PDA

View Full Version : LinearSharpen 3


Pekka
13th of February 2002 (Wed), 15:16
Ok, this time it's a totally new conversion design. I have literally tried hundreds of different variations of the action and still there were some problems with color quality (for example, color corrections made color gradations 'edgy'). I always returned to the drawing board and at last yesterday I finally realized how the linear conversion should be done properly. The ticket to success is "Wide Gamut ICC" which is included in Photoshop. Wide Gamut has been discussed in other forums before, but no one seemed to have any decent results with it.

You need to convert to Wide Gamut, do curves, add some saturation and add to selective green saturation separately. Very simple and fast.

This new method is superior to Canon's 16-bit TIFF conversion and produces an image which holds full highlight detail and extremely natural colors. Canon conversion has color problems (too much yellow on green, the darker the colors get the muddier they look, highlight detail lost, blues look always same tinted and reds are too pink.

I'll place a demo of highlight recovery feature of this method tomorrow. It's really noticeable.

LinearSharpen 3 contains actions for

- conversion only

- conversion + sharpening for low ISO images, with slight jaggyness reduction

- conversion + sharpening for high ISO images, with slight noise reduction and jaggyness reduction

- result to sRGB/8-bit

High ISO sharpening can be used on low ISO images too, but it's much slower due to noise reduction. In this version sharpening is intented to be conservative and of high quality, and main goal was to not mess with the area on 'DoF border' and not to produce any artifacts like pixelisation and jaggies.

NOTE THAT THE RESULTED PHOTO IS IN WIDE GAMUT MODE, AND I SUGGEST YOU KEEP THE PHOTO IN THAT MODE WHEN SAVING (SAVE WITH PROFILE). Wide Gamut mode has a very good quality in e.g. saturation changes so it's adviceable to keep converted "master" photos in that mode.

If you need sRGB for web there is an action included in LinearSharpen 3 which does sRGB conversion are conversion to 8 bits. Make sure you use sRGB when saving JPEG's for the web, otherwise colors are really off on www browsers.

For any conversion OUT of Wide Gamut ICC, please convert to AdobeRGB first, then convert to the desired target. This is because sRGB and many other ICC's are so narrow you'll get color artifacts (e.g. on smooth bokeh) if you convert straight to any low gamut ICC. So if you need e.g. PAL, do Wide Gamut -> AdobeRGB -> PAL.

More tomorrow - download from http://photography-on-the.net/D30/linear/

Let me know how it works for you.

Pekka

Pekka
13th of February 2002 (Wed), 15:37
A couple of converted and sharpened fullsize photos:
http://photography-on-the.net/D30/linear/CRW_2640.jpg
and
http://photography-on-the.net/D30/linear/CRW_7363.jpg

PS. I uploaded a new zip (and .sit) where I deactivated a few almost 'dummy' levels commands (they were for testing only, you can set them pop up if you like).

DWerner
13th of February 2002 (Wed), 16:34
Pekka,

Using "Convert Only" photoshop 6.01 reports this error:
"The object "action" hue and saturation" of set LinearSharpen 3" is not currently available."

Using "Convert and Sharpen" These 3 errors are reported:
"The object "Action "Mask" of set "LinearSharpen #" is not Currently available".

"..."Action "Remove Mask" ...."

"... "action "hue and saturation" ..."

JohnM
13th of February 2002 (Wed), 16:59
Same problem here.

msingewald
13th of February 2002 (Wed), 17:13
Ditto on the missing actions

Roger_Cavanagh
13th of February 2002 (Wed), 17:14
Pekka,

This conversion looks very promising indeed testing on a few of my "standard images". Couple of things:

- Users needs to change the name of the action set by substituing a space for the underscore for the play action to run

- What's with the upsize and downsize in the convert and sharpen actions?

Glad to see you've gone back to sharpening in lab. I think this does gives better results.

Again, well done and thanks.

Cheers,

Roger

Roger_Cavanagh
13th of February 2002 (Wed), 17:15
Anyone who's having trouble running the actions, just change the name of the action set by changing the underscore to a space.

(Double-click on the set name.)

Roger

jzucker
13th of February 2002 (Wed), 20:16
Beautiful. I like the sharpness better than Miranda's actions but the reds still aren't right. Can the sharpening be run by itself from the low iso convert/sharpen?

Pekka
14th of February 2002 (Thu), 00:45
msingewald wrote:
Ditto on the missing actions

Sorry about that, I've uploaded a version with correct action name - it'll work straight away. Thanks for Roger for posting the fix (It was 1 am here when I uploaded the action and I went to sleep for few hours :) )

samson
14th of February 2002 (Thu), 05:17
Pekka,

Great work on Linearsharpen 3. Blues are still slightly oversaturated.

Can you add an option for additional sharpening? Choices of images like you had in Linearchooser would be great.


Thanks.

Pekka
14th of February 2002 (Thu), 07:12
Hi Roger,

Roger_Cavanagh wrote:
Pekka,

This conversion looks very promising indeed testing on a few of my "standard images". Couple of things:

- Users needs to change the name of the action set by substituing a space for the underscore for the play action to run

Yep, thanks for posting the fix. These things seemt ohappen when you upload stuff in hurry.

- What's with the upsize and downsize in the convert and sharpen actions?

It improves D30's problem area a bit: jaggy antialiasing.
In high ISO sharpening it does jaggy removal and also smooths noise a bit. The jaggy improver is slight but it makes sharpening better in quality.

Glad to see you've gone back to sharpening in lab. I think this does gives better results.

I used lab in high ISO, after curves. I'd love to use LaB in linear sharpening (before and between curves) but there is a problem in RGB->LaB->RGB conversion which makes histogram visibly different and you loose some shadow detail. I tried to apply a compensation curve but did not succeed yet to make the result identical to non lab version.

You can check out the full lab sharpening in low ISO version if you download 3.01 from http://photography-on-the.net/D30/linear/ and activate all steps in that action. This gives you the version I'd like to use: Lab sharpening and manual levels adjust before hue changes. Do a lab version and then compare that to convert only version and you'll see the curve difference I'm talking about.

Again, well done and thanks.

Thanks Roger for supporting this in most positive and constructive way!

Pekka
14th of February 2002 (Thu), 07:23
jzucker wrote:
Beautiful. I like the sharpness better than Miranda's actions but the reds still aren't right. Can the sharpening be run by itself from the low iso convert/sharpen?

Thanks,

It wouldn't be the same as the sharpening is done in linear mode before curves, which is the main point for better sharpening.

Pekka
14th of February 2002 (Thu), 07:44
samson wrote:
Pekka,

Great work on Linearsharpen 3. Blues are still slightly oversaturated.

Thanks. How do you evaluate the blue saturation? Can you mail me an example?

Does slight saturation decrease help in your mind?

Also what affects perceived saturation a lot is original light vs. exposured light and how you set overall levels later.

Can you add an option for additional sharpening? Choices of images like you had in Linearchooser would be great.

Yes I'll do that, later. It would be great to try and offer more sharpening methods, too, but Photoshop is very limited in 16-bit mode.

Pekka

jzucker
14th of February 2002 (Thu), 08:08
Pekka wrote:
jzucker wrote:
Beautiful. I like the sharpness better than Miranda's actions but the reds still aren't right. Can the sharpening be run by itself from the low iso convert/sharpen?

Thanks,

It wouldn't be the same as the sharpening is done in linear mode before curves, which is the main point for better sharpening.

OK,

What can we do fix or at least adjust the red saturation problem? If you take a properly exposed picture of a deep red rose or tulip and run it through your action you will see that the reds look neon-like. If we could optionally tone that down, I'd really be happy. I'd never run the other convert action I have again!

Thanks again for all your hard work.

Jaz

Hartmut
14th of February 2002 (Thu), 10:50
Hi!
I 'm not really satisfied with Linear 3!
Her some pics:
The first ist done with "old" Linear Sharpen:
http://www.schulla.at/misc/linear.jpg
The second with Linear Sharpen 3
http://www.schulla.at/misc/linear3.jpg
The third is a crop of Linear 3:
http://www.schulla.at/misc/linear3crop.jpg
You see the Red channel ist different to the original.
Yellow is turned to red in the crop!!!!
Regards
Hartmut

KHogan
14th of February 2002 (Thu), 11:00
Hi Pekka,

Phew, you are way ahead of me!! Had wanted to post some more comments on version 2.1 but then ran into a little snag with my ISP which you helped me with then got busy. Anyway, in case they're of any help, I'll post a few comments mostly from the 2.1 version since I've only just tried your latest version.

I wanted to go back to the sharpening issue with 2.1. I did some more testing and took a closer look at the 5 sharpening versions and ultimately, my favorite of all the choices for sharpening was "D". I found that C & E produced haloing in most images I tried and that B was really oversharpening the detailed areas of the images. D was overall the best for the widest variety of images. Still though, I decided I preferred to do the sharpening afterwards on my own.

Now, a comment about colours which so far also seems to apply to this latest version 3. I have been using some images that have subtle blues which are redder than they are green (i.e. skies shot before and at dusk) because these are examples of shots that I'm having the most trouble converting. My results are showing that a conversion of these images, with both versions 2.1 and 3, are yielding results that are too green/cyan rather than blue/magenta as the original shows. Have you or anyone else found this? Also, I'm getting yellows tending to blue. In other words, I have to subtract blue and green to get back to the original colours. Again, this is based on only a few tests, half a dozen varied images.

On the sharpening, so far I prefer this latest version but I must do more testing with version 3 to be able to say more. :)

Thanks again! I will play some more with LS3.

Kharim

jzucker
14th of February 2002 (Thu), 11:03
What I'd like to do at this point is somehow combine Miranda's Linear Batch with Pekka's sharpening...

Thoughts?

KHogan
14th of February 2002 (Thu), 11:46
jzucker wrote:
What I'd like to do at this point is somehow combine Miranda's Linear Batch with Pekka's sharpening...

Thoughts?

Jack,

I haven't seen Miranda's action, but if it's an action, I can't see any reason you couldn't do that. You'd have to create a new action and mix and match the parts of Fred's and Pekka's actions to get what you're looking for. Of course, I'm saying this without having seen how Fred's set up his action. But it's an idea... ;)

Kharim

Pekka
14th of February 2002 (Thu), 14:23
Try version 3.03.
It does something to red shift issue among other things.

PC: http://photography-on-the.net/D30/linear/LinearSharpen_303.zip

MAC: http://photography-on-the.net/D30/linear/LinearSharpen_303.sit

jzucker
14th of February 2002 (Thu), 14:25
Thanks! I'll try it tonight or tomorrow. Tonight might be kind of hard being valentine's day. I think my GF would kill me if I spent a few hours on the computer!

Pekka
14th of February 2002 (Thu), 14:33
Hartmut wrote:
Hi!
I 'm not really satisfied with Linear 3!
You see the Red channel ist different to the original.
Yellow is turned to red in the crop!!!!
Regards
Hartmut

Hi Hartmut,

The new version is totally different to old one in regard to how conversion is done. This gives some advantages and also some misadvantages but in my mind advantages are very obvious and worth it.

Please note that you should always compare to what you see in that particular lighting and placement of objects you want to color-compare. Any version of linear converter should not be treated as the truth, there are too many variables starting from exposure and white balance. I use Kodak white card and custom WB for testing, it's the only way to be sure. Then I test the action with a couple of hundred shots and see if any unnatural stuff is going on and if skin, sky and plant colors look normal.

Maybe you could mail me the CRW and THM of that photo of yours and I'll take a look what's happening there?

Thanks,

Pekka

colin walker
14th of February 2002 (Thu), 14:39
Hi,

A little D30 Linear Conversion gallery below:-

http://www.pbase.com/colinwalker/d30

I will add a couple more images + Fred's action soon.

Colin

John Boyes
14th of February 2002 (Thu), 17:17
Hi Pekka, do you ever get any sleep! :)

First, thanks for the latest action. I agree that the colours do look more natural, however I find the sharpening isn't as good as LA 2.1

I have been running 2.1 since you posted it and then LA "aftershave" action afterwards to tone down the reds. I haven't tried 2.2 beta or the chooser.

I just downloaded and tried LA 303 against LA 2.1 and I found 303 had the better colours but 2.1 was quite a bit sharper. Is this because 2.1 (and your previous actions all the way back to 1_01) use the "find edges" filter and allow manual tweaking of the edges and mask whereas 303 is the first that doesn't?

Examples: these are full size crops. The LA 2.1 example was converted with 2.1 using 100 - 400 set levels option then "aftershave atn" run followed by the sRGB conversion atn from the 303 set.

The 303 atn was run using 100 - 200 convert then sharpen option and then the sRGB conversion from the 303 set

http://www.btinternet.com/~incamera/test/la_2_1.jpg
LA 2.1

http://www.btinternet.com/~incamera/test/la_303.jpg
LA 303

Has anyone else noticed this?

samson
14th of February 2002 (Thu), 19:16
Pekka,

What is the best way to add sharpening to the image after running linearsharpen 3.03?


Thanks.

samson
14th of February 2002 (Thu), 21:20
Pekka,

One more problem. When I print an image converted with linearsharpen 3.03 the colors (especially blue) are all wrong though the image looked good on the monitor. When converting with linearchooser or any of your previous versions I don't have this problem. What is going on?


Thanks.

Pekka
15th of February 2002 (Fri), 05:12
samson wrote:
Pekka,

One more problem. When I print an image converted with linearsharpen 3.03 the colors (especially blue) are all wrong though the image looked good on the monitor. When converting with linearchooser or any of your previous versions I don't have this problem. What is going on?

You should probably convert to sRGB or AdobeRGB for printing. There are not many printers which can handle gamut of Wide Gamut RGB profiled photos.

Pekka
15th of February 2002 (Fri), 09:08
samson wrote:
Pekka,

What is the best way to add sharpening to the image after running linearsharpen 3.03?

Try 3.05, it has three levels of sharpening, and a chooser action which puts choices as history snapshots. Other than that, plain USM with modest settings might work fine.

http://photography-on-the.net/D30/linear/LinearSharpen_305.zip

PS. the chooser is now for ISO 100-200 only.

Pekka
15th of February 2002 (Fri), 09:15
John Boyes wrote:
Hi Pekka, do you ever get any sleep! :)

First, thanks for the latest action. I agree that the colours do look more natural, however I find the sharpening isn't as good as LA 2.1

Try 3.05

http://photography-on-the.net/D30/linear/LinearSharpen_305.zip

sudaplatov
15th of February 2002 (Fri), 12:48
LinearSharpen 3.03 works very well in 400-800
mode, but very very slow mostly due to resizing.

Pekka
15th of February 2002 (Fri), 15:26
sudaplatov wrote:
LinearSharpen 3.03 works very well in 400-800
mode, but very very slow mostly due to resizing.


Resizing is needed for reducing noise, if this is not done it would be impossible to get such good a sharpening quality which will not make noise worse and pixelize places which are not in focus. Resizing also helps removing jaggyness caused by USM.

Sometimes I set Photoshop's history number to 1, and this speeds big operations considerably.

Anyways, I'll try to see if the speed can be improved.

samson
15th of February 2002 (Fri), 22:35
Pekka,

When I try to convert to Adobe 1998 after running linearsharpen, (image>moce>convert to profile) it seems that the image is already in Adobe 1998, which is my working space. What am I doing wrong? I still get poor colors when printing? How do I convert the image to out of wide gamut RGB?

John Boyes
16th of February 2002 (Sat), 07:10
Hi Pekka, thanks for pointing me to 305. Below is another test using the same image as in my previous post. In this test I've compared LS 1_02, LS 2_1 and LS 305.

Workflows:

LS 1_02: "LS vr 2 RECOMMENDED" setting used - USM levels as suggested in atn - atn converts automatically to sRGB - convert to 8 bit and save for web JPG at max (80) quality

LS 2_1: "ISO 100 -400 (set levels) used - levels set to edges of histogram for clean black lines - LA Aftershave atn run to tone back the reds - "Result to sRGB" atn run - convert to 8 bit and save for web JPG at max (80) quality

LS 305: "100 - 200 high sharpen" used - "Result to web" atn run - save for web JPG at max (80) quality


The test image is a full size 250 pixel @96dpi crop. ISO 100 - custom WB - 17-35L - Sat: Normal, Sharp: Low, Contrast: Low

http://www.btinternet.com/~incamera/test/ls_1_02.jpg
LS 1_02
http://www.btinternet.com/~incamera/test/ls_2_1.jpg
LS 2_1
http://www.btinternet.com/~incamera/test/ls_305.jpg
LS 305

IMHO I still think LS 2.1 is the sharpest with 305 second and 102 third - there seems to be much more detail in the handle of the wooden tool. The colours of 305 are certainly very natural and to my eyes the best of the three.

Being able to select the edges as in 2.1 seems to produce the best control and results for me.

Comments anyone?

Pekka
16th of February 2002 (Sat), 14:41
samson wrote:
Pekka,

When I try to convert to Adobe 1998 after running linearsharpen, (image>moce>convert to profile) it seems that the image is already in Adobe 1998, which is my working space. What am I doing wrong? I still get poor colors when printing? How do I convert the image to out of wide gamut RGB?

Hi samson,

I tested printing on my Epson Photo 750 and it worked just fine (phew!). Let me try to describe the workflow in Windows 2000:

- turn System and Photoshop color management off
- load the linear TIFF (it should have "untagged RGB" status on status bar - see below)
- convert to nonlinear TIFF with LinearSharpen
- convert to profile AdobeRGB(1998) using Microsoft ICM, relative colorometric and black point compensation
- keep the file in 16 bits

Note: You can set status bar in Photoshop to display current color mode (click that small triangle).
In color management you should put on all warning messages on if you're unsure what profiles you are opening/using.

To print:

See in Photoshop's 'print options' that
-SOURCE SPACE is 'Document: AdobeRGB'
and
-PRINT SPACE is 'Printer Color Management'

(there is a 'Show More Options' button under Photoshop's print settings dialog which shows you aforementioned Color Management settings)

Enable in printer settings' ICM color management and you should get excellent results.

Hope this helps.

I'll post workflow screenshots if you should need them.

Pekka

samson
16th of February 2002 (Sat), 16:29
Pekka,

How do I turn system and Photoshop color management off?


Thanks.

samson
16th of February 2002 (Sat), 16:49
Pekka,

I may have figured it out by going to Edit>color settings. WHen I go there and set coloro management off (in settings) my working space becomes my monitor profile ( and no longer Adobe RGB 1998) Is this correct?

samson
16th of February 2002 (Sat), 16:53
Pekka,

One more question while I'm at it:

What should all the settings be in PS 6.01 under Edit>Color Settings, for best results using linearsharpen 3.05?


Thanks again.

Dale
17th of February 2002 (Sun), 08:03
Pekka,

Maybe you could give us a work flow for LS v3.05 and PS.

Thank you.
Dale

samson
17th of February 2002 (Sun), 13:35
Pekka,

Linearsharpen 3.05 seems overall to produce undersaturated images (especially outdoor shots). Do others have this experience?

Roger_Cavanagh
17th of February 2002 (Sun), 14:30
samson wrote:
Linearsharpen 3.05 seems overall to produce undersaturated images (especially outdoor shots). Do others have this experience?

I shall be writing up my tests on my web site in due course, but I thought I would respond immediately to Samson's comments.

Here is an image demonstrating different processing options:

http://www.rogercavanagh.com/images/temp/linear_chooser/combined.jpg

I shot an image of a Kodak Q13 card using a gray card for exposure setting and CWB. I cut and pasted sections from the different processing options to produce this combined image. The five options are:

1. Standard nonlinear 16-bit TIFF settings for contrast, saturation and sharpening were low, low, low.

2.LinearSharpen 3.05 (conversion only)

3. LS 3.05 convrsion only then apply additional saturation with Fred Miranda's Digital Velvia Pro medium.

4. LinearPro 1.3 conversion only.

5. LP 1.3 conversion with normal saturation.

So I agree, Samson, that extra saturation appears to be necessary. I plan to do more testing, but at the moment, I'd say LS305 is looking pretty damn good.

Go Pekka! Go Pekka! :D

Cheers,

Roger

samson
17th of February 2002 (Sun), 15:17
Roger,

What method do you use to increase the saturation following linearsharpen 3.05?

Roger_Cavanagh
17th of February 2002 (Sun), 15:36
samson wrote:
Roger,

What method do you use to increase the saturation following linearsharpen 3.05?

Samson,

I used Fred Miranda's Digital Velvia action. Here's the web link:

http://www.fredmiranda.com/DV/index.html

This does a great job of increasing saturation "realistically", i.e., you don't just move the slider. I didn't consider different options when checking out the results of LS3.05. Like you I though colours looked unsaturated. Since I have DV on my actions palette I tried this first and liked the results, plus there's not much point in using LS (or alternatives), if you have to do anything too complicated to the images after conversion.

Regards,

Roger

samson
17th of February 2002 (Sun), 16:00
Roger,

How do you have "color settings" set up in Photoshop 6.01 for using linearsharpen 3.05?

Roger_Cavanagh
17th of February 2002 (Sun), 16:06
samson wrote:
Roger,

How do you have "color settings" set up in Photoshop 6.01 for using linearsharpen 3.05?



Like this:

http://www.rogercavanagh.com/images/temp/ps_color_settings.gif

Cheers,

Roger

samson
17th of February 2002 (Sun), 16:51
Thanks Roger,

Why does Pekka say to turn off both color management in Photoshop as well as system color management (see above) when converting from wide gamut RGB to Adobe RGB prior to printing?

(how do you turn off system color management?)


Thanks.

Griffin
17th of February 2002 (Sun), 19:47
I barely have time to try out a couple of earlier works with LS3.05 and see the difference, I am currently using Fred's action. Here are some ups and downs and a couple of questions.

The results looks pretty good, but saturation adjustments is not integrated so additional step is required. The bad thing is the speed is too slow! I am going to try out a few more to see how fare it would be.

A number of question for Pekka:

1. Does this action support batch mode? I have less time now and it is important that I can perform batch procession.

2. What kind of noise reduction is used? Is it an in-house or a 3rd party one?


Griffin.

Roger_Cavanagh
18th of February 2002 (Mon), 04:18
Griffin wrote:

The bad thing is the speed is too slow!


In an earlier post Pekka wrote:

It [image upsize-downsize] improves D30's problem area a bit: jaggy antialiasing. In high ISO sharpening it does jaggy removal and also smooths noise a bit. The jaggy improver is slight but it makes sharpening better in quality.

The slowness is almost entirely caused by the upsizing and downsizing. I have not done any comparisons to compare the difference with and without this step. Pekka, would you like to comment more on this?

Griffin wrote:

... but saturation adjustments is not integrated so additional step is required... Does this action support batch mode?

Griffin, there is no problem incorporating, say, Fred Miranda's Digital Velvia Pro as part of an automated workflow with LinearSharpen. I plan to look at the best place to insert DV into the workflow. In LinearPro, DV works either before or after conversion. Although the results are not identical.

Pekka has already commented that "Chooser" cannot work in batch because the history snapshots are lost, but otherwise there should be no problem with a File>Automate.

Regards,

Roger
www.rogercavanagh.com

mbvt
18th of February 2002 (Mon), 13:39
Howdy,

Pekka, thank you so much for developing LS, publishing it unselfishly and taking the time to explain everything patiently.

Part of the steps in LS 3 is upsampling the image, and then downsample it again to the original size.

What should I do if I need to upsample the image anyhow for printing? Should I let the action run and upsample the image again to taylor it to printing needs, or is it better to skip the downsampling steps near the end of the action?


Personally, I'm not yet satisfied with the colors so far but the action does result in very, very sharp images. I still use the LC New profile that's distributed via FredM often because I really like the very smooth, very soft and 'golden' colors it gives.

So, I've have another question/idea: would it be possible to use both the results of the linear image - for sharpness - and the non-linear image - for color rendition? I do this sometimes by stacking the two on top of each other and then blend them, but there surely must be more sophisticated ways.

Isn't sharpness mostly in the Luminance channel? So would it be possible to copy the luminance channel of the LS3-ed linear image over the luminance channel of the non-linear image after converting both of them to lab mode? Or would't this make any sense?


gr. Michel

P.S. Do you still have the latest release of LS 2.1. available on your site somewhere? I think I've only got the one-but-latest release and I feel it suits my needs excellently on some occasions.

mbvt
18th of February 2002 (Mon), 13:42
Howdy,

Just to let you know:

the LS305 action 400-800... isn't working (correctly), because it is calling 'hue and saturation' in LS303, not LS305.

gr. Michel

Pekka
18th of February 2002 (Mon), 15:09
mbvt wrote:
Howdy,

Pekka, thank you so much for developing LS, publishing it unselfishly and taking the time to explain everything patiently.

Part of the steps in LS 3 is upsampling the image, and then downsample it again to the original size.

What should I do if I need to upsample the image anyhow for printing? Should I let the action run and upsample the image again to taylor it to printing needs, or is it better to skip the downsampling steps near the end of the action?


Up and downsampling are part of image quality improvement process (it gets rid of some ugly individual pixels after USM). The resulted image is same size you started with.


Personally, I'm not yet satisfied with the colors so far but the action does result in very, very sharp images. I still use the LC New profile that's distributed via FredM often because I really like the very smooth, very soft and 'golden' colors it gives.

It's good that most actions are different - it's like having a different films for different color tastes. But please check out 3.09, I've make some significant improvements there in color depth and accuracy plus improved sharpening among other things.

http://photography-on-the.net/D30/linear/LinearSharpen_309.zip

So, I've have another question/idea: would it be possible to use both the results of the linear image - for sharpness - and the non-linear image - for color rendition? I do this sometimes by stacking the two on top of each other and then blend them, but there surely must be more sophisticated ways.

Isn't sharpness mostly in the Luminance channel? So would it be possible to copy the luminance channel of the LS3-ed linear image over the luminance channel of the non-linear image after converting both of them to lab mode? Or would't this make any sense?

When Photoshop (7?) let's you use all functions in 16-bit mode it'll be easy to try many sophisticated things. Now you're restricted to only a few commands, no layers, no layer effects, no masks.

Also, I've noticed that in linear mode, conversion to LaB changes contrast and colors.


P.S. Do you still have the latest release of LS 2.1. available on your site somewhere? I think I've only got the one-but-latest release and I feel it suits my needs excellently on some occasions.

http://photography-on-the.net/D30/linear/LinearSharpen_2_1.atn

Roger_Cavanagh
18th of February 2002 (Mon), 16:59
Pekka,

Slow down! You're producing new versions quicker than I can try them out! :D

After a couple of tests, V3.09 is looking tremendous. This is one of my test images with extra-high sharpening:

http://www.rogercavanagh.com/images/temp/linear_chooser/t02-ls309x.jpg

The subtle differences in the various colours really show up. And the base looks much improved with the tiniest bit less red. This is a V3.5, if not V4. :D

Kiitos, kaikkea hyvää!

Roger