LinearSharpen 3.12

[Pekka]

PC: http://photography-on-the.net/D30/linear/LS/LinearSharpen_312.zip

MAC: http://photography-on-the.net/D30/linear/LS/LinearSharpen_312.sit

This version has a totally new new color conversion engine which resolves all problems in 3.11. Thanks to all (esp. Roger and Griffin) for pushing me. I almost got it in one step, but to build good greens and good overall saturation without clipping anything I had to use a lot more steps :)

The color output of 3.12 resembles Canon CRW decode output very much but 3.12 has more open color tones especially on darker side, and a bit more shadow detail.

http://photography-on-the.net/D30/linear/LS/ls312.gif

Actions are:

Info 312
Copyright and support info

Convert only 312
Conversion only - feed this a linear TIFF and wait few seconds.

400-800 convert and sharpen
Simple sharpening and conversion for ISO 400-1600 photos.

400-800 convert and sharpen HQ
High quality sharpening and conversion for ISO 400-1600 photos. This is slow (depends on available RAM) but worth the wait.

100-200 chooser
Does all sharpening actions (non HQ) and places the results in history as snapshots to choose from.

100-200 chooser HQ
Does all sharpening actions in High Quality and places the results in history as snapshots to choose from.

convert (XXXX sharpening (HQ))
All sharpening + conversion actions where XXXX is level of sharpness and HQ means high quality (slower)

aftershave 1 (deeper colors)
After conversion (and sharpening) use this to make colors more 'bright' or 'deep'. Don't use more than twice if you don't want to loose detail in strong color. Try it on different kinds of photos and you'll see what it does.

aftershave 1 (shift yellow)
Use this for converted and sharpened photos. This is a special aftershave which I use for indoor ISO 800 concert hall shots - the auto WB in dark often gives a yellow cast which can be removed by clicking this a couple of times. Try it on different kinds of photos and you'll see how it affects.

result to grayscale
Use this for converted and sharpened photos. It makes three versions of grayscale version for you to choose from history palette snapshots

result to AdobeRGB
Use this for converted and sharpened photos. This converts color space to AdobeRGB which is more suited for printing than WideGamut.

result to web JPEG
Use this when you want to export photos to web galleries. This converts the photo to sRGB color space and then to 8 bit mode.

More aftershaves can be made to suit individual tastes!

All feedback welcome as usual.


Pekka

[SteveB]

Pekka,
What a great conversion,you've cracked it!
I've been using your Linear sharpening versions since way back and I have a bowl of friut photo which I use as a reference and the lemon has never looked so good,no loss of detail and a true colour rendition.I thank you for your efforts and your perseverance it's truly paid dividends.
All the best,
Steve

[John Boyes]

Pekka,

Thanks again for your wonderful efforts. Just tried 312 out on a couple of images and the result is beautifully neutral (no more oversaturated reds on skin tones) and wonderfully sharp. Can't wait to try it out on some landscapes!

I can finally remove 2.1 from my list :)

Many thanks!

John

[Dale]

Pekka,

Looks great on the few images I tested it with. Thanks for sharing with us.

Should all other 3.xx versions be removed? :D

Dale

[igor]

it works very fine,thanks again Pekka

[jzucker]

I'm still getting neon reds with an orange tint.

Sharpness looks great but breezebrowser's or Fred's LP Batch is still getting more realistic reds. Try taking a closeup of a red rose or tulip and you'll see what I mean...

Jaz

[Greg M]

Am I the only one who is getting pink skin?

I still use 2.1 and like it very much. I've tried all of the 3.?? and they all give me various problems with too much red. 2.1 with Fred's AWB has given me very nice results.

If it's something on my end I'd like to know so I could fix it but I'm not so sure that is it. Of coarse if nobody else has this problem then it must be on my end. No matter, I still have 2.1 which works great and the price is even better.

Thanks Pekka for your hard work and generosity.

[jzucker]

Thanks Pekka for your hard work and generosity.

Here, here... I sometimes forget my manners and I need to remember to thank Pekka for all his hard ware.

THANKS PEKKA!

P.S.

Please consider separate sharpening action!

Jaz

[Pekka]

jzucker wrote:
I'm still getting neon reds with an orange tint.

Sharpness looks great but breezebrowser's or Fred's LP Batch is still getting more realistic reds. Try taking a closeup of a red rose or tulip and you'll see what I mean...

Jaz

Hi,

Try this fix:

1. Double click the THIRD 'Channel Mixer' step in action 'WideGamut 312'.

2. A Dialog opens: select 'Blue' from dropdown list.
Type in the new blue values:

Red +33
Green -33
Blue +100

3. Click OK to save the settings.

4. Now, run the action again. Any better?

You can basically adjust red->magenta boundary by adjusting the aforementioned values. The more red (and less green), the more magenta you get. Just keep in mind when testing that in that dialog 'Red+Green' must always be zero).

PS. The whole Channel Mixer pile is like playing Rubik's cube - getting specific changes done means usually staring all over again from the bottom. It good practice for your brain I can assure! :)

[Pekka]

jzucker wrote:
Thanks Pekka for your hard work and generosity.

Here, here... I sometimes forget my manners and I need to remember to thank Pekka for all his hard ware.

THANKS PEKKA!

Thanks!

Please consider separate sharpening action!

It's really not possible as the reason it works so well is that sharpening is done 'inside' the conversion, before it's too strong, and after it has visible artifacts. The way it's there now gives the best overall quality.

[sudaplatov]

Greg M wrote:
Am I the only one who is getting pink skin?

I still use 2.1 and like it very much. I've tried all of the 3.?? and they all give me various problems with too much red. 2.1 with Fred's AWB has given me very nice results.

I prefer 2.1 (and LinearChooser C in 2.2) to 3.x.
However I did not use 3.12 yet - hope I will like it.

[Pekka]

Dale wrote:
Should all other 3.xx versions be removed? :D
Dale

Not neccessary.

Thanks,
Pekka

[Wheelie]

Pekka ,

I have not enough skills to make critics about your actions .
One thing is sure : I ran all of your LS and others and discovered they all provide great results and the fact I have to adjust a little , doesn't mean your LS do not work .
I'm afraid , the average user will expect the actions to fix all the shortcomings of the camera ... and the photographer ....
A universal fixing action will never exist - and it's just right - .
I thank you and admire your work to our benefit .
Let's don't get too lazy and accept some tweakening of our pics is a small inconvenience .
Thanks a lot for your work .
I'm happy with the results you allow me to achieve .

Best regards ,

Danny

[Sean Palmer]

Pekka,

Thank you very much for this action.
I have run it on on about 30 images that have been difficult previously and I am very happy with the results,
ok a little tweak here and there (only using AWB at the moment), but I have found a big improvement with reds, your sharpening option works superbly the best I have seen.

Many thanks

Sean,

[samson]

Pekka,

I have tried all versions of linearsharpen. I still think 3.09 is the best. It gives the most natural color of all the actions. I tried 3.12, and it consistantly gives a greenish cast to the image.

Any comments? Again thanks for your hard work.

[gerry]

I can't wait to spend more time playing with this.
Thanks.

[sudaplatov]

I tried 3.12 - works very nice.
Better than other 3.x.
Pekka, many thanx!
Sometimes I think that you do not exist in a sense
that bunch of people working behind your name
like in some legend about Shakespear :).
However, I saw you picture somewhere.

[Anthony8858]

Pekka:

Great Job!
Thanks.

[john_houghton]

Pekka,

Thanks for 3.12. The colours are much improved, and while they are now nearer to Canon's, the red of my VW parked outside is rendered much more accurately than in the Canon conversion. The sharpening is, of course, also clearly superior. Well done.

John

[Mark Zarn]

Pekka,
Like many of the others who have replied, I have to thank you for the generosity you display in sharing your skills with us.

I am nowhere near your proficiency in Photoshop, yet because of your work I feel that I can concentrate on capturing images that truly represent my emotional involvement with the subject matter, then produce a final product that maximizes the potential of my equipment.

Again, thanks,
Mark

[ajgrossmaniii]

For what its worth...

I am primarily shooting figure models these days with the D30. I had been processing my images lately with LS 305, and I was very happy with the results. CMYK skin tones were pretty close to ideal with Magenta and Yellow equal for the fair-skinned model I was working with.

I downloaded LS 312 and ran a comparison with some images of this one model from one studio session. I have noticed that LS 312 raised the percentage of Yellow and lowers the percentage of Magenta, further moving the skin tones away from ideal values.

Just as an FYI...I am staying with LS 305 for now because the skin tones with LS 312 are not as good.

[Pekka]

ajgrossmaniii wrote:
For what its worth...

I am primarily shooting figure models these days with the D30. I had been processing my images lately with LS 305, and I was very happy with the results. CMYK skin tones were pretty close to ideal with Magenta and Yellow equal for the fair-skinned model I was working with.

I downloaded LS 312 and ran a comparison with some images of this one model from one studio session. I have noticed that LS 312 raised the percentage of Yellow and lowers the percentage of Magenta, further moving the skin tones away from ideal values.

Just as an FYI...I am staying with LS 305 for now because the skin tones with LS 312 are not as good.

Thanks,

As I clearly seem to not have enough studio skintone and flower material :) , I bought a Kodak Q60 Color Input Target today (over 90 euros here in Helsinki!) and yes you're right the skin tones do not match.

To get LS312 more accurate in that section I did a correction (aftershave 3) action and it would be nice if you'd give it a try: http://photography-on-the.net/D30/linear/LS/aftershave_3_LS3122Q60.zip
or for Mac: http://photography-on-the.net/D30/linear/LS/aftershave_3_LS3122Q60.sit

Run it after original LS 312.

This aftershave corrects LS 312 output to very close of Q60 - skin tone's get more accurate and reds get deeper. All colors are affected in some extend.

Downside of aftershave 3 is that it reduces green saturation and generally lacks some vividness of original 312, but you can hit aftershave 1 a couple of times to get some of that back.

If you want to use aftershave 3 always with 3.12 just copy the two channel mixer steps (in right order) of aftershave 3 to the end of 'post 312' section of 312. If you like addition of aftershave 1 permanently, copy it to the end of 'post 312', too.

Tell me if this made LS any better for you.

Pekka

[Pekka]

I have uploaded this new 'Q60'-version as 3.42 (and you know where that 42 comes from, right? :) )

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

[T a z]

Pekka:

You continue to amaze me. Instead of becoming discouraged or defensive when one of us offers a critique or presents a problem we’re having with LS, you jump right in and try to make further adjustments. You then continue to offer them to the D30 community for further comments and critiques. Your ingenuity is surpassed only by your generosity.

Hope you won’t consider this overly dramatic when I say that in today’s world, this type of sharing is so utterly refreshing to see. I would imagine it’s along the same lines of why you wish to share your music with others…using your talents to share and inspire a bit of beauty wherever possible…which, ultimately, is the true measure of an artist.

But having said all that...let me be the first to say that uhmm…no, I guess I don’t know where the 42 comes from. I'm sure I should know, but I can't get it. Care to share this as well?

And one more thing, if I may…I’ve read (and tried to follow as best I could) quite a bit of Timo’s writings on color theory. I’ve also read of his continued crusade against the status quo of the “establishment community” of color theorists (some of the long running debates are quite lively on the Usenet forums).

Since LS is partially based on Timo’s findings, you obviously feel his theories have merit. So I’m curious to know why you chose to use sRGB instead of his Native PC or Native Mac color spaces.

Personally, I’ve found that his color spaces present a more accurate rendering, at least for my images, on the different uncalibrated (which is what those color spaces are designed for) computer screens I’ve tested them on.

Any particular reason you chose not to use his color spaces? Do they just not seem as accurate for you?

Cheers,

-Taz

[Griffin]

Pekka,

Thanks for your kind work. I am just able to sequeeze some time to try out and find it, so far, looks good. But well, it still takes a long time to convert. :) Nevermind, I guess I can fix that.

Here are some samples:

"3.42 convert only"

http://www.pbase.com/image/1323748/medium.jpg

"3.42 using 100-200 chooser"

http://www.pbase.com/image/1323741/medium.jpg

Comments?


Griffin.

[Roger_Cavanagh]

T a z wrote:
But having said all that...let me be the first to say that uhmm…no, I guess I don’t know where the 42 comes from. I'm sure I should know, but I can't get it. Care to share this as well?



42... it is the answer to Life, the Universe, and Everything.

[Pekka]

griffin wrote:
Pekka,

Thanks for your kind work. I am just able to sequeeze some time to try out and find it, so far, looks good. But well, it still takes a long time to convert. :) Nevermind, I guess I can fix that.

Here are some samples:

"3.42 convert only"

http://www.pbase.com/image/1323748/medium.jpg

"3.42 using 100-200 chooser"

http://www.pbase.com/image/1323741/medium.jpg

Comments?


Griffin.

Hi Griffin,

That's very odd. I get identical results with 'convert only' and all sharpening choosers and individual actions.
Check that you have all action steps in WideGamut activated. Remove all other LS's just to be sure they don't affect (they shouldn't but you never know what bugs PS has)

Anyone else have this problem?

[Pekka]

T a z wrote:

And one more thing, if I may…I’ve read (and tried to follow as best I could) quite a bit of Timo’s writings on color theory. I’ve also read of his continued crusade against the status quo of the “establishment community” of color theorists (some of the long running debates are quite lively on the Usenet forums).

Since LS is partially based on Timo’s findings, you obviously feel his theories have merit. So I’m curious to know why you chose to use sRGB instead of his Native PC or Native Mac color spaces.

Personally, I’ve found that his color spaces present a more accurate rendering, at least for my images, on the different uncalibrated (which is what those color spaces are designed for) computer screens I’ve tested them on.

Any particular reason you chose not to use his color spaces? Do they just not seem as accurate for you?

Thanks for you kind words, Taz.

Timo's crusade is interesting - it would be more acceped if Timo would be a bit more diplomatic in the way he presents it :)

I use WideGamut RGB as conversion color space (which Timo also uses, his working space is by his words Widegamut, G1.0, D65 (AIMRGBpro)). I use Eizo F78 sRGB (further adjusted with Adobe Gamma) as monitor/PS working space.

None of the color spaces are 'accurate', but some offer merely wider gamut, which means the image is more tolerant to color shift edits and can represent more color tints. sRGB is as good as any for viewing because no CRT can show real gamut of WideGamut anyway.

You should only convert to sRGB when exporting for web.

One goal with LinearSharpen 3 is that it does not use any external ICC profiles, as this way it'll run on every machine instantly. I'm considering doing an ICC profiled version later, with ICC+action installer program. But one problem with ICC's is that they are a bit 'stiff' to my purposes and to manipulate them effectively you'd need something like GretagMacbeth ColorChecker chart and ProfileMaker Pro 4 which are really expensive. Timo has been kind enough to share me a lot of information about all this.

And one word about conversion: there is no one fixed standard D30 output - I've used three different D30's and all of them had _very_slight_ differences in color reproduction. White balance, also by gray/white card, is always a _bit_ inaccurate in most light situations, and the difference of your camera to camera of whoever made the conversion/ICC can produce small color shifts (hard to notice without direct comparison though). Also, ISO affects characteristics of dynamic range so that ISO 100 has good highligh detail, and higher ISO give better shadow detail (I like ISO 200 most), so in theory the ICC for optimum reality reproduction should be done for each ISO separately.

Then there's the lens and protective filters :) I use Canon 50 f/1.4 without any extra filters for all color tests.

[Roger_Cavanagh]

Pekka wrote:
griffin wrote:
Pekka,

Thanks for your kind work. I am just able to sequeeze some time to try out and find it, so far, looks good. But well, it still takes a long time to convert. :) Nevermind, I guess I can fix that.

Here are some samples:

"3.42 convert only"

http://www.pbase.com/image/1323748/medium.jpg

"3.42 using 100-200 chooser"

http://www.pbase.com/image/1323741/medium.jpg

Comments?


Griffin.

Hi Griffin,

That's very odd. I get identical results with 'convert only' and all sharpening choosers and individual actions.
Check that you have all action steps in WideGamut activated. Remove all other LS's just to be sure they don't affect (they shouldn't but you never know what bugs PS has)

Anyone else have this problem?


Pekka,

I tried only one image and there is no problem. I've had a look at the action steps. Logically, there appears to be _no_ way that Griffin should get different results from the convert only option. I haven't checked every line yet, but so far no references to old versions in 342. It's very odd.

I have assembled a bunch of images to compare to see whether 342 is as good as it looks. I'll post my comaprisons when I'm done, but so far 342 looks to be another great leap forward!

How often can I keep saying the same thing? Thanks and well done!

Cheers,

[T a z]

You should only convert to sRGB when exporting for web.

I realize that sRGB is the conventional color space for web posting, but as Timo has tried to point out, because 98% of people view web pages on uncalibrated displays, the gamma of sRGB won't be optimized for them.

That's why he develped his Native PC and Native Mac color spaces. He's made an attempt to find a closer match for the gamma of most of the worlds computer displays...displays which are not calibrated to sRGB's gamma.

Sounds good in theory, and obviously I can't speak for more than just a handful of displays I've tested. But for those uncalibrated systems where I have been able to view a Native PC (or Native Mac) embedded image vs an sRGB embedded image....Timo's color space wins out every time.

But then, this is a whole different discussion than what LS is about.

And Roger....from what I've seen so far from 3.42, I'm going to take your explanation of "42" at face value, because with 3.42 I feel I have found that "it is the answer to Life, the Universe, and Everything"...at least as far as color correcting my images are concerned :)

Cheers,

-Taz

[Pekka]

T a z wrote:

I realize that sRGB is the conventional color space for web posting, but as Timo has tried to point out, because 98% of people view web pages on uncalibrated displays, the gamma of sRGB won't be optimized for them.

That's why he develped his Native PC and Native Mac color spaces. He's made an attempt to find a closer match for the gamma of most of the worlds computer displays...displays which are not calibrated to sRGB's gamma.

Sounds good in theory, and obviously I can't speak for more than just a handful of displays I've tested. But for those uncalibrated systems where I have been able to view a Native PC (or Native Mac) embedded image vs an sRGB embedded image....Timo's color space wins out every time.


NaticePC profile produces an image which has yet more shadow detail, but a bit less saturated colors (and 3.42 is already quite properly saturated). Looks good. But it's not a very general profile to me, as you have separate Mac and PC target profiling.

I could put a link to Timo's download page and version 2 of 'export to www' which would use Timo's ICC's. There would be one version of that step for PC and one for Mac.


And Roger....from what I've seen so far from 3.42, I'm going to take your explanation of "42" at face value, because with 3.42 I feel I have found that "it is the answer to Life, the Universe, and Everything"...at least as far as color correcting my images are concerned :)


Thanks fro your compliment. But hey, let's not try to influence Rogers opinion! He's doing an independent evaluation to his nice website, right ;)

PS. Could someone hint on dpreview.com that 3.42 exists in, please? Thanks. I don't feel comfortable advertising it there every other day 8)

Pekka

[T a z]

About providing a link to Timo's site from within the action...that's a nice thought, Pekka, but I doubt that it would benefit the majority of your action's users.

I'm sure I'm part of a tiny minority that care to even try Timo's color spaces...and even then, I don't do all that much posting to the web for a "non-calibrated" audience so it's not that big of an issue. I was mainly interested in your thoughts on Timo's color spaces from a colour theory perspective.

No...I would not want to distact you from doing what you're already doing with the actions, although I don't know how you could improve upon them as it is.

And by the by, I agree that the folks at DP Review owe themselves a look and have extended an invitation for them to drop by and give LS 3.42 a try.

Thanks once again,

-Taz

[Griffin]

Pekka wrote:
That's very odd. I get identical results with 'convert only' and all sharpening choosers and individual actions.
Check that you have all action steps in WideGamut activated. Remove all other LS's just to be sure they don't affect (they shouldn't but you never know what bugs PS has)


Pekka,

I think I have removed all other LS actions from the "Action" panel, not the physical file. I am not sure if I have done wrong in procedure. I will try to make up some time for that cause I am very busy right now -- less than 6 hours sleep since Friday night. That is one of those unfortunate IT people. :D

Skin tones are always my 1st concern in various linear profiles. I don't think other type of photographs would be so "demanding".

Keep up the good work!


Griffin.

[Greg M]

I left a message about getting pink skin using 3.12. I have since found that it did that only on some pictures and that more came out looking very good.

Your latest version does a teriffic job on even those that left pink skin in 3.12.

Thank you very much for sharing your hard work. I'd rather take pictures than try to learn PS enough to get these results. You make that possible.

[oops]

Wow. Very busy post just now! Want to sneak a "Thanks Pekka".

Glodilocks is one busy person these days. (Just Right!....Nope. Just Right!....Nope. Just Right!....Hmmm.)

:D

[samson]

Pekka,

Is sharpening different in 3.42 than in 3.09?


Thanks.

[fotosaccion]

Is linear sharpening good for Canon 1D photographs also?? I am new to all of this and own a 1D. I am trying to figure all of this stuff out.....
THANKS

[BigOsmo]

Hello all, hate being repeticious,but I gotta know.... I recently purchased a G3 and have been reading up on it and came across "Pekka's Linear Sharpen 3.12". Am I correct in that I can download this action into the actions pallette in PS7? Then what do I do? Run it on the converted tif? or on the raw image? I would really appreciate some enlightening. Thanks,
Bob

[D60DIETER]

Hi,

may be an uninteligent question! Can use this for RAW pic of the D60 too???

Dieter

[jmublueduck]

way to bump up a very old topic & confuse the crap out of me.

do a search for LS-D60.

answer to your question: NO. (edit, ok, I'm wrong. I meant the D60 version. sorry for confusion.)

[Roger_Cavanagh]

D60Dieter wrote:
Hi,

may be an uninteligent question! Can use this for RAW pic of the D60 too???

Dieter

Dieter,

No, but check here http://photography-on-the.net/forum/showthread.php?t=7503

Regards,

Roger

[martcol]

Can someone tell me what this is about - it looks interesting

[Roger_Cavanagh]

martcol wrote:
Can someone tell me what this is about - it looks interesting

Check here:

http://www.rogercavanagh.com/helpinfo/12_ls342.htm

and here:

http://www.rogercavanagh.com/actions/05_lsm.htm

and here:

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

Regards,

[Leo Reinhard]

I just tried this action on a D60 image and compared it to LS_D60 and Adobe ARC.

Although this is suppose to be for D30, I think the result of LinearSharpen 3.12 is better.

The ARC results in a lot od noise in the shadows and the LS_D60 results in to strong reds.
Has anyone else tried this on a D60 image?