:35 pm:   Edit Post Delete Post

Not only is this possible, but it's much easier with PWP than with PS. Just extract the saturation channel (either HSV or HSL sat) and use it as a mask on color curves, boosting just the saturation of those parts of the image that are low sat (white slider = 0, black slider > 0). Makes washed out skies really nice and blue.
Top of pagePrevious messageNext messageBottom of page Link to this message

Richard de Garis
New member
Username: Rdegaris

Post Number: 32
Registered: 10-2002
Posted on Tuesday, June 22, 2004 - 03:25 am:   Edit Post Delete Post

I was thinking about this too and I have an alternate suggestion to try out; simply use the Transformation - Color - Curves tool to selectively alter the saturation for particular portions of the saturation histogram.

The principle rationale behind this suggestion - and I may be totally wrong - is that the transformation is made at full i.e. 16-bit colour depth. The extracted saturation channel unfortunately has to be reduced to 8-bit B&W before it can be applied as a mask.

Now I've done absolutely no testing of these two techniques, so I can only speculate that if - and I stress if - saturation banding is evident in a clipped channel image - like the example on Luminous Landscape - then one may have more success in counteracting it by manipulating the image at full colour depth. Anyone got an example to try this out on?
Top of pagePrevious messageNext messageBottom of page Link to this message

Richard de Garis
New member
Username: Rdegaris

Post Number: 33
Registered: 10-2002
Posted on Tuesday, June 22, 2004 - 07:33 am:   Edit Post Delete Post

Okay, I've had a bit of a play and while Color - Curves can be a very powerful tool, I actually preferred the method suggested by Bob. As to my fears about colour depth, I couldn't discern any adverse effects in the examples I tried, although they were hardly stressing cases (I'll have to see if I have a suitably mis-exposed slide, that I haven't already thrown away, to scan in and experiment on).

In any case, Bob's point is well made - Saturation Masking appears to be a royal pain to achieve in Photoshop, yet a simple breeze in PWP.
Top of pagePrevious messageNext messageBottom of page Link to this message

Kirk Benson
New member
Username: Valerie01

Post Number: 24
Registered: 03-2004
Posted on Tuesday, June 22, 2004 - 01:07 pm:   Edit Post Delete Post

Just to confirm I have Bob's technique correct:

1) Use Extract Color|Channel transform selecting the HSL Saturation or HSV Saturation option. These do not seem to be the same. HSL mask seems to be preferable, but why is S-channel different in the two views?

2) Convert to 16 bit so as to be useable as a mask.

3) Use Color|Curves with the mask

I believe I'm correct that it is not necessary to invert the mask, as the curve transform's sliders take care of whether the black or white mask bits are being sampled.

Top of pagePrevious messageNext messageBottom of page Link to this message

Jonathan Sachs
Board Administrator
Username: Admin

Post Number: 1798
Registered: 08-2002
Posted on Tuesday, June 22, 2004 - 01:46 pm:   Edit Post Delete Post

HSV S and HSL S are different because the color spaces are different. HSV has its maximum saturation when the V component is 100%; HSL has its maximum saturation when the L component is 50%.

Only 8-bit B&W images can be used as masks.
Jonathan Sachs
Digital Light & Color
Top of pagePrevious messageNext messageBottom of page Link to this message

den
Unregistered guest
Posted on Wednesday, June 23, 2004 - 09:56 am:   Edit Post Delete Post

I have been trying to duplicate the saturation masks of the flower and boats shown in the Luminous tutorial using PWP's HSV,S and HSL,S channel extractions starting with the shown unenhanced images. There is a significant difference between the tutotial derived masks and PWP's. Does anyone know why?
Top of pagePrevious messageNext messageBottom of page Link to this message

George McLelland
New member
Username: Georgemc

Post Number: 10
Registered: 09-2003
Posted on Wednesday, June 23, 2004 - 12:14 pm:   Edit Post Delete Post

Can someone please tell me how to create a saturation mask OR tell me where to find the instructions for this.
Thanks, George
Top of pagePrevious messageNext messageBottom of page Link to this message

Jonathan Sachs
Board Administrator
Username: Admin

Post Number: 1803
Registered: 08-2002
Posted on Wednesday, June 23, 2004 - 12:23 pm:   Edit Post Delete Post

Transformation/Color/Extract Channel -- select HSV S or HSL S.
Jonathan Sachs
Digital Light & Color
Top of pagePrevious messageNext messageBottom of page Link to this message

Bob Walker
Unregistered guest
Posted on Wednesday, June 23, 2004 - 01:03 pm:   Edit Post Delete Post

Responding to den's question. I also tried duplicating the LL tutorial images using PWP, and the masks shown there. (I don't use PS much any more.) PWP Sat masks (S from either HSV or HSL) do not look anything like the tutorial. My eyes glaze over a bit trying to follow the tutorial, but it seems evident that he is not constructing a simple mask, but is editing the mask to adjust it, especially for the flower.

However, I can do much better at reproducing the final images using the PWP masks, especially for the boats. So, I suspect there is some difference in the definition of what the gray-level mask intensity means between the PWP and PS masks. Jonathan may have some insight there.

In the case of the flower, however, if I were to try to accomplsh the final result (where the tutorial adjusts not only saturation but also hue and brightness), it seems simpler to me to think about using the color correction tool, maybe even using the new feature where you construct a mask from the affected color region (so you could later edit it and restrict changes to the flower petals only, not to any other parts of the image).
Top of pagePrevious messageNext messageBottom of page Link to this message

George McLelland
New member
Username: Georgemc

Post Number: 11
Registered: 09-2003
Posted on Wednesday, June 23, 2004 - 11:42 pm:   Edit Post Delete Post

Thanks, Jonathan
Top of pagePrevious messageNext messageBottom of page Link to this message

den
Unregistered guest
Posted on Thursday, June 24, 2004 - 11:59 am:   Edit Post Delete Post

Bob -- thanks for a response.

I agree with you that there has been editing of the masks shown in the tutorial in addition to just saturation contrast selection.

I have only been concentrating on the flower image, mainly because of the clipped R channel recovery aspect. In analyzing all channels, R, G, and B, not only is the R channel clipped but also the shadow detail of the B channel, particularly at the flower's center. With both R and B channels clipped, PWP would produce a high contrast saturation mask either HSV,S or HSL,S which is no where near the tutorial shown mask which is low contrast showing increased petal detail at the flower center.

In trying to duplicate the tutorial enhanced flower image, the saturation mask issue was ignored and attention given to R and B channel clipped image data recovery.

A very similar outcome to the tutorial enhance flower image was achieved by the following:

1. extract R, G, and B channels from the tutorial’s un-enhanced flower image.

2. modify the B channel using composite - lighten where the input image = B, input image mask = B, input wht = 0, input blk = 15, overlay image = G, overlay amt = 100. This modifies the B channel tone values of 0 to about 10 [used the ReadOut tool in HSV mode] and adds slight center detail.

3. modify the R channel using composite – darken where the input image = R, input image mask = R, input wht = 50, input blk = 0, overlay image = G, overlay amt = 100. This modifies the R channel tones values near the center of 100 to 90 [used the ReadOut tool in HSV mode] and adds slight center detail.

4. recombine R-modified, G, and B-modified channels. This produces an image very similar to the tutorial’s enhance image except for the stamen end saturations.

5. Preferring more saturation and contrast, the re-combined image’s contrast and saturation was increased by using the Brightness Curve transform in RGB space with control points at 20% and 80%. Curve ends were adjusted for full tone range and the 20% control point decreased by 10% for the added contrast. The saturation increased as a result the of the mid-tone decrease [normal for RGB space].

Ray -- Thank you for bringing this tutorial to the PW message board. The clipped data recover learned here will help me improve some images that showed okay on the camera’s histogram but actually had lost detail due to apparent over saturation, particularly red.
Top of pagePrevious messageNext messageBottom of page Link to this message

Jonathan Sachs
Board Administrator
Username: Admin

Post Number: 1810
Registered: 08-2002
Posted on Thursday, June 24, 2004 - 05:17 pm:   Edit Post Delete Post

One other technique I have seen used to restore lost highlight or shadow detail in similar situations is to copy detail from one of the other channels. Once the detail is gone, I'm not really sure how else you could recover it.
Jonathan Sachs
Digital Light & Color
Top of pagePrevious messageNext messageBottom of page Link to this message

den
Unregistered guest
Posted on Thursday, June 24, 2004 - 08:40 pm:   Edit Post Delete Post

Right Jonathan.... The steps outlined above in reality blend [copy] portions of the G channel detail into the burned out areas of the R and B channels. By using the R and B channels as their own masks and by varying the associated amount, one can blend G channel details into the burned out R and B channel areas without making it too obvious when the modified channels and the G channel are recombined. I just arbitarily chose to modifiy the original R [100% to 90%] and B [0% to 10%] channel tones. This seemed to result in good blended boundaries.

If one follows Steps 1 thru 4 and then compares the histograms [in RGB space]of the original image and the resulting Step 4 image, you will see that the narrow spikes at 0% and 100% tones have broaden and moved inward from 0% and 100% tones for the modified R & B channel recombined image.

This may be hard to visualize from the written description. If anyone would like screen shots, I can post them and/or email them.
Top of pagePrevious messageNext messageBottom of page Link to this message

Bob Walker
New member
Username: Rwalkernm

Post Number: 2
Registered: 06-2004
Posted on Friday, June 25, 2004 - 01:49 am:   Edit Post Delete Post

I am finished fiddling with the flower and saturation mask. I came up with 3 workflows that reproduce the final image in the LL tutorial well enough to suit me.

The first workflow strategy was to reproduce the "saturation mask" shown in the tutorial. That mask simply cannot be an HSL(S) mask in the PWP world. It is closer to a HSL(L) mask. See the tortuous workflow below. Screen shot can be found here.

The second strategy was to make a mask of the saturated yellow petal area, blur it, reduce saturation, and fine tune. Much simpler, more straightforward, equally satisfying result for me. Screen shot found here.

The third strategy was to use color correction tool. One step to success. See text file below, and screen shot here.


text/plaintext file of workflow
workflow.txt (1.5 k)



Top of pagePrevious messageNext messageBottom of page Link to this message

johnn
Unregistered guest
Posted on Friday, June 25, 2004 - 11:07 am:   Edit Post Delete Post

1. Using the new CC transformation click in an area with clipped reds. I used an area below and to left of the critter.
2. Click 'save as mask.'
3. Open CC again and, with the mask active, click on approximately the same location and drag toward center to reduce saturation to appropriate level.
4. If necessary open color/saturation and reduce overall saturation to suit.

In the originals there is very little difference between PWP and LL/PS - actually the PWP looks better.
Top of pagePrevious messageNext messageBottom of page Link to this message

den
Unregistered guest
Posted on Saturday, June 26, 2004 - 08:09 am:   Edit Post Delete Post

After some more thought……..

This tutorial procedure is creating is Saturation Contrast Mask, not separating out an image’s HSV,S or HSL,S channels. Hence the difference in appearance between the tutorial’s “saturation mask” and PWP’s HSV,S or HSL,S channel extractions.

The tutorial procedure does this by taking advantage of the HSL space characteristics where full saturation occurs at the 50% gray level; adding and subtracting 50% Gray to an image; and then taking the difference. This difference image is further edited to B/W and full tonal range to become the tutorial’s “saturation [contrast] mask”.

LL Tutorial Saturation [Contrast] Mask Procedure converted to PWP:

1. Click on the original un-enhanced image [Image1].

2. Open Composite – Lighten mode where Image1 is the Input Image at 100% and 50% Gray [click on the Overlay button, click Select Solid Color, click on 50% Gray on the color bar] is the Overlay at 100%, click OK [Image2].

3. Click on Image1.

4. Open Composite – Darken mode where Image1 is the Input Image at 100% and 50% Gray is the Overlay at 100%, click OK [Image3]. Image3 should be the active window [if not click on it before proceeding to the next step].

5. Open Composite – Absolute Difference mode where Image3 is the Input Image at 100% and Image2 is the Overlay Image at 100%, click OK [Image4]. Image4 should be the active window [if not click on it before proceeding to the next step].

6. Open Color -> Saturation – HSL space, move slider to –100%, OK [Image5]. Image5 should be the active window [if not click on it before proceeding to the next step].

7. Open Grey -> Levels and Color – HSL space, click full dynamic range, OK [Image6]. Image6 should be the active window [if not click on it before proceeding to the next step].

8. Open Convert, select 8-bit Black and White, OK [Image7].

Image7 is the Saturation Contrast Mask and should look something like this for the tutorial flower image:

SatCntrstMask1
Top of pagePrevious messageNext messageBottom of page Link to this message

johnn
Unregistered guest
Posted on Saturday, June 26, 2004 - 08:18 am:   Edit Post Delete Post

Forgot to mention that the sliders in the mask, Step 3, have to be reversed.

Top of pagePrevious messageNext messageBottom of page Link to this message

Bob Walker
New member
Username: Rwalkernm

Post Number: 3
Registered: 06-2004
Posted on Saturday, June 26, 2004 - 08:01 pm:   Edit Post Delete Post

den, you got it right, thanks for the improved explanation. Considering the mathematics of the transformations involved (I reread the Help pages on exactly what the transformations in composite are), you can skip a couple of steps and get to exactly the same result. No need to blend in 50% lighten and darken before taking absolute difference.

(1) Take original image, then Transformation/Composite, select 50% gray as overlay image, take absolute difference. All sliders at 100%, say (OK).

(2) Then do steps (6), (7), (8) as indicated above by den. Exactly same result, works for both the LL flower image and the LL seaside image.

I agree with johnn that Transform/CC is easier for achieving the final result for this image, and that the final result with PWP looks better than the PS image. But I like knowing better now what the editing options are.
Top of pagePrevious messageNext messageBottom of page Link to this message

den
Unregistered guest
Posted on Sunday, June 27, 2004 - 12:51 am:   Edit Post Delete Post

Bob … The consolidation of Steps 1 to 5 is eloquent and demonstrates the subtleties of PWP for one who is determined enough to explore…. Me….. I was just able to be literary minded at 2:30 am one morning.

This has been a very informative and thought provoking discussion. Thank you everyone.
Top of pagePrevious messageNext messageBottom of page Link to this message

meow
New member
Username: Meow

Post Number: 49
Registered: 10-2003
Posted on Sunday, June 27, 2004 - 05:10 am:   Edit Post Delete Post

I love these threads. I'm learning so much from them. I suggest we have at least one a weak. Deal? ;-)

Actually, I'm a little proud since I still feel very green. I sort of figured the middle gray and absolute difference out but I had done god-knows-what to the image first so it didn't look right at all. I was on the right track anyway and that's an improvement.

It's remarkable how much more easy to understand the PW process is. When I read your posts above I understand what's happening all the time. That's cool. To learn a process and to understand what it does at the same time. I think that was what made me choose PWP, that I intuitively felt that it doesn't hide things from me. It feels like a text editor compared to DreamWeaver or Word, if that comparison means anything to you. I want to learn but I don't want to learn which 'action' I should use in a given situation I want to learn what's needed and then how to do it.

Sorry for babbling and thank you for sharing.
Top of pagePrevious messageNext messageBottom of page Link to this message

Ray Milewski
New member
Username: Rmilewsk

Post Number: 86
Registered: 01-2004
Posted on Tuesday, June 29, 2004 - 12:20 am:   Edit Post Delete Post

I think that's what most of us want.

Topics | Last Day | Last Week | Tree View | Search | Help/Instructions | Program Credits Administration