Saturday, January 25, 2014

HDRsoft's New Photomatix's Version 5.0.1

Late 2013, HDRsoft released version 5 of their popular Photomatix software. The update is free to current users who purchased Photomatix 4.0 after October 2010; otherwise it’s $29. The most significant new feature relative to landscape photography is a new Tone Mapping method called Contrast Optimizer. It produces realistic looking images while still enhancing shadows and highlights. I found it a better alternative to Exposure Fusion/Natural for natural rendering; however, I didn’t see it challenging HDR Efex Pro as my primary HDR software. Another new feature of note is Fusion/Real-Estate, intended for rendering interior scenes with an outside view through a window.

Contrast Optimizer Example
To test out Contrast Optimizer, I processed an image of Lundy Lake shot at dusk with three frames bracketed at ±2-stops. This was a very high contrast image and was aided with a grad filter. Nevertheless, the frame to capture the shadows was slightly underexposed. That made for a good test case to see how well Exposure Fusion could restore the shadow detail compared to the other processes. Below I show the results with each process: HDR Efex Pro 2, Contrast Optimizer, Exposure Fusion (natural), and Details Enhancer. The horse race was really between HDR Efex Pro and Contrast Optimizer. Exposure Fusion had the most problem with extracting shadow detail and Details Enhancer had the most unnatural look (especially noticeable in the foreground rocks). To be fair, when the frame that captures the shadows is better exposed (i.e., more exposed than in this example), Exposure Fusion can be a more formidable challenger. In the past, it was my most often used method whenever I used Photomatix.

The winner is HDR Efex Pro, due largely to its superior 32-bit editing capability; for example, the curves and graduated filter tools. As you see in the new Contrast Optimizer control panel, it’s relatively light on editing controls. On the other hand, a simpler interface may appeal to many. I could narrow the difference in Photoshop, but I would loose the 32-bit editing advantage. Though you can’t judge HDR performance based on one image, I can definitely say Photomatix has significantly closed the gap in this new version. If you have Photomatix 4.2 (or earlier), I highly recommend you upgrade to version 5.


Contrast Optimizer Control Panel
A simple and easy to use control panel.
However, the spartan panel also means less
editing capability compared to HDR Efex Pro.

HDR Efex Pro
Overall, a better rendering than Contrast Optimizer, but not dramatically. The differences can be evened up a little in Photoshop. Note that though the shadows are a bit darker than Contrast Optimizer, that was intentional when I added a bit more contrast.
Contrast Optimizer
A bit more touchup in Photoshop, and you have a well-rendered image. However, you loose the 32-bit advantage when working in Photoshop.
Exposure Fusion/Natural
Shadows were blocked up, but given more exposure in the shadow frame, the results would have been much better. Previously, when I was primarily using Photomatix, I used Exposure Fusion almost exclusively and was generally happy with the results. That said, Contrast Optimizer is a definite improvement.
Details Enhancer
I could have diddled more to improve this image, but why bother. There were times I had to resort to Details Enhancer over Exposure Fusion, but even then I wasn't getting the most satisfying results. Simply put, when HDR Efex Pro 2 was introduced, it blew away Photomatix's tone mapping for natural landscape rendering. Now with Contrast Optimizer, Photomatix is back in the fight.

Fusion/Real-Estate Example
I tested this new method on an old church mission interior shot and found it did a good job in naturally rendering the bright outdoor detail. On that basis alone, you can argue that Photomatix did a better job than HDR Efex Pro, except it's not that simple. HDR Efex Pro better captures the effect of a brightly window-lit room. In this image, the lighting effect is more important than the uninteresting outside. The real benefit of this new Fusion method is described by its name: Real-Estate. If you are an agent who wants to highlight a living room that has an ocean view, then you want the outside to be well exposed and colorful. This also holds true for certain landscape images; for example, using an old barn's interior window to frame an outside scenic view. Here you are using the window only as a framing element, so you want the outside scene to be properly exposed. You could argue it's better to shoot such a situation with two exposures only: one for the outside and the other for the inside of the barn, and merge them manually in Photoshop. That may well be true, so my advice: shoot it both ways!

HDR Efex Pro
If I want, I could improve the outside detail using Control Points. However, the objective of this image is to capitalize on the window light's interior illumination effect.
Exposure Fusion/Real-Estate
The outside lawn and shrubs have more detail and saturation than the HDR Efex Pro version. Although I prefer the HDR Efex Pro's image better, there are situations I explain in the above text when Exposure Fusion may be the better tool. Even with this example, it's still a well-rendered image and a little more work in Photoshop will easily make this a wall hanger. 

Monday, September 2, 2013

Apps for the Landscape Photographer


I recently acquired some iPhone apps that may prove useful for landscape photography. I have yet to field test them, so the following are only first impressions after some playing around. I’m planning several photo trips starting next month and I’ll update this blog if my opinion changes with actual field use. The apps are: PhotoPills, The Photographer's Ephemeris (TPE), CameraAngle, and Pocket Light Meter. TPE is available for all platforms, including desktop, while the rest are iPhone only.

PhotoPills and The Photographer’s Ephemeris (TPE)

The essence of these two very similar apps is to indicate the sun and moon coordinates during any time of day at any global location. TPE has been around for some time and most photographers are probably already familiar with it. PhotoPills is a new challenger that adds some impressive features. PhotoPills is $10 while TPE is free for desktops, $5 for Android and $9 for iOS. Note that PhotoPills' iPad version is just a ported iPhone version and doesn’t utilize the larger screen.

The screenshot below of TPE (iPad version) shows the purpose of these apps. I placed a red pin on one of my favorite spots to photograph the Court of the Patriarchs in Zion National Park, specifically the Abraham Peak with the Virgin River in the foreground. The direction of the rise and set of the sun and moon are shown by the radiating lines from the red pin for the selected date of 19-Nov-2013. Orange is sunset, yellow is sunrise, dark blue is the moonset, and light blue is the moonrise. Moving the time scale below rotates thin yellow (for the sun) and blue (for the moon) lines to show their position anytime during the day. All the ephemeris data is displayed in the bottom panels. I also placed a grey pin on Abraham Peak to determine its distance and elevation, which is indicated in the lower right panel. The grey pin aids in determining if and when either the sun or moon is at some desired coordinate relative to a landmark.


PhotoPills does essentially the same thing and additionally throws in extra tools relating to depth-of-field, exposure conversion, star trails, and time-lapse photography to name a few. Beside the extra frills, what PhotoPills really brings to the table is its superior search capability for determining when (and if) the sun or moon is in a particular coordinate. If you ever had a favorite site that you always wanted to photograph with a full moon precisely positioned within the scene, then this is the must-have app.

PhotoPills, being a first version, does have some drawbacks. One is its location database and search engine. While most major landmarks can be found, it’s hit-and-miss on lesser ones. For example, TPE had no problem finding two popular state parks along Lake Tahoe (Sugar Pine and Sand Harbor). PhotoPills couldn't find them. This is most frustrating when, say, you’re at a particular location and trying to “pin” a nearby landmark to assess elevation and distance, but PhotoPills can’t find it. Instead, you have to try to pin the landmark using the satellite imagery, which can be difficult (if not impossible). The iPhone-size screen and other interface annoyances only compound the frustration. Presently, I see PhotoPills more a long-range planning tool while TPE is better for everyday field use.

Example using PhotoPills
I live near a 19th century lighthouse that I would someday like to photograph with a long lens capturing the upper portion of the lighthouse with a full moon next to it. (See my blog post, Moonshot, for more info on shooting the moon.) The free TPE desktop version lacks any automatic search for a predetermined sun or moon position. Instead, you have to manually hunt for it. The tablet versions have a search capability, but only in the azimuth direction that doesn’t include elevation. Though finding a predetermined position is easier with the tablet version, you still are left with some manual hunting to find the desired elevation. However, if you can accurately find and “pin” your desired shooting spot as well as whatever landmark you’re aligning to, then PhotoPills is the perfect tool.

In the PhotoPills screenshot below, the red pin shows my intended shooting location. I placed the black pin (equivalent to the grey pin in TPE) to trace the line-of-sight I want the full moon to be, which is slightly to the right of the lighthouse. Also, for the moon to be in proper alignment with the lighthouse, its elevation needs to be about 10-degrees. How did I know that? I’ll explain that when I get to the CameraAngle app. Usually, you can place the black pin on a landmark (a mountain peak for example) and the app indicates the elevation angle and distance (same as TPE’s grey pin). But building heights are not in the database and must be determined separately.


Now, based on the black pin’s position, I have PhotoPills search when the moon will be at that azimuth and elevation. I can control the search window size to as much as ± 5-degrees in either azimuth or elevation. In this example, I set the search field from 9 to 13 degrees elevation and ± 2 degrees in azimuth. PhotoPills then lists all the moon phases for that coordinate. You can sort the listing and show all the full moons first. The listing is color coded to indicate if the moon is at night, during the day, or during the golden hour. I want it during the golden hour since that's the most photogenic time and the contrast issues with the moon are minimal. From the screenshot below, you can see that the next best opportunity isn’t till May 2015. In the above screenshot, notice that the full moon (thin blue line) on 2-May-2015 aligns both in azimuth and elevation with the dotted black-pin line at 7:37pm, which is exactly at sunset–perfect! (To see the black pin's coordinates you  finger-swipe the top banner showing the sun and moon's rise and set times.) If you can tolerate more slop in the moon’s position, then widening the search area will yield more results that may occur sooner.


Recommendation
TPE is the must-have field tool that should satisfy 99% of your needs. I feel that for determining routine sun and moon ephemeris data, TPE is more user-friendly and the larger screen support (for tablets and desktops) is one main reason why. The free desktop version is sufficient by itself, but if you don’t carry a laptop out in the field, then you’ll need the tablet or smartphone app. To be fair, if you are relying only on a smart phone in the field, then PhotoPills may be a better choice (if you have an iPhone). The reason is TPE was also annoying to use on a small-screen phone. And if establishing a precise sun or moon position in your compositions is an important part of your photography, then PhotoPills is the best tool for that job. Also, the extra tools may have some appeal. Just beware that PhotoPills can still be frustrating to use at times.

[Update. I recently tried a new app called LightTrac by Rivolu Pte Ltd. It is available for both iOS and Android for $4.99. My first impressions are it's essentially a lite version of TPE. It is user friendly and a few bucks cheaper. This might appeal to someone wanting something with no frills and is easy to use. Otherwise, I didn't see any reason why it would replace TPE for me, which is more versatile and nearly as easy to use.] 

CameraAngle by Geometry ($0.99, iPhone only)

In the above lighthouse example, I knew in advance that I wanted the moon to be about 10-degrees in elevation. Often, the only way you can precisely position the moon or sun within a composition is to be there. But, you still need a way to measure elevation and so you need an inclinometer. There are many inclinometer apps, but sighting an angle with them is rather cumbersome. CameraAngle solves this by using your phone’s camera. Simply line the crosshairs where you want the moon (or sun) to be and read the angle. In the screen shot below, you see approximately where I wanted the moon to be relative to the lighthouse. (Note that it shows the angle as negative for reasons I can’t explain.)



Pocket Light Meter by Nuwaste Studios (free or $0.99 to remove adds, iPhone only)

I can imagine many DSLR users asking why would I ever need a light meter? Well, generally you don’t, but it can useful in certain situations. In my free ebook on HDR, I talk about using a light meter to help determine the dynamic range of a scene so you can set the appropriate bracketing range. I have a spot meter that is the perfect tool for that task, but to be honest most times I determine bracketing based on experience and a little trial-and-error. Still, with a light meter conveniently in your hip pocket (or wherever you keep your phone) you may find it useful to quickly size up a scene’s dynamic range. Though Light Meter is not a spot meter, it’s close enough to be useful.

Now I can imagine another question: why not just use the DSLR’s internal light meter (assuming you have a spot mode). The problem is you want to know the range in stops. Cameras normally indicate exposure by shutter speed and f-stop. Trying to figure the stop difference between two exposures takes a little math. Instead, you want to read exposure in EVs (Exposure Value). That way it’s easy to mentally subtract the two numbers in your head to determine the dynamic range in stops. That’s exactly what Light Meter can do for you.

In the screenshots below, you see two readings of a daylight scene in a local park.  The stop difference between the sunlit and shadow areas is: 14.21 – 10.08 = 4.13 stops. You can tap the red rectangle anywhere within the screen or simply repoint the phone to measure the desired area. But there is a drawback: the dynamic range is actually more than 4 stops and probably closer to 5 stops. The problem is the large measuring area. This makes it difficult (if not impossible) to pinpoint either the highlight or darkest shadow without being thrown off by surrounding light. This is why the 1-degree spot meter is more adept at this job. The silver lining, though, is this drawback may work to your advantage.



The reason is the true dynamic range of a scene isn’t necessarily the bracketing range you need. As I explain in my HDR ebook, you need only pull back the highlights and shadows about one-third from their histogram ends. Any light meter, instead, wants to place them roughly in the middle, which is too far. So for the above example, though 5-stops may be the true dynamic range, a 4-stop bracketing range may be more appropriate (i.e., set your camera’s auto bracketing to ± 2 stops). So it’s possible that most times merely subtracting the two readings is close enough for government work. Still results can vary, so you may need to empirically determine a correction factor that works for you. Remember that precision accuracy isn’t necessary. You just need to get close enough so, at most, you need only tweak the bracketing range a little after previewing the histograms.

As for the general accuracy of Light Meter, first off, displaying EV to two decimal points is totally ludicrous and a complete false sense of accuracy. Nevertheless, I compared Light Meter to my Minolta spot meter and most times the readings correlated reasonably well. Actually in one case, while my spot meter was thrown off by the color of a reflective surface, Light Meter nailed it. I’m guessing Light Meter may use the color information to correct for any nonmidtone-like colors. However, Light Meter could get quirky at times. The red rectangle is not a precise metering area and sometimes seemed to exhibit a hysteresis-like variability in its readings relative to its red boarder. But that’s needless quibbling since, as I said before, all you need is a reasonable approximation of the dynamic range. Plus Light Meter costs less than a buck while a good spot meter is over $500!

Wednesday, May 15, 2013

Photoshop's 32-Bit Mode vs. HDR Efex Pro 2


HDR has typically involved tone mapping with its often garish or unnatural results. Then Photomatix offered Exposure Fusion that excelled at natural rendering, though it was not tone mapping or true HDR. Since then, tone mapping has improved at producing natural-looking images as demonstrated by the latest version of HDR Efex Pro 2. Now there's another viable option lurking within either Lightroom or Photoshop CS6.

When you lift the hood on Photoshop CS6, you’ll notice a new feature that changes the HDR equation. I’m not talking about Photoshop’s HDR Pro tone mapping, which frankly is not up to par with Photomatix or HDR Efex. Instead, it’s the ability to simply take multiple exposures and merge them into a 32-bit floating-point file that is editable in either Camera Raw or Lightroom. This means you don’t need separate HDR software and have to struggle with all their arcane controls; and best of all, it’s great at natural rendering. But now the $64 question: is it really better?

To answer that question, I processed several HDR images and compared them side-by-side on screen. Since the editing options are so varied between the two processes, each with their own unique advantages, I took a Machiavellian “ends justify the means” approach. In other words, I used whatever tools available in HDR Efex, Camera Raw, Color Efex Pro, and Photoshop to produce the best end-result for both processes. At the end, it's what looks best that counts and not how amazing the process was to get there.

How to generate a 32-Bit File
I’ll describe how to generate a 32-bit file based on my workflow environment using Adobe Bridge and Photoshop. In Bridge, I select the HDR images and click: Tools → Photoshop → Merge to HDR Pro. After processing in Photoshop, I set the Mode to 32 Bit and click OK. Except for optionally checking the deghost box, that's all you do. When the image opens in Photoshop, it won’t look anything special, but don’t worry. I then save the image as a TIFF file using the following TIFF settings (which should already be the default): 32 bit (Float), LZW (or None) compression, Interleaved, and IBM PC. Then close or exit Photoshop and return to Bridge. From Bridge, select the TIFF file and right-click: Open in Camera Raw. (Important: make sure in the Camera Raw Preference the JPEG and TIFF Handling option is not set to Disable TIFF Support.)

Camera Raw is now your substitute for HDR tone-mapping software. It’s like editing a normal raw file, except one with enormous dynamic range. That means you’ll initially start with radical settings in the Basic Panel; specifically, near -100 for Highlights and +100 for Shadows, and very large values for Clarity. You can then proceed to use the rest of Camera Raw’s features to edit the image. When done, click Open Image or Open Object (depending on the default) and continue in Photoshop. Now you can use Photoshop, or as I did with my test samples, start first with Color Efex Pro and then fine-tune them in Photoshop.

Comparison
For the sample image below, I selected a classic daytime high-contrast image with a bright and cloudy sky, a shaded cliff side, and day-lit foreground trees. Though you can’t judge a process based on one image, I think this image is somewhat representative of the general differences.

When I processed the 32-bit mode image, it took Camera Raw’s HSL/Grayscale Panel to improve the autumn tree colors to compete with HDR Efex Pro’s rendering. Furthermore, it took Photoshop’s Shadows/Highlights tool to improve the cloud rendering and deep-shadow detail to compete with HDR Efex Pro. In the end, it took a little more finessing in Camera Raw to even things up, but at least you’re using familiar tools and not dealing with the arcane tone-mapping controls in HDR software.

When you view the two images below, you’ll note a modest, but by no means dramatic, difference. Unfortunately, the differences may be difficult to observe in these low-res images, but they are more obvious when viewing the original file or a large print. Overall, HDR Efex retains the tone mapping advantage of extracting better detail and contrast within the highlights and shadows and better overall midtone contrast and color saturation. But the 32-bit mode is superior when it comes to noise suppression and—the bane of tone mapping—halo artifacts. I zeroed-out the Structure setting in HDR Efex and still couldn’t match the 32-bit mode’s near-zero noise in the blue sky. And though this HDR Efex example didn’t suffer from halo problems, when compared to other images that had halo problems, the 32-bit mode versions were immune.

There were other fundamental differences I noted with the other test photos. The 32-bit mode process doesn’t compress the tonal range as well as HDR Efex. Below is a portion of a photo of the Zion Watchman illuminated by the setting sun. HDR Efex better captures the detail and color saturation while the 32-bit mode tends to blowout the highlights with higher contrast. This illustrates not to go overboard with the Clarity slider, which tends to blowout the highlights. The deghost option in Photoshop’s HDR Pro was also not as effective as HDR Efex’s; but that said, try not to use either whenever possible. Finally, I noticed in a few samples where the 32-bit mode picked up some banding artifacts in the sky while HDR Efex was clean of that, but was still prone to its usual haloing tendency.

32-Bit Image
HDR Efex Pro Image
32-Bit Image
HDR Efex Pro Image
Conclusion
If you have either Lightroom or Photoshop CS6, I think you could comfortably survive using the 32-bit mode as your sole means to process HDR, especially if you don’t want to invest either the money or learning time in more software. However, given the new lower $149 price on the Nik Collection that includes HDR Efex and Color Efex (and other highly desirable plugins), cost and learning curve may not be a deterrent. In that case, with my philosophy to always extract the last drop of “fine” in fine-art photography, HDR Efex Pro will continue to be my first tool of choice. That said, the 32-bit mode is still an invaluable backup and now rivals Photomatix’s Exposure Fusion as my backup if HDR Efex Pro should ever go squirrely with a particular image. Besides landscape photography, there are many other photo subjects done with HDR that, by their nature, may possibly render better in 32-bit mode (or Photomatix for that matter). For example, though HDR Efex may have the advantage in tone mapping, Camera Raw (or Lightroom) has some 32-bit mode editing advantages. Specifically, the neutral-gradient density filter tool is superior to the one in HDR Efex Pro, and the ability to edit hue/luminance/saturation by color or set white balance is lacking in HDR Efex Pro. In the comparison photos above, note that I was able to emphasize the blue sky more in the 32-bit mode image. 

I have experienced times that when using HDR it can be highly unpredictable. That's why I recommend to always try to shoot a backup using a neutral-density gradient filter if possible. Lacking that, having multiple options to process HDR does relieve me of some anxiety when I do have to totally rely on HDR. So it's worthwhile to consider investing in more than one HDR process.

[Notice: If you have an older version of Photoshop, you no longer can upgrade to CS6 (this does not apply to Lightroom). CS6 has been replaced by Photoshop CC (Creative Cloud) and is available only through subscription. Visit the Adobe website for more info.]

Tuesday, May 7, 2013

Nik's Dfine 2 Noise Reduction Plugin


Up to now I’ve posted reviews on Nik’s HDR Efex Pro 2, Color Efex Pro 4, and Sharpener Pro 3; and all were highly favorable to the point they are now a permanent part of my workflow. Now I’ve looked at Dfine 2 to see if it keeps up the tradition. But first: what is Dfine? This is another Photoshop, Lightroom, et. al., plugin that’s dedicated to noise reduction. It seems to me that Dfine’s attempt to distinguish itself from other noise filters is by being a one-click automatic operation. It will analyze the image and apply what it determines is the best noise reduction. Nik’s training video touts that this should handle almost all situations. You do have the option to switch to manual control, but I decided to take them at their word and evaluate it strictly on automatic mode.

Below is a screen shot of Dfine and it is similar to the rest of their plugin interfaces. A few notable features are, as in all of Nik's plugins, the Control Points (aka U Point Technology) that allow selective noise reduction. However, as useful as I find Control Points in Nik's other plugins, I'm not sure I see this as particularly useful when dealing with noise. When I want to apply selective noise reduction, I really need the more sophisticated masking tools in Photoshop. Another feature is the ability to control noise reduction by a specific color; which is nice, except for me not a "die-for" feature.

Once you launch the filter (in my case, through Photoshop’s Filter menu), you wait for it to analyze the image and then click done. Notice the two square outlines on the image in the screenshot below. The software looks for a continuous tone area with little or no detail to analyze, and once defined, automatically applies what it determines is the best global noise reduction. A one-click noise solution would be a truly great feature—if it really works. I’m sure you have found it frustrating, as I have, to fiddle with multiple sliders in other noise filters when trying to find an optimum setting. So to find out how well Dfine works, I took an image I consider my worst-case noise problem in “normal” practice and compared it with Camera Raw’s Noise Reduction filter and Photoshop’s Reduce Noise filter.


Nik's Dfine 2 Screen

It’s important I explain my “normal” practice so that you understand any bias in my opinions. I almost always shoot at ISO 100 and use a full-frame DSLR to keep noise down to the point where I usually didn’t need noise reduction. Lately, however, things have changed due to my increased usage of HDR software and Nik’s Color Efex Pro 4 plugin. Either of these programs can dredge up noise from the netherworld that you otherwise would have never known existed. This has now forced me to make sure my images are squeaky clean before I use either program. This also means that the amount of noise reduction is modest and, as such, I didn’t evaluate the noise filters on their ability to fix super-noisy images. My worst-case image is one shot at ISO 800, which is usually the highest I’ll go and only when I’m contending with wind and/or depth-of-field issues.

Below is the example image I used: sunset at California's Torrey Pines State Park that was shot at ISO 800. The image is only a little noisy with the sky as the biggest offender, though there was some minor color noise in detail within the tree grove. Noise within continuous tones, like the sky, is really my biggest concern and the most noticeable flaw on my prints. For the comparison images, I selected a portion of the nosiest part of the sky and a portion of the tree grove. Each image is a screenshot at 200% magnified view. You may need to click on the image to get a better view.


Test image: Torrey Pines State Park, California. F/16,1/60 sec, ISO 800.

In all my examples, after the noise filter was applied I then used Color Efex Pro 4 to spice up the image; which in turn should magnifiy any latent noise not cleaned up by the filters. Within each group, the first image is without noise reduction, second is using Camera Raw’s Noise Reduction filter, third was Photoshop’s Reduce Noise filter, and finally Dfine. The Camera Raw noise reduction settings were: Luminance 15, Luminance Detail 60, Luminance Contrast 50, Color 25, and Color Detail 60. The Photoshop (CS6) settings were: Strength 6, Preserve Detail 30%, Color Noise 30%, and Sharpen Detail 20%.

Sky: no noise reduction. Noise is clearly visible after Color Efex Pro.
Sky: Camera Raw Noise Reduction Filter. Adequate noise reduction.
Sky: Photoshop Reduce Noise Filter. A little different structure than
Camera Raw's, but both are sufficiently cleaned up not to be noticed
in a large print.
Sky: Dfine. Clearly the best, but at a cost (see next images).
Tree Grove: No noise reduction. Some minor color noise in darker tones.
Tree Grove: Camera Raw Noise Reduction filter. Very good noise
reduction with slight loss of detail.
Tree Grove: Photoshop Reduce Noise Filter. Excellent color noise
reduction with imperceptible loss of detail due to the Preserve
Detail and Sharpen Detail slider options in the Reduce Noise filter.
Tree Grove: Dfine. No color noise at all, but noticeable loss in fine detail
compared to Camera Raw and Photoshop.
Conclusion
I extracted a slice from each test image and printed them at an equivalent 11x14 inch size after sharpening with Sharpener Pro. In the image without noise reduction, the noise was apparent but not overly objectionable. All the noise-reduced images effectively eliminated the noise. The Photoshop and Camera Raw noise filters were indistinguishable, as you would expect. As for the Dfine image, it took careful study to notice the slight softening in some detail within the tree grove. In other words, despite the obvious differences noted on screen, the printer has a way of equalizing minute screen differences due to it's blurring factor. When viewed casually at a normal distance, I doubt you'll be able to tell the difference.

In the end, recommending Dfine boils down to if the convenience of one-click noise reduction is worth the minor detail softening that is likely unnoticeable to most observers, unless explicitly pointed out. Speaking for myself, I tend to want to squeeze out that last drop of "fine" in fine-art, and the Camera Raw or Photoshop noise filters are clearly the better tools. Yes, I can manually crank down the noise reduction in Dfine, but I still wasn't happy with the on-screen results; plus having to manually tweak the settings defeats the one-click convenience.

My workflow will continue as it has, and that's to always clean up any noise first in Camera Raw before proceeding to either Photoshop or my HDR software. It's possible, depending on how aggressive my image adjustments were in the HDR software or Photoshop, that a bit more noise reduction is needed. In that case I may apply the noise reduction selectively rather than globally. But no matter the method, I would certainly use Photoshop's Noise Reduction rather than Dfine. If I was to apply noise reduction to just a continuous tone area, such as the sky, then I might consider Dfine, but that would be the exception. Otherwise, I don't see a home for Dfine in my normal workflow.