Untitled Document

Hi,

On page 385 figure 16.4 of PP5E, Dan shows an image that he describes as a "natural" open in Camera Raw. I believe that by "natural open" Dan means with all of the "auto" boxes unchecked and with the curve set to linear.

So I unchecked all auto boxes and saved the unadjusted image. I then proceeded to set the LAB curves as Dan describes in the book. When I set the curves in the Lightness channel, I blew the highlights in the image out. So I went back into Camera Raw and noticed that even though the Brightness setting was unchecked, the brightness was set to 50 by default. When I set brightness to zero and then applied Dan's curves to the Lightness channel, the image looked fine. So I assume that Dan's image had the brightness set to zero as well.

So, here's my question. When using Camera Raw, should you change the default for Brightness to zero or leave it at the default value of 50? I always assumed that the Brightness setting was set to default at 50 because it represented a neutral uncorrected brightness setting, kind of like leaving the middle levels setting in Photoshop set to zero. Is that correct or is zero the neutral Brightness setting?

Thank you for your help?

John Arnold

___________________________________________________________________________

Hi All,

May I also ask that to get the most original capture, should I do the following Camera Raw settings :-

White balance = As shoot; Exposure = 0; Shadows = 0 ; Brightness = 0; Contrast = 0; Saturation = 0; Cutrve = linear ?

Thank you in advance !

Wai-hong Chung from Hong Kong

___________________________________________________________________________

I'm working exclusively with the new Camera Raw in CS3 Beta because it has vastly expanded capability compared with Camera Raw in CS2, hence what I say here applies to CS3 in particular, but may be generally valid. Brightness can range between -/+150. As you adjust, it shifts mid-tone values in a non-linear manner much like the grey input slider in Levels. The default setting of 50 is Adobe's guess about a setting that would be a liveable starting point for image correction in Camera Raw. It means nothing more than that, and in particular there is no such thing as "neutral" when it comes to Brightness. I generally find it one of the least useful correction tools in Camera Raw and hardly ever use it - especially since Curves became part of the Camera Raw arsenal with CS2, and all the more so with the Parametric and Point Curves options in the new Camera Raw CS3 Beta. I have no idea where Dan started from in the image to which you refer - I'm merely making the point that "default" settings should be treated as starting points and they are only as useful as the images require.

Mark Segal

___________________________________________________________________________

Wai-Hong,

Let's start with the approach - what are you trying to do - what the camera saw and what you want it to have seen can be and often are two different things. So Camera Raw is one tool in the arsenal for making the image look like what you want it to look like. If you accept that, it follows there is nothing religious about the original raw capture itself except that you want to have set the exposure to maximize the amount of information without clipping. Then it follows from that there is nothing religious about the default Camera Raw settings - it is all a matter of your approach and what is the most convenient starting point for image correction. For example, in my case I set my defaults as you have them below except Brightness is 50 and Contrast 25 only because they are workable starting points for most images. I then generally leave these alone and use Parametric and Point Curves, Recovery and Shadows (CS3 Beta) for dealing with contrast and luminosity issues. I like

starting the life of the Curve at Linear like you suggest below, simply because I don't want or need Adobe's suggestions about initial image contrast. CS3 Beta has tons of stuff for colour adjustment - the Vibrance slider in the first tab plus multiple options in the H tab being the most interesting.

Mark Segal

___________________________________________________________________________

The Brightness setting in ACR (and LR) is not what you©ˆd think of when you think of Brightness in terms of Photoshop. This is due to the linear nature of raw data where half is in contained in the first stop of highlight data and due to the design. You want to set either end of the tone using the Exposure and Black sliders (use the option/alt key when sliding). New in ACR that you will not see in Dan©ˆs book is Fill light and Recovery. You should do ALL corrections in ACR from top down, left to right! Set Exposure, Recovery (for highlights), then Fill light and then Blacks. Brightness is a minor tweak that produces an S shapped curve OVER the above edits due to the editing order. The Œdefault©ˆ is 50 out of the box but you can set it to anything you want and make a new Camera Raw Default of course (meaning, it©ˆs tough to know what a default is other than an initial setting when the product is first used). Contrast is simple (make the blacks/whites blacker or whiter).

In most cases, you©ˆll never need to do much to Brightness and if you do, it©ˆs going to be a tiny tweak. Work with the four main tone sliders. Then you might want to consider the Parametric curve.

ACR is going to radically change compared to the old version you©ˆre reading about. 3.7 is public beta, 4.0 is private and a major change over 3.X!

Andrew Rodney

___________________________________________________________________________

The +50 brightness setting is a default setting that Adobe came up with to emulate the cameras own internal processing for rendering Jpegs. Just about every camera manufacturer lies about the true sensitivity of their chip. There good reasons for this (Dan alludes to these reasons in his book) – by lulling users into underexposing they insure , somewhat, against the possibility of clipping highlight values. How you interact with any default depends on how you routinely expose your images. If you are going to be fully correcting in Photoshop anyway, it makes sense to default to a totally flat, un- enhanced, raw process setting (every slider set to zero) so that you are dealing with as much of the raw data from shadows to highlights as possible. I always test my camera at these settings to find the ideal ISO for exposure of the chip and thus end up with the best possible data. The camera meter, however, is almost always set up for an ISO where there has been some level boosting post capture so you have to cheat the meter or enter some kind of custom compensation (bracketing) to get around it. I generally use a hand held spot meter for serious work and use the camera in manual mode.

regards,

Lee Varis

President, LADIG

Photographer and Digital-Photo-Illustrator

___________________________________________________________________________

Andrew,

Thank you for your explanation. So is setting Brightness to +50 Adobe's way of compensating for the linear nature of raw data? That's kind of the conclusion I am arriving at. And this is probably a question for Dan, because he is the one who applied the curves in the book. But I am wondering why he chose to substantially darken the image (at least that's the way it looks to me)and move away from the ACR default, even though the default initially appears to IMHO look like the better exposure where the overall brightness level of the image is concerned?

I'm probably overanalyzing the situation. But if there is some kind of best practice behind the move, I would like to know what it is.

Thanks.

John Arnold

___________________________________________________________________________

Mark Segal" wrote:

Mark,

Thanks for the reply. Do you or does anyone else know "why" they have chosen +50 as a liveable starting point? As Dan mentions, they use +25 as a starting point for Contrast because most images require an expansion of the range in the midtones. I was wondering if there was also some kind of rationale for the Brightness setting. My concern is this. Does leaving brightness set at the default of +50 in essence apply a curve to the master channel and if so, shouldn't we avoid that based on Dan's comments in chapter 16?

It means nothing more than that, and in particular there is no such thing as "neutral" when it comes to Brightness.

I would agree, neutral is a relative term when speaking of brightness. However, I remember reading in the late Bruce Fraser's book "Real World Camera Raw" that cameras do a "linear" capture of data when it comes to brightness, and that it did not correspond well to human perception. So I am wondering if Adobe's +50 is an attempt to show that they have applied some kind of gamma correction to the data. Whereas leaving it at zero would possibly be the equivalent of a linear capture?

I generally find it one of the least useful correction tools in Camera Raw and hardly ever use it - especially since Curves became part of the Camera Raw arsenal with CS2, and all the more so with the Parametric and Point Curves options in the new Camera Raw CS3 Beta. I have no idea where Dan started from in the image to which you refer - I'm merely making the point that "default" settings should be treated as starting points and they are only as useful as the images require.

I too have no idea where Dan started from. I just noticed that when I applied the LAB curves as they are presented in the book, that I blew the highlights out completely. It was only after I set Brightness to zero that Dan's Lightness channel correction made sense.

Thanks again for you thoughts on the matter.

John Arnold

___________________________________________________________________________

Lee Varis wrote:

So in other words, are you saying that you should test your camera by shooting some test shots and then determine how much in-camera bracketing is required to get what appears to be a correctly exposed image when all ACR settings are set to zero?

John Arnold

___________________________________________________________________________

On 2/3/07 8:55 AM, "John Arnold" wrote:

To begin with, there is a difference in the Brightness settings between older versions of ACR (undoubtedly the copy used in Dan©ˆs book) and newer versions (3.7 forward) since there needed to be consistency with ACR and Lightroom. Also note, in these newer versions, you can edit non-raw files like JPEG and, very important, the settings used are applied differently to existing rendered files versus Raw files. An extra negative range is required to edit JPEG/TIFF files, and DNG files created from JPEG/TIFFs. So whatever you©ˆre reading is probably out of date or will be real soon now (sorry).

Yes the default (what is better known as the Neutral Starting point) is set to +50 primarily due to the fact that a zero setting produces too dark a rendering with most raw images. The default state of all converters will be different of course. The idea is to provide a fairly decent rendering as a starting point for you to now work on the image. You©ˆll notice that in ACR and LR there©ˆs a setting called Auto which is not the Neutral Starting point but rather, much like the rendering in-camera, a guess as to what would produce a pleasing image. You can use that, or Neutral or set anything you like as a new ACR default. According to Mark Hamburg, the +50 = one stop and this neutral staring point is necessary for both a good looking starting point and to work with the controls above it (primarily Exposure, Blacks etc).

Additional changes from the book, Brightness and Contrast controls are moved upstream of the parametric curve in the processing pipeline. You may notice there are two Histograms provided (main and in curves). The B&C controls are reflected in the background histogram in the parametric curve, but not the tone curve graph.

As mentioned earlier and by others working with the product, leave it alone unless after working with the controls above it, which are placed there in that order for a reason, they doesn©ˆt produce a tone curve you desire such that a subtle S shaped curve is needed. I don©ˆt know what Dan©ˆs talking about here (I don©ˆt have the book) and the bottom line is, whatever he©ˆs talking about is old news since 3.7 is available now with far greater control, a totally different imaging pipeline and version 4.0 is coming real soon with even more rendering options. All this applies to Lightroom as well.

There is simply no correct way to render the images other than following the controls in the order they are provided. You are working with scene referred data and using this converter to produce output referred data. There are no rules in any of this other than make the image appear as you wish (again, doing so in the order provided or you'll just chase your tail in getting the preferred rendering).

Contrast is the last setting you should ever need to touch based on it's order in the processing pipeline in ACR. Again, top down, left to right (using the various panes).

Contrast in ACR isn't linear like we have in CS2 but rather another S curve and affected by Brightness so again, work there first. And unlike Photoshop, altering these controls doesn't affect color as you'd see in Photoshop rather just luminance (another advantage of doing all this on Raw data).

Not zero across the boards. Here's what has worked well for me. I take a precise external light meter (a Minolta Flashmeter III) under controlled lighting and setup a Macbeth Color Checker. I know the meter is accurate (within 1/10 of a stop) and get a incident reading at ISO 100. I then set the camera to shoot a bracket at the minimal amount shooting 2 stops over, 1 stop under the recommended F-stop. I bring all images into ACR and set Auto off (the old Neutral Setting again). Now I examine the white patch and look for an exposure in the group that is as close to clipping without doing so. In the case of my Canon 5D, I found that actually shooting at ISO 100 provided the correct values above but if it's off, I simply set exposure compensation on the camera to produce that exposure. You want to expose for the highlights in digital due to the linear capture. Ideally you want to exposure to the right, that is, as close to clipping a highlight that isn't specular without clipping as possible. Note that the Macbeth white isn't as white nor specularly neutral as I'd like so I'm also using a white tile that is ( http: //www.babelcolor.com/main_level/White_Target.htm). With this pup, if I can get 245/254/254, I'm in really good shape.

Now the issue is correlating the meter in the camera with the scene since such meters are kind of dumb, thinking everything they 'see' is 18% gray. Using a spot meter mode in the 5D and understanding this allows me to handle exposure pretty well. So for example, if you know have your camera meter ISO nailed using the above technique, if you point the meter at say a white dog on snow, the meter will under expose the scene by about 2 stops, you simply compensate. The key is nailing the ISO in the camera meter, then understanding what it's looking at as 18% gray and if necessary, compensating for this. Of course you could carry around an external incident meter and just use that as your exposure guide. But at least you know the chip sensitively and correct ISO to expose for the highlights.

Andrew Rodney

___________________________________________________________________________

John Arnold writes,

No, I didn't darken it. Camera Raw wanted to substantially *lighten* it, and I did not give it permission to do so.

An inexperienced person attempting to correct an file often creates problems for the next person who has to handle it. This is why a professional retoucher will always ask for the *original* file to work from, rather than something that somebody else has attempted to improve. The "corrected" version probably looks better than the original does--but the person who did it likely engineered in some problem that makes the file more difficult to improve than if we started from scratch.

The same thing can happen when acquiring a digital image, whether from a camera, Camera Raw, or any other acquisition module. Artificial intelligence often attempts to "improve" the image before we see it, by forcing a white and/or black point and by increasing midtone contrast at the expense of highlights and shadows.

These features wouldn't be in there if they didn't work most of the time, but in a lot of images they actually make matters worse, such as when highlight and/or shadow detail is critical, or where there are bright colors that have to retain shape. In those cases, they may make subsequent correction more difficult (as in the image you're talking about) or even, IMHO, impossible (as in the flower image in the same chapter).

The utility of raw capture modules is that they can bypass these "corrections" when necessary. To take advantage of them doesn't require a recipe--I use a zero setting as a matter of convenience. It just means choosing very conservative numbers to ensure that the image isn't damaged when you open it.

If we're fortunate enough to have a neutrally correct (or nearly so) capture, then opening the endpoints in Camera Raw won't hurt anything. But if there *is* a cast, then opening the endpoints makes life unnecessarily difficult. If we do, in addition to making the endpoints lighter and darker, Camera Raw forces them to be more neutral, a bad idea.

Yes, if you compare the two images as if they were final products, the one with the full range looks better. But that's not the object of the game. The question is not which one looks better *now*, but which one looks better after we've corrected it. The choices are starting with a relatively flat image that has a uniform cast at all levels of darkness versus a relatively contrasty one whose highlights and shadows are neutrally correct but is wrong everywhere else. The first takes seconds to fix and no great skill. The second looks better now but is rather difficult to improve further.

Dan Margulis

___________________________________________________________________________

"John Arnold" wrote:

What Dan said is he ZEROED everything in ACR and then open the image in PS.

Alcides Pomina

___________________________________________________________________________

Andrew Rodney wrote:

This makes sense to me now that you mention it. In other words, if you reset endpoints etc., some sort of brightness move will have to be made.

I didn't know that the Brightness slider doesn't affect color. That makes it a lot more of a desirable control.

Thanks very much for the great explanation. It shed a lot of light on the matter.

John Arnold

___________________________________________________________________________

Andrew,

Like all recipes that may have some general validity, this one should not be followed slavishly. For example, beyond the White Balance, the Exposure slider would generally not be my preferred starting point. I find most images have problems that need to be more specifically targeted than possible with the Exposure slider. I find myself dealing with highlight and shadow values using Recovery, Fill and Black before tinkering with Exposure. And if there are "Exposure" problems, I find I can generally solve these better in terms of overall tonality by going straight to the "T" tab and working on either the Point or Parametric Curve, depending on the nature of the correction to be made. I agree with you that the main objective is to make the image appear as desired, but I haven't found myself chasing my tail by disobeying the "order provided".

Mark Segal

___________________________________________________________________________

On Sat, February 3, 2007 3:43 pm, MARK SEGAL wrote:

I haven't spent much time yet with the new ACR, but I tend to work "top to bottom" as well. I fix the color temperature and exposure, then work my way down. Slavish? Nope - just weems to work well. If I don't fix exposure errors first, I seem to end up in a loop.

I'll be spending a lot of time with CS3 and Lightroom the next two weeks -I'll see if my habits change given the new tools. For me, setting the exposure and shadow is analogous to setting the black and white points on a scanned image. It would seem strange to not adjust these first.

--Rich Wagner

___________________________________________________________________________

On Sat, February 3, 2007 1:27 pm, alpom111 wrote:

I still don't understand the rationale for doing this. You have a 16-bit, wide-gamut internal working space to optimize the image in before converting it to an output space (of your choice). Any corrections possible should be made in that space (with the desired output space selected when making corrections) - that's what it was designed for. PS should be used for corrections that are not possible in ACR.

--Rich Wagner

___________________________________________________________________________

I appreciate all the responses to my question. All lot of good answers. The one that comes closet to what I had in mind was written by Lee Clawsen.

I am in the fortunate position of being able to hire someone who is not expert, or even particularly experienced, in Photoshop. I'm willing to train him or her. But I need to have someone with a good eye, and I don't know how to train someone to "see" what I want them to see. I can explain color, contrast, perspective, etc., to some degree but that only goes so far.

There are (at least) two problems finding a person like that: I don't know how to do a proper evalution, and even if I did, an untrained person might not have the knowledge of how to express himself regarding the image concepts I'm trying to evaluate. I think I may be looking for the type of testing a psychologist might do.

In the meantime, I'm going to start using Lee's method.

Jonathan Clymer

___________________________________________________________________________

Rich,

I think what I hear Dan saying is that many of the controls in ACR are somewhat primitive compared to Photoshop at this point because they do not allow for channel by channel corrections. As a result, even though ACR delivers a 16-bit wide gamut working space, the potential casts that could be introduced by using ACR in the areas that Dan cautions us to avoid has now created problems that could, depending upon the image, be far more difficult to correct than if you had just taken the image in it's completely neutral zeroed out state, and worked with it from there.

John Arnold

___________________________________________________________________________

Rich, it no doubt depends on the image, and the user. Different courses for different horses. The end image is what matters to Dan, not so much how one gets there (my perception is that Dan thinks the market hype on camera raw is overstated). A less skilled user of Photoshop than Dan, may get better results doing most of the work in the camera raw converter, than in Photoshop (even more so with later raw converter software versions). Dan may get better results with a "flat" zeroed image, despite the theory saying that his non linear gamma encoded 8 bpc small gamut RGB edits and lossy LAB/CMYK conversions will lead to an inferior result. Dan can make a silk purse out of a sow's ear - a zeroed raw conversion may be nowhere close to a pig.

Back in scanning, there were two main camps - those that did as much work as possible when setting up the prescan, or those that scanned flat with headroom and edited in Photoshop using a custom scanner profile as input to an editing space. Scanners even introduced a 'digital negative' or archive that one could use the scanner software on, even after the original flat raw scan.

It may not be fair to compare a high bit wide gamut scanner RGB editing workflow with linear high bit digital camera raw images, but some do use a similar general approach.

Obviously Dan would take advantage of the RAW converter where necessary and where it offers benefits such as highlight recovery, there is nothing stopping one from combining various flat or tweaked raw conversions into a composite in Photoshop (just as is common for bracketed exposures).

As mentioned in recent posts, automated corrections can make later edits harder. Although raw conversion edits may be human performed and not as bad as automated ones, in some images and edits Dan may get better results with a flat image. Your mileage may vary!

P.S. An 'output space of your choice', as long as it is one of four hard wired profile choices in ACR and not any installed profile as in Photoshop proper (better than the model T option I guess).

Sincerely,

Stephen Marsh.

___________________________________________________________________________

On 2/3/07 3:43 PM, "MARK SEGAL" wrote:

You are more than welcome to work with the product in a method that isn©ˆt recommended by those who built it. The bottom line is, can you render the image as you wish? But usually its a good idea to handle the big imaging issues first. Yes, white balance is the recommended first correction and Exposure is the way to set the white clipping after which, you can use the recovery to bring back any clipped white values that may have clipped in the process or exposure assuming one of the three channels has any data. Try that with an existing rendered image; ain't going to happen!

The highlight recovery, the new Fill Light and all the other controls below Exposure ARE based on the settings you apply in Exposure. Operating in a different order can work but you will likely find you©ˆre taking two steps forward, one step back due to the processing order of the edits in the two converters being discussed.

You may or may not chase your tail depending on the order you apply the edits and the degree of the edits. For example, it is generally suggested that you fix white balance before exposure or a color cast before applying a saturation adjustment. That©ˆs usually the case with rendered image corrections in Photoshop as well as in a raw converter. But If you can produce a desired rendering by working backwards, so be it. It©ˆs just a bit more work to adjust say saturation before setting black and white point or fixing a color cast. That isn©ˆt to say you can©ˆt do this, it©ˆs usually a lot more work.

Thomas Knoll built the tools in a specific order but if you find you©ˆd prefer to work differently, by all means do so but be aware that corrections are happening in a fixed order in ACR and Lightroom.

On 2/3/07 5:25 PM, Dan Margulis wrote:

Considering that raw data has no color, its Grayscale data, I don©ˆt understand how a cast could be an issue unless you make a cast based on your rendering decisions. The recommended first correction in ACR and LR is white balance the correct or 'neutral' setting being whatever you wish it to be. The white balance at capture has absolutely no bearing on the white balance of the rendered Grayscale raw data. It©ˆs a suggestion you can apply or completely ignore. The ONLY area where the raw data is affected is ISO and exposure.

Corrected it? You're building a color image from data that has no color. The best, fastest and most flexible approach is to do all the heavy lifting at the raw conversion process because you©ˆre dealing with linear encoded data, the corrections are totally non damaging and applied in high-bit and you can change your mind about the rendering as many times as you wish (look at Virtual Copies once you get your hands on the final version of LR or when using ACR with Smart Objects). That isn't the case with a pre existing rendered image.

Try this: make two renderings of the same raw, one for shadows, one for highlight and even play with differing white balance. Drag and drop the rendered images in Photoshop on top of each other (use shift key) so they are in pin registration. Double click on top layer and play with blend if options using feathering. You'll produce a vastly superior tonal range than one flat raw conversion with excessive Photoshop edits for tone. Here's where you want to now use Photoshop until ACR/LR has such blending layer options (it will come in time).

It doesn©ˆt have to be an either or situation. Setting ACR or LR to produce a flat appearing image using fast, metadata corrections to then have to Œfix©ˆ it as a full resolution pixel based image in Photoshop is totally non-productive for anyone working with raw data, certainly if they have dozens of similar images. This Œfix a big image©ˆ one pixel at a time in Photoshop is very 20th century thinking and it might appeal to a few but not anyone shooting seriously who has to render lots of images. Do the big work in ACR or LR, bring a corrected file into Photoshop for the pixel polishing, a process in which it was designed.

A raw image and the process of rendering isn't anything like correcting a rendered image! The tools, the data encoding and the workflow are not at all the same. If you have an ugly transparency, you can attempt to correct the issues at the scan stage but there's only so much one can do. A raw image initially has no color, its scene referred and your job is to produce an output referred image which isn't the same as 'correcting' an image since the rendering options are vastly more variable and powerful.

Producing a flat image in ACR to then tone map it in Photoshop is like setting a scanner in a default mode and doing corrections after the scan instead of tweaking the controls to produce a corrected scan. I don't understand why anyone would do this.

This is a lot like the differences in handling a color neg versus a color transparency. The rendering on the color transparency is baked into the chrome, you only have so much leeway in what you can fix. Not the case with a color neg, you have an infinite number of possible filter packs to handle the color rendering. A raw file has vastly more tone and color possibilities than any color neg.

Andrew Rodney

___________________________________________________________________________

Rich,

As with everything in Photoshop we each use what seems to work best for ourselves, but when a response like this comes back I have a bad habit :-) of saying "now wait a minute, what am I missing here?". As it happens, I'm now processing a large photoshoot I did in Barcelona last October. So I have these 340 raw images tabulated in Bridge and I ran through them all to find one that is truly and unambiguously under-exposed - i.e. not just highlight and shadow problems but really throughout the range under-exposed - because I said to myself if there were a sure-fire reason to use the Exposure correction first, it would be to address true overall under-exposure. Interestingly, I could only find ONE of 340 images that fit this definition (which indicates I don't under-exposure very much, which is perhaps one of the reasons I seldom need this tool). The under-exposure was on the roof of La Pedrera (Gaudi's Casa Mila) - easy to happen there because the overall brightness is so high that unless one is really careful the camera can be fooled. The image has traces of blue sky, billowing clouds ranging from near-white to below mid-grey, and those beige-ish Gaudi chimneys and vents resembling abstract sculptures (which should occupy a range around the mid-tones). The general objective is to shift the histogram to the right, the binding constraint being without blowing-out highlights. There are at least two ways: Exposure Slider and Curves (Point and/or Parametric). So I tried each starting from "default" settings of Brightness 50, Contrast 25 and Linear Curve. The Exposure slider certainly worked, but by the time the brightest points of the clouds were about to blow-out, the mid-tones were wishy-washy such that tonal modulation of the sculptures was unsatisfactory. So I went back to square-one, opened the "T" tab, went to the Point Curve, grabbed the upper right handle and simply shifted it leftward, steepening the curve until just before the highlights in the clouds were ready to blow. Far better result - nice mid-tones with far better modulation of the sculptures and gorgeous clouds. (Of course this is happening because the curve affects brightness AND contrast simultaneously due to the slope - with less impact on the mid-tones than results from a direct attack with Exposure.) To improve it a bit more I went to the Parametric Curve, slid the break-point between Lights and Darks further to the left of Default (to better target the Darks) and shifted the Dark parameter positive 9. This just breathed a little more life into the three-quarter tones. A happier image.

I described this in a bit of detail, because it is typical of the kind of outcomes I have discovered with this workflow. Just after Christmas I was processing a wedding shoot (a favour for a relative) where photo-flash produced all kinds of exposure issues that were challenging to deal with, and here too I found the Exposure slider on the whole less useful than going straight for the "T" tab.

I've taken an interest in Camera RAW CS3 workflow, so I'm now starting an archive of screen-captures showing "what happens when". If you would like to see the three screen captures of the La Pedrera shot in ACR (original, Exposure Correction, Curves correction), just send me your email address.

Mark Segal

___________________________________________________________________________

The clinchers are "primitive" and "at this point". "At this Point" is no longer CS2, because the new ACR is so much richer than its predecessors that all the books about this need to be re-written. Lightroom and the new ACR offer so much refined control over colour and tonal correction that they really require us to re-think important aspects of workflow.

Mark Segal

___________________________________________________________________________

Andrew,

Thanks for your permission to work with the product in a manner that isn't recommended by those who built it, but as the product is in Beta and has no instructions on paper or on-line, they haven't explicitly recommended anything about workflow for this version of the product. That much said, you are probably correct one can infer that they intend this workflow from how they set-up the tools - at least as far as the logic of handling the biggest-picture adjustments first. I don't believe. however, this necessarily means - or even they necessarily think - that other workflows are inferior. I can ask Thomas about this next time I see him, but until then my experience and common-sense prevails.

Whether I use the "T" tab or the Exposure slider first, I am handling the big issue that comes next after White Balance (on that we agree), except in a different way, and I have NEVER YET chased my tail - at least not in photo image editing - would that the rest of life were so straight-forward! :-)

The ostensible evidence fromm trying to do similar things in different ways indicates to me that the corrections are happening in the order I implement them, regardless of how the tools are laid-out in the inter-face.

Mark Segal

___________________________________________________________________________

I couldn´t agree more. For instance, on some images, I prefer to set the contrast at zero, or near it, and increase the contrast by moving the Shadow slider. Some images come better this way, some not. In conjunction with some moves on Brightness it´s a way to have more control over shadow contrast, instead of the general Contrast slider.

Regards,

Rodolpho Pajuaba

___________________________________________________________________________

On 2/4/07 9:11 AM, "John Arnold" wrote:,

Actually many of the controls in ACR/LR are vastly more powerful than what you©ˆd find in Photoshop! Some, not all (and vise versa). Take a look at Vibrance as just one example. There's nothing in Photoshop like it.

You may need to edit color channels in baked rendered color images. That's not what's happening in this case. Until you export the data based on the metadata instructions, there's no color, and you can produce as many variations as you wish without damaging (or even creating) a pixel. And do it REAL FAST!

These are different tools that provide different capabilities and someone who's spent too long falling in love with Photoshop's ability to correct images without really looking deep into what raw processing offers needs to take the blinders off quickly. Again gang, raw has no color. You decide what you want the initial color to be long before pixels are built for editing in Photoshop.

Photoshop can be the ultimate turd polisher but the question becomes, why start with a turd? True, you could under or over expose the raw data and be stuck with potentially very poor data. Photographers tend to try not to do this. One problem with raw data for some is, your likelihood of having a trud to polish is greatly diminished if you simply exposure correctly. White balance off? Not an issue. Color cast? You made it based on the rendering controls you affected. Blown out highlights? As long as there's one channel of data, the other two can be rebuilt. Raw files can provide up to an additional stop of tonal data for rendering, a rendered image is what it is.

As I stated, you can pretty much render the Grayscale data to appear anyway you wish IF you understand how the controls work. Also, many of these tools that affect tone do not affect color shifting due to how the raw data is handled. Something that's promoted here as a 'big deal' using LAB. Lastly, anything written about the beta of Lightroom or the old version of ACR in a book you currently have is outdated by a pretty significant degree.

On 2/4/07 9:39 AM, "MARK SEGAL" wrote:

Actually this workflow has been documented almost as long as ACR has been a product. While like Dan's book, it's now seriously outdated, Bruce Fraser's Real World Camera Raw (1st and 2nd edition) discusses this concept. And again, you don't have to use the product as designed but for those reading these posts who may not have your technical abilities with the product, its good advise to START top down, left to right. But as I've said from day one, the beauty of working with scene referred raw data is as long as you achieve the color and tone you wish, anything is fair game. Its fast, its non destructive, its infinity variable.

From purely a technical standpoint in how the product applies the edits, that's incorrect but, if you're happy with the results, that's all that matters. Play around with the various tone controls and look at BOTH histograms and I think you'll see that there are fixed orders to how the edits are applied to the data. That's by design of course.

Yup!

I've been begging for soft proofing in LR since day one and hopefully that will come since you can print (very effectively) out of Lightroom. In fact, it makes printing out of Photoshop seem prehistoric when you work with the templates which can hold every print parameter (paper type, profile, output rez etc). Mac Holbert at Nash Editions is not printing most of his work out of LR due to this ease of use.

However, I'd be hard pressed to edit anything but a virtual copy for output once I do have soft proofing. That is, I'd like to produce and maybe render a master file that's corrected for all subsequent output needs as I currently do in Photoshop, then load a soft proof and edit based on the output device.

What I hope to see in LR (we Alpha's are working hard on new feature requests) is a print history in which you could click on a history state in LR for any printing and corrections for that output device and be done. So it would be like adjustment layers in layer sets per output device but all this is accomplished using metadata instructions instead of a pile of huge, pixel based images. It would save a ton of time and HD space. Obviously if you need to do Photoshop work, you'll have to render the file first. With the new cloning and healing tools in version 1.0, there's a lot of work that can be accomplished solely in Lightroom, never having to be rendered as a PS file. But this is a 1.0 version product. Lets not forget what Photoshop 1.0.7 was capable of and cut LR a bit of slack. Time will come where 80%+ of many users work will be accomplished completely in LR instead of Photoshop.

Resist the urge to render the files in Photoshop just because that's been the way we've worked for 16 years. Yes, this will be necessary but how often? Less and less I'll bet.

Andrew Rodney

___________________________________________________________________________

Mark,

Please send the screen captures - my email address should be at the top of the CT message header, but in case it's not, use Rich@WildNaturePhotos.com I'll be spending a *lot* of time with CS3/ ACR after I get home in 2 days - I'll give your workflow a try. It's certainly possible that you've hit on something that even the designers of ACR may have missed, and there's certainly benefit to experimentation. Because the workflow "made sense" to the engineers in the order they've set things up, they may not have even tried your method. I've spent little time up until now with the new ACR, which is greatly changed from the old ACR, so until I try what you describe, for me, all bets are off.

It is nice to have these new, powerful tools coming out, although it will take some time to really get to know them. And I agree - soft proofing in Lightroom/ACR would be the coup-de-grâce to Photoshop from the photographer's perspective.

Thanks,

--Rich Wagner

___________________________________________________________________________

OK Andrew, I tried what you said (hope I understood you correctly) and I don't see how it shows any fixed order for doing things. Whether I start editing luminosity in the "B" tab or the "T" tab, the histogram adjusts in the expected manner in both places, as one would expect it should. So what have I proven (or not proven :-))?

Mark Segal

___________________________________________________________________________

RAW Developer v1.6 has curves for individual RGB channels (and individual LAB channels, too). It's not clear to me from their documentation whether these individual channel adjustments actually are some kind of white balance movements, or represent post-processing after rendering, or what. Can anyone explain what RAW Developer's individual channel curve adjustments really are doing?

- George Machen

___________________________________________________________________________

On 2/4/07 5:55 PM, "markds0" wrote:

The Histogram in what you©ˆre calling the T tab (which in my version, there are icon©ˆs (don©ˆt ask)) isn©ˆt affected by a change in Brightness slider while the upper Histogram is. There are two different Histograms! That©ˆs due to the order in which the tone curves are applied. If you©ˆre not seeing this we should talk off list because there may be some minor NDA issues going on here (the version of LR and ACR I©ˆm using may not be the same as you).

There are two paths in the engine where tonal edits are applied, before or after the parametric curve that's my point. And this was done by design. You don't have to follow the logic of this approach of course but, there are reasons it was implemented. There's a reason why two proprietary profiles are used. There's a reason why the encoding color space was chosen. You get the idea.

Andrew Rodney

___________________________________________________________________________

Andrew,

Yes, correct, there are two histograms in the T tab - the upper one which shows colour channels and the lower one which is grey. In the B tab there is only the upper histogram which is identical and behaves identically with the upper histogram in the T tab. Now, based on what you say about what changes when - in the version I am using, which I believe is the most recent public Beta, the grey histogram in the T tab does not change when doing point and parametric curve edits in that tab - but the upper one does. If, however, I go back to square one and say increase the Exposure in the B tab, then revert to the T tab, the lower grey histogram has indeed shifted to portray the change made to Exposure in the B tab. All that means to me is that the grey histogram in the T tab was put there as an aide-memoire so that anyone proceeding to the T tab will be reminded where they've reached in the editing trail. But so what? I might be terribly thick this evening (or perhaps always, who knows :-), but I still don't see what this proves about the optimum order of the workflow. We can continue this off-line if you prefer; I made this observation here because I suspect it is generic regardless of version details; but as an Alpha you would know that better than me. I'm not even a Theta - just a User (with two eyes and some scene-referred grey cells :-).

Mark Segal

___________________________________________________________________________

A rejoinder on the workflow issue - and something that hasn't been mentioned yet. At the raw conversion stage we don't have a printer profile. Therefore we cannot fine-tune luminosity in a soft-proof mode. Hence until there comes a way to soft-proof in ACR or in Lightroom, we will still need Photoshop to make the final pre-printing adjustments. This of course is completely consistent with the point made by others that neither ACR nor Lightroom are intended as complete substitutes for Photoshop; however, the addition of soft-proofing capability to either the new ACR or Lightroom, if technically possible, would eliminate one more need to revert to Photoshop.

Mark Segal

___________________________________________________________________________

Mark,

In thinking about this evolution of workflow, the other essential missing component from LR (AFAIK) is sophisticated multi-pass sharpening, as with PhotoKit Sharpener. Unless I can do capture sharpening, creative sharpening, and output sharpening from within LR, I'll be making trips to Photoshop, although all but creative sharpening would be batch processed. If it's not there now, I'm sure it will just be a matter of time until it is. I guess my other thought would be RIP support - I hope the major RIP writers are paying attention.

--Rich

___________________________________________________________________________

Rich,

It wouldn't surprise me in the least to see Photokit Sharpener appear one of these days as a Lightroom module. It is a natural. And you are right - for me as well the ability to use this tool will continue to require a trip back and forth to Photoshop until it does become a module. Fortunately there is flexibility in the Photokit Sharpener workflow about when one applies Capture Sharpening. Where I disagree though is on the point about batch processing. This would only be recommended if all the images had similar characteriistics from a sharpening perspective. If you haven't read Bruce Fraser's book on sharpening yet, let me suggest it is a worthwhile read. It deals with all the principles behind Photokit Sharpener (but not the product itself) and reading this one learns why the various choices the program offers are meaningful and important to set correctly for optimum results. As for RIP support, over time I can see the Print module of Lightroom displacing more and more functions that a RIP provides.

Mark Segal

___________________________________________________________________________

On Feb 4, 2007, at 6:34 PM, MARK SEGAL wrote:

They do, or they don't get batched.

I have it, but have not quite finished reading it. I agree, it is an excellent book, and I'd say it is currently the definitive book on image sharpening. Obviously, I don't batch creative sharpening. If there's any question regarding output sharpening, I run the batch and save the images with layers, then inspect each, adjust as necessary, flatten and save.

--Rich Wagner

___________________________________________________________________________

There has been considerable smoke blowing and NDA threatening discourse going on around working methods for ACR. The basic operating principles of this plug-in haven't changed but that also doesn't mean that new features might introduce new ways of working within the interface. The fact is that the order that you make slider adjustments is somewhat flexible in that nothing really happens until you click Open or Save! Certain slider adjustments have an interdependent relationship however so that Curve adjustments and Exposure/Shadow/Brightness/Contrast affect the image together. If you start with Curves that will affect where you start with the sliders - the "tabs" do not force an order otherwise when you back up to a previous tab the image would revert to a previous state. The is some logic in the way the controls are laid out but the interface is not so fascist.

Everything you do in ACR affects everything else to a certain degree and the biggest problem with ACR is that often its hard to figure out the interactions. When you are in the Exposure/Contrast slider area you can't really tell what's happening to the overall curve except to see the result in the image – if you go to the curve after doing slider adjustments you see a preset shape or a linear curve (depending you how you set your defaults) this curve does not reflect the adjustments you've made elsewhere! You also can't evaluate the adjustments on a per channel basis except by looking at a histogram overlay. There is little precision feedback provided – you can only see RGB numbers from the chosen output profile – no LAB, no CMYK. Hue and saturation moves are similarly intertwined in different "tabs". If you are seriously trying to optimize the image using the controls in ACR you almost HAVE to bounce around, back and forth, from tab to tab and back until you have the look you're after. At this point, again, if you're serious about optimizing the image – you open the image in Photoshop, look at individual channels, read LAB and/or CMYK numbers and figure out where you really are!

Given the limitations of the interface (and please, this is obvious regardless of features like Vibrance) Dan has suggested taking a conservative approach to the adjustments you apply in ACR. ( Andrew, before you jump on this, please READ Dan's book) This makes perfect sense in a "one image at a time" workflow. However, for many photographers this is not practical when dealing with a large volume of images that have to be delivered to a client. Fortunately, if you've done your homework and tested your camera to optimize your ACR settings (at least visually) to your shooting style, you shouldn't have to do that much in ACR to get a reasonably good image. For most commercial applications ACR can deliver a good enough image to satisfy most clients with minor slider adjustments. Problem images – underexposed, overexposed or unusual/uncontrolled lighting conditions WILL require different strategies in ACR/Photoshop to fully optimize the image and I would not try to do everything in ACR just because it has new Hue/Saturation adjustments or a Vibrance slider.

regards,

Lee Varis

President, LADIG

___________________________________________________________________________

On Feb 3, 2007, at 8:05 AM, John Arnold wrote:

Yes...

regards,

Lee Varis

President, LADIG

___________________________________________________________________________

On 2/4/07 6:51 PM, "George Machen" wrote:

First off, I©ˆm a big fan of this product with respect to beautiful rendering it produces. I asked Brian your question, here©ˆs his reply:

Now we can debate again the effects of RGB to LAB to RGB on 8-bit conversions <g>.

On 2/5/07 9:01 AM, "Lee Varis" wrote:

Caution, not smoke blowing but thanks for the kind words. Not knowing your relationship with Adobe as either an Alpha or Beta tester, let me simply say that I know for a fact that the discussion of Lightroom beta 5 (now RC) can be discussed publicly while ACR 4.0 may not necessarily be discussed even though both share the identical imaging pipeline. I©ˆd prefer to remain conservative with NDA©ˆs, you©ˆre free to say whatever you like of course. That doesn©ˆt change the facts of how the products are designed.

For those of you that want timely info about Lightroom from those who©ˆve been working on it long before that name, check out http://lightroom-news.com/

Not really and I©ˆm actually suggesting here that one use the basic operating principles as the designers have suggested! Adobe purchased and implemented a good deal of the RSP©ˆs code into ACR and LR and made significant changes to the imaging pipeline. I have dozens of emails from a very select alpha list (of which I©ˆm pretty sure you©ˆre not privy to) that discussed the evolution of how and WHY controls where placed where they are, in addition to having compatibility with older versions of ACR. So, your statement above is grossly simplified and not too accurate with respect to ACR 4 or LR.

The processing order has been designed in a specific fashion from day one. You can if you wish, bump saturation or Vibrance then white balance or set clipping but I am suggesting that it©ˆs not the most effective way to work with the products. Same could (and should) be said for image corrections in Photoshop. By all means, alter saturation then fix a color cast if you wish. Or sharpen then add noise reduction if you wish. It would be useful if you would explain why such operations are viable.

Yes as I said, the order is flexible but it was designed for a reason. Its not like going outside the order will make your computer explode. However, I thought the point of this list was best practices in using certain imaging products. If you©ˆd like to discuss how to circumvent these concepts, and why its a good idea, go for it. I©ˆm sure Mr. Knoll and Mr. Hamburg would like to know why as well.

Yes, Exposure/Recovery/Fill Light/Blacks/ (there is no Shadow any more in this pane)/Brightness/Contrast affect the image together and that©ˆs why they are lumped together in a single pane. Highlights/Darks/Shadows/Black are in a different pane, and control tone in a different order. There are tone controls that take place at the beginning of the pipeline and those that take place after hence the grouping order (which you don©ˆt have to follow but should when first working with the product) that being start top down, left to right.

I find it amazing that by simply suggesting you follow a recommended process, one discussed for years by the designers of the product, a controversy is again generated on this list. I guess it depends on who here makes the suggestions, not necessary the strength of the recommendation nor where the recommendation originally came from.

As to the two Histograms, the output histogram is at the end of the processing pipe and uses the sRGB response curve. The curves histogram is prior to the tone curve but also uses an sRGB response curve.