Friday, October 2, 2009

More Canon G10 Infrared (IR) Images

On a couple of puffy-cloud days this week I worked with the IR camera again. This time I used DXO Optics Pro to convert the raw files. Since it works based on profiles for specific camera/lens/focal length/ISO, many issues that might otherwise be bothersome are dispatched with ease.

This Coupeville house just looked interesting with the cloud behind it. I used the auto white balance setting provided by the camera, used DXO's ability to do film look emulations, then applied medium-high contrast. The B&W film look I chose for this was Ilford XP2.

For this image of the barn at Jenne Farm, I simply applied a curves adjustment, then in the HSL section (hue, saturation, lightness) I set saturation to 0 to convert to black & white. Remember that the IR images have some color content as they come from the camera.

For this shot of the same barn I tweaked quite a bit: highlight recovery strong, black point 99, local contrast 43, then in color mode I chose the sepia gold toning with medium-high contrast and grain 0.

Finally, for this shot of the Jenne farm house and barn, I used color mode b&w, high contrast and daylight white balance. Note that without the b&w conversion, daylight white balance produces a major red cast as one would expect of a color IR image.

Labels: , , , , , , , , ,

Saturday, March 28, 2009

High Dynamic Range Photography ... HDR

Background
High Dynamic Range (HDR) photography combines multiple shots at bracketed exposure (constant aperture, variable shutter speed) to render extended dynamic range and, often, an exaggerated look. I'm not an early adopter at high dynamic range (HDR) photography. Let's say I've been exercising some restraint; that sounds better than I'm lazy. Actually, I hate to buy software that I use infrequently and forget that I have, so I've been waiting to be sure I'll really do this. Last night, with the tax info off to my preparer, my reward was to dive in and try HDR.
I have been shooting bracketed sets of shots in anticipation of HDR imaging ... for about two years. That means I have some material to work with: Blue Hen Falls in Cuyahoga Valley National Park, the Jacob Ebey House in Ebey's Landing National Historical Reserve, the windmill in Windjammer Park in Oak Harbor Washington, sunrise across Saratoga Passage from Dugualla Bay on Whidbey Island.

I suppose I am an early adopter in a way; a good fifteen years ago, I shot a series of three exposures of a scene in Zion National Park on 4x5 film. I had the lightest and darkest drum scanned and later combined them in Photoshop. My Photoshop skills were limited, Photoshop tools were crude, and it took a lot of work to mask areas. Still, the result was pretty good. I may try to find the scans and do it again.
My approach to selecting an HDR program will be to try a quick conversion with each program, then select the top candidate and explore at greater depth to confirm that it suits my needs. So please keep in mind that the results and comments you see here are at the quick and dirty level, therefore subject to revision.

Photoshop CS4
This is to establish a baseline, not a suggestion that you'd want to do it this way. Can you tell how I was unimpressed with Photoshop's ability to creat HDR images?
Starting in Bridge, I selected the lightest, darkest, and good middle image. The full set is five images, but I don't think this scene requires all of them. Then it's TOOLS>PHOTOSHOP>MERGE TO HDR. When prompted, save the file as 32-bit. Then in Photoshop IMAGE>MODE>16 BIT (or 8 bit if you prefer). At this point, local adaptation seems to be the choice. Play with the sliders and the tone curve until you get something close. After processing, use Photoshop curves.
PHOTOSHOP TEST EXAMPLE
In Photoshop I applied the Strong Contrast preset in the curves dialog to achieve the look above. To me this still suffers from the flat look of a native HDR image.

Dynamic Photo HDR 4 ($55US)
The web site does a good job of selling Dynamic Photo HDR. It seems to be easy to use with gobs of flexibility. I was particularly taken with the apparent ability to create a range of looks from subtle to extreme HDR. Following are my initial impressions ... with no attempt to RTFM (read the manual).
It seems that aligning images is the expected first step in creating an HDR image. The assumption is that even with a tripod there can be some misalignment of images. I went with the flow. Working with a five image set (shot on a tripod), DPHDR needed manual help aligning the darkest image pair ... even with my windmill test which included blades of the windmill against the sky (boucoup contrast). It allowed me to manually adjust, but I found the interface difficult and not fabulously intuitive.
Step two is to walk away and make a cup of coffee while the HDR image is built. The resulting file contains the full dynamic range of information from all included images. When this is done, the fun begins.
DPHDR is a real playground when it comes to tweaking the image for output. HDR images contain much more information than monitors can represent, than printers can print, than eyes can see. Adjusting and compressing the information allows us to preserve detail ranging from shadows to hightlights that would be impossible to record within the limited dynamic range of digital capture (or film, for that matter). The result can be natural looking or exaggerated, usually with extreme local contrast. Each approach can be pleasing in its way.
DYNAMIC PHOTO HDR TEST EXAMPLE
My tests came out showing more digital noise than I like (I worked from NEF raw files) but I was able to achieve just the HDR look I wanted, on the verge of extreme but not over the top. The realtime preview of adjustments was responsive, allowing me to have a good time trying variations. It does take a little while to render the final tonemapped image.

Photomatix Pro 3 ($99US)
It always seems that when I spot HDR images in a gallery or when I'm judging a show, the author uses Photomatix. I'm willing to call it the gold standard based on the results I've seen. It strikes me as pretty spare, with a simple workflow and reasonable range of adjustments.
PP3 had no trouble aligning my test images. It offers just two types of tonemapping, with nine sliders to adjust parameters to get the desired look. It is necessary to make adjustments, release and wait for the preview to catch up: not real time preview. Although noticeably not as responsive as DPHDR, the delay was not too irritating.

PHOTOMATIX TEST EXAMPLE
I was not able to achieve as much "HDR look" with Photomatix, but found a pleasing lack of edge artifacts and grit in the result. I should note that I again worked directly from raw files. The publisher recommends converting to TIFF first, admitting that other raw converters are better than what's built in. I found that it produced a cleaner result than DPHDR.
At this point, I am inclined to go with Photomatix even though I'd like to be able to have the option to sometimes get a more extreme look. Clearly, this program will produce images that will sell, that fulfill my artistic ambitions, and that convey the mood of time and place.

FDR Tools Advanced (
39 €)
FDR Tools has a slightly quirky interface, opening separate windows for various dialogs. When I launched it I saw the main window, plus windows for progress, Tools and Navigator. Navigator is actually the preview of your finished image, though the individual shots show up as thumbnails with histograms in the main window. This could be interesting.
FDRT was not initially intuitive, so I followed the steps outlined in the online manual, importing images, then clicking "edit," which creates the HDR raw image. For this trial run I chose the compressor method of tone mapping, with compression 10, contrast 6.4, smoothing 7, and bringing curves over to the darkest and lightest pixels, darkening the lightest ones a bit.
Opening the resulting 16-bit TIFF in Camera Raw 5.3, it was much flatter looking than the preview in FDRT. The red channel was clipping a bit so I lowered exposure, then cranked up clarity to +30. Using the default curves points in ACR, I left highlights at 0, lights at +60, darks at -20, and shadows at -80.
FDR TOOLS TEST EXAMPLE
For some reason the trial version of FDRT did not leave a visible watermark, as promised. As you can see, I was not able to get as punchy an image as with Photomatix. The good news is that the result seems pretty clean and halo-free. I would say that my primary gripe about FDRT is the inscruitable interface.

Essential HDR Community Edition (free)
It seems that the free "community edition" of Essential HDR is identical to Essential HDR Standard Edition ($48.99) except that it shrinks the image to less than one MB. I say it seems so because the full program cannot be downloaded until after purchase, so I have to go with this assumption for now.
EHDR is fast in all operations, but since it shrank my image, I cannot compare directly with other programs. The interface is intuitive and simple. Dabbling, I was able to create a range of looks, from moderate to extreme. Halos are a little more obvious in the final image than in the preview.
Once it has combined your source images, select the tone mapping method you prefer: Detail Revealer (my choice) or Fast Tone Balancer. Tweak the Details, Brightness, Color Saturation, and Fill Light sliders for the desired look. I set black point and white point is also available. Color balance can also be adjusted.
ESSENTIAL HDR TEST EXAMPLE
No Photoshop touch-up here; this is the result straight from the program. As I look more at the results I'm getting, I realize that the challenge in this image is to prserve the color of the tilled soil and dark forest.

Ariea HDR Max ($129US)
Well, HDR Max won't run in 30-day trial mode. I downloaded and installed it, but get an "unknown error" message when I click the trial radio button and next. After having the install program repair it, same crash.

Picturenaut HDR 2.12 (free)
Since Picturenaut will not accept raw files, I converted the three images to TIFF format in ACR first. The program quickly built the 32-bit HDR file. After trying both Adaptive Logarithmic and Photoreceptor Physiology tone mapping methods, I went with adaptive. I was struggling to get an interesting result when Picturenaut crashed. Play as I might with the sliders, I was not able to achieve anything remarkable. I will say that Picturenaut produced a clean image, free of grit and halo. But it really was no improvement over what I can get with careful processing of a single raw file.
PICTURENAUT HDR TEST EXAMPLE
This looks a bit like the Photoshop CS4 HDR image.

Photo Acute Studio ($119)
Photo Acute Studio seems to have two primary functions: building HDR files and extracting additional detail from a series of more than four shots. I did not explore the second function.
The interface is straightforward. On my first attempt, I brought in the three raw files I've been using. After setting everything up, PAS told me it wanted a DNG (Adobe) raw file. I went to the site a verified that the Nikon D300 and 50mm f1.8 lens used for these shots are supported in NEF format.
Plan B was to use my TIFF conversions of these images. The process was easy. First, I brought the images in, selected all three, and highlighted the one with no blocked highlights, as directed. I had to tell the program what camera and lens I used. I then chose the program's High Dynamic Trange Tripod Mounted preset and the processing took about two minutes. The result had a gross white halo around dark objects, but I saved it anyway.
PHOTO ACUTE HDR TEST EXAMPLE
I cannot show you the results of this test because Photoshop can't open the resulting 16-bit TIFF. Went back and processed again, saving as a JPEG this time. The results were no better, but at least you can see what it did. Since the result was rather dark, when I was resizing in Photoshop I applied the curves "lighter" preset.

What's Next?
I've covered the HDR programs of which I am currently aware, so the next step is to narrow the choices in a new blog entry. I plan to look closer at:
  • Dynamic Photo HDR because the results were interesting if exaggerated. If it can be moderated, it's a great value at $55.
  • Photomatix because the results were interesting if too tame.
Did I mention that these were the favored candidates at the start of my research?

Labels: , , , ,