Camera Raw and Sharpening

[Roger_Cavanagh]

There was some talk about the sharpening in ACR vs LS in the first long post discussing ACR. The view was expressed that LS sharpening was so good, it couldn't be beat.

I've been spending some time looking at options. I've reached no definitive conclusions as yet, but I decided to post this and get the ball rolling. Rather than embed lots of images in this post, go to this page http://www.pixelpixel.org/acr_vs_ls.htm for the example images. Note, there's a total of around 650kb.

I've used an image that has lots of edges and "texture" and looked at different ways of sharpening ACR converted images. On the samples page, I've posted 3 full-size crops and one crop upsized 4x.

Ignore the colours, I was sloppy about WB.

This is an assessment based on screen evaluations not print, as I'm trying to choose the best candidates for pirnt evaluation before blowing printer ink. :)

The four sets of images show:

LSM sharpness 3: levels 3 and 4 are both OK on screen, although 4 is quite an aggressive setting. Four is equivalent to high/strong is LS342. At 100% there are no obvious artefacts, but zooming in does show some haloes.

After testing various ideas, I narrowed down my choice to use a low/moderate level of sharpening in ACR followed by USM in Photoshop using the edge sharpening technique in LS.

The three variations posted are:

ACR 25 USM: sharpness slider set to 25, 0 smoothing, then two pass USM (125, 0.5, 0 and 50, 0.5, 0)

ACR x2 25 USM: sharpness and smoothness as before, but image upsized to max allowed for D30, downsized to 2160x1440 upon conversion, then USM as before

ACR x2 35 USM: sharpness 35 and smoothness as before, but image upsized and downsized as before, then USM as before

I think the examples indicate this approach has some promise. I particularly draw your attention to the difference in the second crop (the orange splodge) between LS and the ACR versions.

Comments and insights from own your tests, please.

Regards,

[AJSJones]

Roger, Thanks for the post and the info. I've not been as organized as you but have arrived at a similar (although temporary) conclusion that 25-35 in ACR with 0 smoothness is a good start (that doesn't do too much damage) and follow it with similar USM (I'll try the 2-pass method also) in PS.

For 400 and 800 ISO on my D60, I add a certain amount of smoothness to begin to tackle the noise. What would be really neat, after taking the time to finalize the values for each ISO, would be if the next ARC could read the EXIF data and select the ARC settings as part of a batch/action. My most recent trip had me changing ISO pretty often to accommodate the limitations of my lenses. In the meantime, do you (or anyone) know of an EXIF tool that could sort images based on EXIF information so I could put the different ISO shots in folders based on that info and then run batches with different ISO settings for each (I'm on a Mac and am just beginning to get serious about things like iView and other utilities. I'd also like to reset the creation date for my files based on the EXIF info since my computer has intermittent problems keeping the correct time!) Right now, I'm processing pretty much each one on its own merits and learning about ACR but it's getting old already!
Excuse the ramble
Andy

[Roger_Cavanagh]

Andy,

AJSJones wrote:
Roger, Thanks for the post and the info. I've not been as organized as you but have arrived at a similar (although temporary) conclusion that 25-35 in ACR with 0 smoothness is a good start (that doesn't do too much damage) and follow it with similar USM (I'll try the 2-pass method also) in PS.

For 400 and 800 ISO on my D60, I add a certain amount of smoothness to begin to tackle the noise.

I haven't look at high ISO at all yet (anything over 200 for me :) ) since I don't shoot very much at that speed.

What would be really neat, after taking the time to finalize the values for each ISO, would be if the next ARC could read the EXIF data and select the ARC settings as part of a batch/action. My most recent trip had me changing ISO pretty often to accommodate the limitations of my lenses. In the meantime, do you (or anyone) know of an EXIF tool that could sort images based on EXIF information so I could put the different ISO shots in folders based on that info and then run batches with different ISO settings for each (I'm on a Mac and am just beginning to get serious about things like iView and other utilities. I'd also like to reset the creation date for my files based on the EXIF info since my computer has intermittent problems keeping the correct time!) Right now, I'm processing pretty much each one on its own merits and learning about ACR but it's getting old already!
Excuse the ramble
Andy

You might want to have a look at PS Script Builder. I've written a bit about it here http://www.rogercavanagh.com/software_reviews/sc_pssc.htm#newstuff, and the links to the site are included. It does have the ability to process actions conditionally on ISO number, but I must confess, I've never actually tried that feature shooting - as I said - almost exclusively 200.

Regards,

[Georgees]

I almost added ISO and Camera combinations to the Automated (Un-Attended) Adobe RC scripting command in PS Script Builder but decided not to on this go around. I have had a couple of people mention just that in the last week now so I am of a mind to add it sometime next week.

I am in Los Cruces New Mexico right now and though I can add it here (I really never take a vacation and haven't for 30 years) my internet access speed is so slow I am hesitant to change it due to the excessive upload time.

However I will be back to the office Next Wed so I will likely code it and ready it for implementation as a new build before I get back. Look for it later in the week next week.

Of course with execution keys like that you would not need to separate anything, PSSB will just execute the correct command.

Regards

George Smith
www.ksscinc.com

[pigasus]

Roger_Cavanagh wrote:
After testing various ideas, I narrowed down my choice to use a low/moderate level of sharpening in ACR followed by USM in Photoshop using the edge sharpening technique in LS.

Great job so far, Roger. One question - by 'edge sharpening technique in LS' do you mean that you used the LS sharpening mask prior to the first USM and then applied the second USM to the full image?

[pigasus]

I've had a good look at your examples now, Rog. What is causing the x2 examples to brighten up so much? To my old eyes, and given the relatively small examples, it seems to me that highlight detail is lost in both the x2 variations. On the other hand the ACR 25 variation looks very good to me. Better than straight LSM sharpening. By the way, were the LSM sharpening variations done with full LS conversion or applied after ACR conversion?

[AJSJones]

Thanks Roger - I always shot my chromes at 100 ISO but with the D60 the difference in noise between 100 and 200 is very small, so I prefer 200 . For birding with my slow (5.6) 400 mm lens I often have to go to 400 or 800 to get anything decent, hence the need to deal with the noise.
George, PSSB looks neat and I will look forward to a Mac version that is "planned"!

Andy

[Roger_Cavanagh]

Sally,

pigasus wrote:
Great job so far, Roger. One question - by 'edge sharpening technique in LS' do you mean that you used the LS sharpening mask prior to the first USM and then applied the second USM to the full image?


Yes, I basically just lifted the sharpening routines including mask building from LS342/LSM and dropped the curves and channel mixer commands. I have it set up as a separate action that can be applied to an ACR conversion.

I've had a good look at your examples now, Rog. What is causing the x2 examples to brighten up so much? To my old eyes, and given the relatively small examples, it seems to me that highlight detail is lost in both the x2 variations.

I think there are two things going on here: I was sloppy with WB and EV adjustment. The image does need brightening, but I cannot be 100% certain I applied the same EV adjustment each time. :eyes But I got bored testing variations, so I posted something to generate discussion. Also most of my conversions were done in ignorance of Jeff Schewe's advice about 3/4 tones. The examples will need to be redone from scratch for a "formal review".

I don't think I agree with loss of highlight detail. I think that ACR sharpening at 25 at 4096x2731 just gives less sharpening, but I admit the comparison is made more difficult by the different WB.

On the other hand the ACR 25 variation looks very good to me. Better than straight LSM sharpening.

I agree it does look promising.

By the way, were the LSM sharpening variations done with full LS conversion or applied after ACR conversion?

Not sure whether I answering the right question here: the sample labelled LS3 is straight LS: conversion with YP+ARF2 and LSM quality 3, sharpness 3.

Regards,

[pigasus]

Roger_Cavanagh wrote:
Also most of my conversions were done in ignorance of Jeff Schewe's advice about 3/4 tones.

Was that the discussion about not using a gray card to white balance with? Or another bit of advice I've missed?

I'm definitely going to have to stop holding out and get ACR. I want to play too! :D

[Roger_Cavanagh]

pigasus wrote:
Roger_Cavanagh wrote:
Also most of my conversions were done in ignorance of Jeff Schewe's advice about 3/4 tones.

Was that the discussion about not using a gray card to white balance with? Or another bit of advice I've missed?

I'm definitely going to have to stop holding out and get ACR. I want to play too! :D


Yes, JS says you're not supposed to use a gray card because the WB is being done in linear space and so a 3/4 tone white object is the proper thing to pick.

[Dale]

Roger_Cavanagh wrote:

Yes, I basically just lifted the sharpening routines including mask building from LS342/LSM and dropped the curves and channel mixer commands. I have it set up as a separate action that can be applied to an ACR conversion.
I agree it does look promising.



Roger, could I please get a copy of your PS action referred to above?

[Roger_Cavanagh]

Dale wrote:
Roger_Cavanagh wrote:

Yes, I basically just lifted the sharpening routines including mask building from LS342/LSM and dropped the curves and channel mixer commands. I have it set up as a separate action that can be applied to an ACR conversion.
I agree it does look promising.



Roger, could I please get a copy of your PS action referred to above?

Dale, sent you an email.

Regards,

[Georgees]

I just thought I would add a reply to this thread since I had repllied a few days ago.

I have added Camera and ISO execution keys to the 'Automated (Un-attended) Adobe RC" script command.

This means that you can have different saved settings files from the Adobe raw converter plug-in that are camera and ISO related.

By entering a camera model and/or ISO execution keys to the command you can ensure that those settings are only executed on that camera/ISO image (taken from the image EXIF).

This also means you could do some camera/ISO's in un-attended mode but have other camera/ISO's opened with the Adobe RC plug-in preview window. Images that do not have camera model/ISO EXIF information that match an un-attended command will be opened in the preview mode.

See more on my site if you are interested. www.ksscinc.com.

Regards

George Smith

[LooseCanon]

Roger_Cavanagh wrote:
[quote]Dale wrote:
[quote]Roger_Cavanagh wrote:

Yes, I basically just lifted the sharpening routines including mask building from LS342/LSM and dropped the curves and channel mixer commands. I have it set up as a separate action that can be applied to an ACR conversion.
I agree it does look promising.



Roger - Sorry to be a pain, but would you mind also sending me a copy of your modified action? I've had a look inside the originals but can't sort out which bits do what!

Thanks and regards

Dr Andrew Pearce

[Francis]

Roger,
Seems to be it's gonna become a general demand. Also like the results of your sharpening process (even if you say it's not completely polished) but misses how to use the last step (Arc is ok, photoshop sharpening ok but miss how to mask). A temporary link to your action would be very nice.
Francis
P.S. would it work with d-60 images ??

[Dale]

georgees wrote:
I just thought I would add a reply to this thread since I had repllied a few days ago.

I have added Camera and ISO execution keys to the 'Automated (Un-attended) Adobe RC" script command.



George,

Works well. PS Script Builder and PS Side ClicK Pro are great PS tools.

[Dale]

With delivery of the 10D only days/weeks away I wonder if Adobe will be upgrading ACR soon and if there will be an additional $$99 price tag on it. I have the 10D on order :) :) and will be selling my D30 so I'm hoping for an updated ACR very soon.

[Roger_Cavanagh]

francis wrote:
Roger,
Seems to be it's gonna become a general demand. Also like the results of your sharpening process (even if you say it's not completely polished) but misses how to use the last step (Arc is ok, photoshop sharpening ok but miss how to mask). A temporary link to your action would be very nice.
Francis
P.S. would it work with d-60 images ??

Francis,

If you send me an email (go to rogercavanagh.com there's a contact link), I'll forward the same email, I sent Dale. You're welcome to try it. It's not magic solution, basically, it's an edge sharpen on top of ACR sharpening, but it seems to be working quite well.

Yes, it will work for D60 images, but (isn't there always one :) ) the USM settings may need adjustment to suit the D60 with the extra pixels.

Regards,

[Georgees]

Dale,

There is a good chance that the 10D raw files are the same as D30/D60.

I have a 10D raw (.crw) file but can not get it through ACR. I suspicion that is because I do not have the associated .thm file which contains all the EXIF data. Obviously ACR uses EXIF data or at least it seems so.

If anyone has a complete set of 10D .crw and associated .thm let me know.

Regards

George Smith
www.ksscinc.com

[etcetera]

By patching a few bytes of a 10D CRW file it runs fine with ACR as A D60 file - but not with Canons RIC...

Wheter ACR gets it right? Looks like it does, but there could be changes too...

Magne

[Francis]

Roger,

Just trying to get in touch with you through your contact link on your website since march 6th following your suggestion to send me an E-Mail with your edge sharpening action to use with ACR. Unfortunately I must be doing something wrong because got autoreply two times with no personal E-mail.
Could you help please
Thanks a lot
Francis

[Roger_Cavanagh]

Francis,

Sorry about that. It's not you doing something wrong. The autoreply proves that. Emails to rogercavanagh.com should be redirected by my ISP; usually they are, but for some reason yours never arrived.

Send me a mail at:

roger_cavanaghATfastmail.fm

Making the appropriate change, of course, and I'll forward the action.

Regards,