This content has been marked as final. Show 79 replies
Camera Raw 4.1 is doing a more advanced version of the "blend if" logic, which was tuned with the help of Jeff Schewe, who stepped in to complete Bruce's consulting work after his untimely death. You don't need to "adjust them according to the color profile and system gamma", since the processing is done while the data in a fixed internal color space.
Thomas Knoll wrote:
> Camera Raw 4.1 is doing a more advanced version
> of the "blend if" logic ...
Thomas further wrote:
> You don't need to "adjust them ..."
That's even better!
Thank you for the clarification.
Is there any documentation on how the these camera raw preferences work? The photoshop help was somewhat terse.
- make defaults specific to camera serial number
- make defaults specific to camera iso setting
My question is - how do I set camera raw defaults for each combination of camera SN and ISO for the detail settings?
>how do I set camera raw defaults for each combination of camera SN and ISO for the detail settings?
Calibrate your camera and save the specific results.
>Calibrate your camera and save the specific results.
Not what I was asking, I believe. According to Jeff Schewe the ACR detail pipeline is sensitive to the preferences for each SN and ISO, so I want to know how to save default settings for the details (sharpening and NR) for each ISO by camera.
I think Ramon already answered that question. You don't save "defaults" for each ISO, etc.. You save a subset for each of those ISO settings that you require.
>I think Ramon already answered that question. You don't save "defaults" for each ISO, etc.. You save a subset for each of those ISO settings that you require.
Thanks Jim. So what do these ACR preferences do? What do you mean by saving a subset (save settings)?
From the way you talk, it seems that you understand how to save camera raw defaults. Those settings will be applied to any image that you open in camera raw. Additionally, you can set the sharpening for specific ISO settings, go to the far right tab in Camera Raw and save those specific settings and give it a name, for instance, "Sharpening - ISO 400". Then, when you have some ISO 400 images that need to be sharpened you just go to that same tab and choose the appropriate set to apply those settings to that image.
Actually, Rory _IS_ correct...
The way you do that in CR 4.x is while Camera Raw dlog is open, click on the preferences icon at the top in the tool bar.
Under Default Image Settings
Select "Make Default specific to camera serial number" if you have two or more camera bodies and you want Camera Raw to keep track of them and then make it's defualt specific to the serial number...and also "Make defaults specific to camera ISO settings" if you ALSO want Camera Raw to tracking multiple settings and set the defualt based on the ISO.
If you have multiple cameras you would have to open a raw from each camera and adjust all your settings that you wish to be set-remember, you don't have to change all of them-just the ones you want to be different than CR's own default-such as calibrate. You would do this for however many camera bodes of the exact same model (this doesn't ally to say a Canon 1Ds and a 1DsMII-those already have different defaults because they are different models).
You would then need to open images that have different ISOs and adjust the settings at those ISOs that you want different such as noise reduction...
Pretty sure the matrix could become pretty confusing if you wanted different defaults for both serial numbers _AND_ ISOs, but you could do it.
Once you have the settings set the way you want on a serial number or ISO basis, (and AFTER selecting the new options in preferences) click the flyout from the Basic panel and select "Save new Camera Raw default".
This is really very powerful...and by serial and by ISO is new for 4.x.
(Pretty sure Thomas will correct me of I got a detail wrong)
Thanks very much Jeff - I owe you a beer if you ever make it to Vancouver Island...
> "by serial and by ISO is new for 4.x".
It is in ACR 3.7. (But I believe not earlier versions).
I can understand default by serial number, but what does default by ISO do for you except perhaps allow for a larger noise reduction adjustment? This can also be done as a preset save. Are there subtle changes in parameters which can be normalized as a default setting? If so, then setting both s/n and ISO as one would be indicated.
Another question: In "Preferences", a check mark is available for "Prefer CR for jpeg and Tiff", but in Camera Raw Preferences we see "Always open jpeg... and Always Open Tiff.... as selectable, which I like better than ACR4. Now, it's confusing to have one checked but not the other. I have both Tiff and jpeg unchecked in Camera Raw preferences but the "Prefer... one in "Preferences is checked.
The result now is that some tiffs still open in ACR, but most don't, unless they both are checked. I haven't experimented with leaving the Camera RAW preferences checked but "Preferences" unchecked, reversing my current settings.
Are you thoroughly confused? I am!
My suggestion: Allow the choice to be made in Bridge as a constant. Sometimes, Open in RAW is lit with a tiff highlight, sometimes not. Forget the presets and give us ongoing selection at will.
Not confused at all. One thing is opening a file through Bridge or ACR hosted by Bridge, and another one is opening a file through Photoshop or ACR hosted by Photoshop.
The creator of Camera Raw knew what he was doing. By all means keep them separate.
As for the default settings by ISO, all one has to do is calibrate the camera at different ISO settings to realize the obvious benefits.
Now, if someone doesn't see an appreciable difference, there's no reason to bother with it. It all depends on what your results show you. It will vary from camera to camera (which is why we calibrate in the first place).
If this were film, the benefits to calibration with different ISO ratings would be obvious and necessary. With digital, well if it is truly necessary, then I worry about the camera design.
So far with the D80, I have mixed ISO in the same job, sometimes adjacent shots when it became necessary to do so, and I cannot tell the difference below 100% view Visible noise). So, I guess I don't need it!
Speaking of obvious, it is far from obvious that the two preferences refer to different hosts, at least to me. They both appear in Bridge, no RAW preference appears in PS.
Since unchecking Tiff in Camera RAW Preferences stopped most tiffs from opening when Bridge is the host, I assume that refers to the host prefs. But "Preferences" has the Control-K hot key, and the assumption would be that that is the Bridge Preference settings.
A detailed explanation here would be most helpful. A bit more than the Creator knew what He was doing!:D
Quips aside, it really isn't all that obvious.
One can, and it seems today one can only, experiment to see what's what and not depend on anything authoritative for reference. Trouble is, you can get it seriously wrong that way. Witness all the installation and operational problems showing up in the other threads.
It's the "Toss them to the wolves" syndrome so prevalent in today's technological world. Things have rapidly moved from merely complicated to Complexity. Few know what to do with it and many that do are protecting turf.
Jeff and Thomas are doing a great job helping sort this stuff out. Thanks to all who do! :-)
Off to my other work.
I understand that you've put some good work into the new sharpening, and I DO like the new options... But I still prefer Smart Sharpening (deconvolution), and its sliders for controlling halos. Even with the new options in CR 4.1, I still can't control halos as well as I'd like, and I think this is a key for what is called 'capture sharpening'.
I look at it this way: By having the two sets of preferences affect different workflows (one being opening a file through Bridge or ACR hosted by Bridge, and another one opening a file through Photoshop or ACR hosted by Photoshop), I get the choice of opening a JPEG or TIFF in ACR or directly by Photoshop without having to change preferences for one or the other. Just use the contextual menu or double click on the file.
It it were a single, global preference, I'd have to go into Preferences and change them according to how I choose to open a given file.
Smart Sharpening was a definite improvement over USM, but neither of those two, or the new sharpening options in ACR 4.1 give me nearly as much control as PhotoKit Sharpener does.
In other words, I agree with you as to Capture Sharpening. PhotoKit Sharpener excels at Capture + Output sharpening.
Different strokes Ramon. I suspect that we both could have our options, if it were made available.
We can always ask!
The bad part for me with PhotoKit is it only runs in RGB. That was a large file size cost to me for sharpening grayscale, which at the time of it's intro, was all I was doing. I spoke to perhaps Bruce at the time; they agreed it was something to consider and asked if I would be willing to look at a beta of that.
It never happened.
I haven't looked lately at it's configuration.
>Smart Sharpening was a definite improvement over USM, but neither of those two, or the new sharpening options in ACR 4.1 give me nearly as much control as PhotoKit Sharpener does.
That may be because you do not know how to use the tools in Photoshop to control sharpening. If you refer to Bruce Fraser's Real World Sharpening book, he explains his work flow. So far as I know PhotoKit sharpener adds no tools of its own, but makes use of those already in Photoshop. If you master Bruce's book, you should be able to get more control than with PKSharpener, albeit with a lot more effort. You could exercise more control over the radius and amount of the USM and have more control over the edge mask. Of course, you wouldn't have the PK presets with the magic numbers Jeff Schewe talks about, but Bruce gives general guidelines in the book.
>If you refer to Bruce Fraser's Real World Sharpening book, he explains his work flow.
Which Bruce himself, Jeff and the other gurus at Pixelgenius built into PhotoKit Sharpener.
Just your usual ********, Janes. Too bad I didn't realize who was the poster before I bothered to type a reply. :/
>Just your usual ********, Janes. Too bad I didn't realize who was the poster before I bothered to type a reply.
A well reasoned reply refuting all my assertions in a tactful and dignified way, Castaneda. :)
I will not consciously bother with you, monkey man.
Guys, do me (and Bruce) the honor of canning your disagreements from this thread, ok? I really hate to have bad vibes surrounding stuff that Bruce worked on, ok...
I would appreciate a lucid explanation of the interactions of the choices between the various Preference settings, two in Bridge and one in PS. Having that many seems to muddy the waters.
Huh? I think Lawrence is continuing a discussion from the Photography forum re: preferences.
Back to the topic at hand, as Olaf says, and as I've mentioned above, there's not enough control of the halos in CR in my opinion. I've just tried sharpening a series of lightning shots, and as you might imagine, there's a lot of contrast in there. Sharpening in CR leaves a black halo around the lightning. If I try to compensate for it, the rest of the image does not receive enough sharpening.
In Smart Sharpen, we can control the sliders more or less like blend-if sliders in advanced layer blending to control the halos, and as expected, these images sharpened up fine with Smart Sharpen.
Mathias Vejerslev wrote:
> I understand that you've put some good work into the new sharpening, and I DO like the new options... But I still prefer Smart Sharpening (deconvolution), and its sliders for controlling halos. Even with the new options in CR 4.1, I still can't control halos as well as I'd like, and I think this is a key for what is called 'capture sharpening'.
I'm glad you brought up the topic of deconvolution, and I think it deserves further discussion.
Deconvolution is conceptually preferable to USM, since deconvolution actually restores detail, whereas USM merely gives the illusion of increased sharpness. Roger Clark compares the results of the Richardson-Lucy deconvolution algorithm to those of USM on his website.
The problem with the deconvolution techniques is that the mechanism responsible for the blurring, described by the point spread function, must be known prior to the application of the deconvolution as noted on Wikipedia. In his post Roger uses a point spread function (PSF) represented by a 7*7 kernel in the Richardson-Lucy algorithm of ImagesPlus, but he gives no information on how he determined those values. When the point spread function is unknown, some newer algorithms attempt to deduce it through a process of iteration.
The Richardson-Lucy algorithm was used by NASA to restore images taken by the Hubble telescope prior to its repair during a shuttle mission. The problem was spherical aberration and an appropriate PSF could be chosen.
The smart sharpen deconvolution of Photoshop allows one to select from three point spread functions: Gaussian blur, lens blur, and motion blur. The main hurdle on the use of smart sharpen for most of us is that we don't know how to adjust the parameters for optimal results. In his Real World sharpening book, Bruce Fraser showed us how to greatly improve the use of the USM filter, but no such resource for smart sharpening is available. In his initial experimentation, Bruce was not blown away by smart sharpen, but perhaps he did not have enough experience with it at the time he wrote the book.
Other deconvolution filters are available, such as Focus Fixer. Their website contains a demonstration comparing their filter to smart sharpen and USM. They promise to publish a paper entitled "Has USM had its day?", which should be interesting.
Deconvolution is beginning to be widely used in microscopy and this link demonstrates some of the results that can be achieved. However, the discussion gets heavy pretty fast, getting into such topics as the Fourier transform, Wiener deconvolution, Regularized Least Squares, Linear Least Squares, and Tikhonov-Miller regularization.
The purpose of my post is to encourage further discussion of these deconvolution techniques that few of us understand well enough.
Your new book, "Real World Camera Raw with Adobe Photoshop CS3", that will be coming at soon, will that be addressing the latest version of camera raw, ACR 4.1?
Also do you think someone like yourself will be producing an updated book of, "Real World Image Sharpening with Adobe Photoshop CS2," to address CS3, or do you think most of what's in the CS2 book can apply also to the CS3 version of Photoshop?
I'm glad you took my bait :-)
What exactly were Bruce's arguments against Smart Sharpen? I mean, he must have known the advantages and pitfalls of deconvolution sharpen. And I imagine he also used the blend-if sliders for the Photokit sharpeners - a method I also employed before Smart Sharpen (often with a High Pass layer).
I've described a very simple method of employing the blend-if sliders to control halos with USM in this Quick Tutorial.
I also described the method I like to call 'Contrast Coating' - analogue for 'Clarity' slider in CR - and 'Contrast Masking' here.
Most all of this stuff is the legacy of Bruce Fraser as far as I'm concerned.
I like tech - I really do - but I'm a visual guy first, and I judge my sharpening not on the algorithm, but in the real world. And I have to say that I have not used USM for sharpening since I learned to use Smart Sharpen. Nowadays, I mostly call up USM for 'Contrast Coating'.
Exactly right, Mathias. I have USM set to a default Contrast Coating and have a hot key for it.
I spent some time a year or so ago with Richardson-Lucy and found it as Bill Janes explained it. Without the proper knowledge of the functions on which R-L operated, the results varied from some sharpening to "Huh? did anything happen?"
The restoration of detail is really a resolution problem and sharpening, as I understand it here, is analogous to optimizing rise time in oscilloscope measurements. If it introduces overshoot or undershoot, the measurement would be considered invalid (speaking of optimizing the measurement capabilities)
OTOH, resolution is a function of bandwidth,, and so it becomes intimately related.
From my point of view, with RAW at it's current capabilities, I would want to retain all the detail from the original supplied in RAW. Algorithms extending that detail so small items might be identified is of less importance. If I want that kind of detail retention, I'll go back to 8x10 film!
Mathias, I only brought up the preferences here because they were here and I want to encourage that to be resolved. I don't expect to see it in this thread.:-)
Thanks, all, for the links.
"Your new book, "Real World Camera Raw with Adobe Photoshop CS3", that will be coming at soon, will that be addressing the latest version of camera raw, ACR 4.1? "
You bet...since I knew Camera Raw 4.1 was gonna be a pretty BIG upgrade I told the publisher we're just gonna have to wait...
""Real World Image Sharpening with Adobe Photoshop CS2,""
That book only came out in Sept 06 and I think it's still "fresh"...and there's really nothing in CS3 to alter the "equations", ya know?
Hey thanks, Jeff I'll be putting your new book on my list.
That's one of the nice thing about PS CS3, most, if not all the books and tutorials written for CS2 can be applied to CS3.
> ""Real World Image Sharpening with Adobe Photoshop CS2," That book only came out in Sept 06 and I think it's still "fresh"...and there's really nothing in CS3 to alter the "equations", ya know?"
Not the equations. But 2 things in particular would probably appear in a new CS3 version of that book:
1. Smart filters. Before I bought that book, I got into the habit of doing sharpening and noise reduction and blurring using smart filters.
2. ACR 4.1 Detail tab. (And eventually the equivalent in Lr 1.1). What is the relationship with Bruce's capture sharpening?
My current intention is to do my capture sharpening in ACR 4.1, then migrate it to Lr 1.1. I am trying to deduce the equivalent in my thread:
I would LOVE to see some sort of mapping! I'll never be able to do an authoritative one myself, so I need one from you or someone else. It may be in a "Camera Raw" book rather than an "Image Sharpening" book. But I would hate to buy that book and find that it didn't bridge the gap between those two topics.
>What exactly were Bruce's arguments against Smart Sharpen? I mean, he must have known the advantages and pitfalls of deconvolution sharpen. And I imagine he also used the blend-if sliders for the Photokit sharpeners - a method I also employed before Smart Sharpen (often with a High Pass layer).
Well, with the Gaussian blur PSP function, he did not find that smart sharpen had much to offer over USM. For removal of motion blur, there is no similar function with USM, so smart sharpen would be the logical choice in that case.
That leaves the remove lens blur mode. He thought that smart sharpen worked well for small files such as would be used for on screen viewing, but that it was too slow for print sized files. It is not clear to me if he used the blend-if sliders to limit sharpening of the shadows and highlights, but rather he focused on the fade controls that are available in the advanced mode in the shadow and highlight tabs. He did not think that the amount of control that he needed was available. He also made no mention of and use of edge masks with smart sharpen.
>I've described a very simple method of employing the blend-if sliders to control halos with USM in this Quick Tutorial.
>I also described the method I like to call 'Contrast Coating' - analogue for 'Clarity' slider in CR - and 'Contrast Masking' here.
Those examples are very helpful. Similar tutorials on the use of smart sharpen would also be helpful.
"2. ACR 4.1 Detail tab. (And eventually the equivalent in Lr 1.1). What is the relationship with Bruce's capture sharpening? "
CR 4.1 sharpening embodies Bruce's concept of capture sharpening. The detail slider adjusts the halo suppression and the masking allows concentrating sharpening on the edges.
As of now, YOU will need to figure out the appropriate settings for different cameras (MP density) and image types (edge frequency) on your own...
I would suggest a working radius of between .8 and 1.2 for 10-16MP cameras. If you are working on "landscape images" (with lots of high frequency texture) try a detail of around 40-50 and no masking.
If you are working a portrait try a lower detail, say 15-20 and masking of 50-60.
Then bring the amount so that at 100% (1:1) the sharpening "looks good" (meaning no over/under sharpening no ugly stuff-deffinately not "slightly crunchy).
A lot of this is just starting to be determined...this is a new tool after all. And it doesn't-with all 4 sliders in play-really act like anything out there. Yet...Lightroom will bring all this to the table when it ships.
Use the Option/alt technique to see what the sliders are working on...remember that Detail will slide between no halo suppression (100) to max halo suppression (zero) and masking will slide between zero masking (zero) and almost total masking (100).
Getting the right combo of detail and masking will be image size and image content sensitive...then adjust the amount to taste to get it to look "good" at 100% zoom.
That's what I've been doing...and it seems to work well with PhotoKit Output Sharpening....
I hate to ask this dumb question but, what are the "blend-if" sliders?
In Photoshop, it's the Layer Properties that allow you to modify the way one layer blends into another layer (the sliders at the bottom of that dlog).