"Pay attention to the world." -- Susan Sontag
 

Category: atlanta

Iris japonica (1 of 2) / Notes on Image Reconstruction

 , , , , , , 0

From “Iris Japonica” in The World of Irises by The American Iris Society, edited by Bee Warburton and Melba Hamblen:

“The much frilled and fringed flowers of Iris japonica lie almost flat and are 8-10 cm (3-4 inches) in diameter. The multibranched stalks, up to 60 cm (2 feet) in height, carry as many as 24-30 blossoms, two to each spathe. Generally four to six are open on each stalk, but when plants are well grown there may be many more. The pale lavender flowers have finely traced orange markings on their falls consisting of three slight ridges. The margins appear on a white ground outlined with splashy dots of deeper lavender shade. Smaller orange dots are contained within this area. The lavender style arms are finely fimbriated above the lip….

“Iris japonica is a native of moist woods and grows best outdoors in cool, frost-free conditions…. Since it blooms in early spring, early to mid-March in climates where temperatures seldom go below 3 degrees C (26 degrees F) and freezes are rare, it enhances the azalea and camellia garden….

“Because of their shallow root growth, some kind of mulch is desirable at all times except under circumstances of excessive heat and moisture. New plants emerge some distance from the main rhizome… traveling on the surface of the soil. The damp mulch provides anchorage for these proliferations and encourages their rooting….

“These irises have no genuine dormant period since the foliage remains evergreen the year round.”

From “Japonicas and Hybrids” in Iris for Every Garden by Sydney B. Mitchell:

“Iris japonica, sometimes also called I. fimbriata, is the most distinctive and beautiful of the crested irises. From fans of thin, bright evergreen foliage it sends up two-foot widely branched stems which bear over several weeks a succession of somewhat fugitive but lovely pale lavender flowers, fringed, and with orange crests. A single stem will bear dozens of blooms, so light and airy and so like orchids that this particular species is often referred to as the ‘orchid’ iris….

“Japonica is not in the least demanding as to soil, though its preferences are for one that is not too heavy and that contains humus and leaf mold. In contrast with almost all other irises, it succeeds far better in considerable shade than in sunlight and often will be found growing luxuriantly in the dappled shade of overhanging trees. Unfortunately japonica cannot be grown outdoors in very cold climates, but all along the Pacific Coast and through the southern states it is well worth trying.”

Hello!

This is the first of two posts with photographs of Iris japonica from Oakland Cemetery, that I took on one of my early spring photoshoots. Iris japonica is known by several common names, including Japanese Iris, Fringed Iris, or Butterfly Flower. You might also find it called Shaga or Shaga Flower, “Shaga” being a translation of its Japanese name, one that was used in early horticultural trade before European botanists assigned it scientific names. Both the Butterfly Flower and Fringed Iris names connect to the flower’s visual appearance: Butterfly Flower has Chinese origins reflecting how its spread flower petals resemble butterfly wings; Fringed Iris descends from one of its earlier scientific names, Iris fimbriata, where the Latin word “fimbriata” referred to the distinctive fringed edges of the flower petals. In iris classifications, Iris japonica is included in a group of visually similar crested irises, more formally known as Lophiris.

While they’re members of the same family as upright irises like Iris germanica or other tall irises — the Iridaceae family — their growth pattern is much different. Iris japonica grows close to the ground, as an understory plant that I’ve often found surrounding the trunks of Oakland’s massive magnolia and oak trees. Their early spring bloom time under those trees means that the plants emerge above layers of discarded leaves from the prior autumn as well as built-up winter debris. The evergreen leaves are abundant and large compared to the sizes of individual flowers; and the flower stems push through the layers of natural mulch to bloom just a few inches above ground.

In my last two posts about Camellia japonica (see Camellia japonica (1 of 2) and Camellia japonica (2 of 2)), I mentioned that I photographed the Camellia plant because I wanted to take on the challenge of properly rendering the red and magenta color blending in the flower petals. I took these Iris japonica photos to pursue a different challenge: to see what I could do with their untidy backgrounds, which we tend to ignore when viewing the flowers in real life but dominate any photo featuring both the flowers and their surroundings.

Here’s one of the photos in this series as it came out of the camera, where the flowers look just fine but the rest is a mess — a visually disturbing mix of broken, cut, or winter-damaged leaves that overwhelm the image:

In the olden days of a few years ago, I would — and occasionally did — produce photos of these flowers by zooming in on only the flowers, hiding the backgrounds as much as possible, or converting them to black. Despite how tedious that was, converting them to black — which involved carefully detailed masking around the edges of flower petals — would create a presentable image of the flowers, but only the flowers and maybe a few stems and leaves. I used that black background technique often with different kinds of plants (see here), and my black background period taught me a lot about how to use Lightroom’s masking tools to emphasize the subjects in my photos. Its biggest drawback, however, was the loss of environmental and botanical context since the images isolated only the blooms and eliminated everything else.

When Adobe added Generative AI Remove to Lightroom in 2024, the capability was introduced as a “distraction removal” tool to get rid of spots or unwanted objects in an image. It indeed does that (and does it well), but for my nature photography, I’ve adopted it as a reconstruction tool that’s perfect for conditions like those present in these Iris japonica photos. I can use it to rebuild or reconstruct severely damaged leaves; or, as I’ve come to think of it: replace damaged elements with what might have been there, if the damage had not occurred.

Lightroom previously had (and still has) its traditional healing and cloning tools, which operate by replacing something you select with pixels from another section of the photo. That means, in effect, you could select something like the tip of a broken leaf and replace it with the tip of another leaf that wasn’t damaged. But that would only work if you could find a suitable match. Scroll back up to my sample photo and see if you can find structures, colors, or textures — which matter a lot with botanical photographs — that would match those with any single piece of damage you might want to eliminate. It would be like trying to assemble a puzzle from pieces out of several boxes: most of them just wouldn’t fit.

Like a lot of photo editing tools, the original healing and cloning tools did exactly what you told them to do. Their behavior was literal and restricted to the image you were working on, with no broader context. Generative Remove, on the other hand, isn’t restricted like that; it can draw on an enormous amount of information beyond what’s present in a single image. While I don’t have technical knowledge about exactly how Generative Remove does what it does, I can see — from experimenting with it — that it recognizes patterns in the colors, forms, and textures of botanical (and other) subjects. It may or may not know anything about Iris japonica specifically, but manages to take selections I make and suggest replacements that are aligned with the natural appearance of parts of the plants.

Equally important, I’ve learned that I can influence Generative Remove’s replacement suggestions. If I want to replace the flatly torn end of a leaf in this image with a properly pointy Iris japonica leaf, I can make a roughly triangular selection and the replacement will look more like the tip of an unbroken leaf. And if I make an angled or curved selection aligned with the directional flow of the leaves, the replacement will mimic those angles or curves. These then lead to a more complex form of influence: the sections of the image I reconstruct first help establish how Generative Remove “sees” the image as I continue the reconstruction work, which enhances the quality and effectiveness of its additional replacement suggestions. That led me to develop a specific reconstruction workflow: remove smaller defects first so that some of the leaves are restored early on, then proceed to those with less damage, then finish up with those that are the biggest messes of all.

In this image, for example, I’m about halfway through leaf reconstruction. I’ve finished fixing those on the right side of the flowers, which had less damage than those on the left; and those in the upper right corner now match the natural appearance of Iris japonica leaves when they’re fully intact. These patterns then become additional context that Generative Remove can use as I move to the more seriously damaged leaves on the left side of the photo…

… which, when completed, look like this, and show why I think of this reconstruction as “what might have been there, if the damage had not occurred.”

Here you can see the transition from the original image, to partial reconstruction, to the final reconstructed version — that final version including color adjustments to reflect the flowers’ subtle blend of blue, violet, and purple tones. Select the first photo to view the series in a slideshow, where it should be evident how well the tool (with two hours of my help!) has managed to generate a botanically accurate and naturally plausible background for the flowers.

That the final image doesn’t reproduce exactly what was present at Oakland when I took the photo isn’t, of course, the point. All images — whether drawn or painted, produced by a film camera and retouched in a darkroom, or captured and enhanced by a digital camera’s creative modes — are interpretations rather than precise representations. A tool like Generative Remove provides a way to approach image editing differently, because the tool infers what should occupy any area I select from what’s actually present in the surrounding scene. That makes it possible to render a composition as our human visual system and memory experienced it, where distracting elements like damaged leaves in the background are filtered out in favor of the subject that attracted our attention to begin with. And — unlike the hyper-realism that often surfaces in an image created by an image generator, or even one where the background is replaced — this kind of editing preserves fidelity to the experience of taking a photograph and recalling what we found significant when we did that.

That the Generative Remove tool’s behavior can be influenced is one of the most interesting things about using it for reconstruction like this. But it’s also very different from tools that do exactly what they’re told, as the influence isn’t obvious at first and the results it produces seem ambiguous or random until you repeatedly observe how your influence can work. I don’t tell the Remove tool what I want to do in words; instead, I have to make selections, see what results it produces, decide whether they fit visually and look natural, then use them or modify them — with those choices (and the accumulation of many choices) influencing what the tool does each time I make another selection.

The closest description I can come up with is that it’s like a picture-based dialogue between a human and a machine toward a creative goal. This ambiguity and uncertainty can feel uncomfortable at times, yet embracing it means that the nuanced responses produced by AI-based tools — photo editing tools or even language models like ChatGPT or Claude — are not necessarily mysterious, but are within realms we can understand if we step back from the kind of literal instructions and expectations we’re accustomed to, when interacting with these new technologies.

Thanks for reading and taking a look!

Camellia japonica (2 of 2)

 , , , , , 2

From “Fascinating Immigrants from the Far East” in Japanese Gardens by Wendy B. Murphy:

“The beauty of Japanese plants — from the satiny shimmer of a camellia to the gnarled elegance of a Japanese black pine — has fascinated Western gardeners for centuries. As early as 1694, Engelbert Kaempfer, a German doctor and naturalist with the Dutch East India Company, returned from Japan with a collection of azaleas, camellias and tree peonies that stunned botanists in Europe….

“Kaempfer gathered his plants surreptitiously and at great personal risk, for shoguns sharply watched the movements of early European traders and explorers. One plant collector was thrown out of Japan for obtaining maps of the islands. Other collectors were in mortal danger of encountering the daimio, feudal princes who believed it their patriotic duty to kill foreigners….

“Despite such obstacles, a slow but steady stream of plants trickled from Japan into Europe and eventually across the Atlantic. That trickle became a flood after Commodore Matthew Perry appeared in Tokyo Bay with American warships in 1853, forcing Japan to open trade with the West. Japanese bamboos, azaleas, hydrangeas, hostas, evergreens, lilies, peonies and flowering fruit trees all found adopted homes overseas….

“Today many Western gardeners regard these plants as native, using them liberally in landscapes without realizing their Japanese origins. The camellia, for example, has become so entwined with the folklore of the American South that its Japanese heritage (camellias first arrived in America about 1820) is all but forgotten.”

From “A Camellia Falls” by Yosa Buson in A Haiku Garden: The Four Seasons in Poems and Prints by Stephen Addiss with Fumiko and Akira Yamamoto:

          A camellia falls
spilling out
          yesterday’s rain

Hello!

This is the second of two posts with photographs of Camellia japonica, one of several Camellia species plants at Oakland Cemetery. The first post is Camellia japonica (1 of 2), where I wrote about this specimen’s especial color characteristics.

As this was the first time I’ve created a series of Camellia photographs, I had not previously learned about the plant, its botanical or cultural history, or its use in Victorian gardens like Oakland’s. I still haven’t learned enough about it, but it’s gotten my attention now and my interest in it will continue as I amble through the gardens on subsequent visits. What I find — especially what I’ll eventually discover about the two Camellia trees I mentioned in the previous post — will emerge here in the coming months. Their blooming periods — whether spring, early summer, fall, or winter — will help me identify the species, point me toward a research direction, and determine when I photograph and write about them.

The excerpt from Japanese Gardens by Wendy B. Murphy that I included at the top of this post establishes a pretty nice historical framework for Camellias to start from, taking us from the eras before Western exploration of Asia, to European exploration and imperialism beginning in the 1600s, then to the 1800s, then to the nineteenth and twentieth centuries. Any of the four paragraphs could be used as a starting point for research; what direction I go will be determined by which Camellias I actually encounter and photograph — and it should be a lot of fun to uncover the ways in which Camellias “became entwined with the folklore of the American South.”

While digging into Camellia history with the Japanese Gardens excerpt in mind, I came across this…

Timeline of American Garden History

… produced by Smithsonian Gardens, a program of the Smithsonian Institution. The timeline presents the evolution of horticulture, landscaping, and national parks in the United States since the seventeenth century, something of interest to anyone learning or writing about botanical history.

If you scroll to the 1860-1890 decades, you’ll encounter the part of the timeline around which much of my photography and writing here revolves: “Victorian Gardens in the U.S.” The design of these new gardens embraced landscape and architectural themes that were heavily influenced by European Victorian culture. Scroll backward from that, and you’ll discover key historical events that led directly to the development and design of Oakland Cemetery and its continued evolution today:

  • 1820, with the establishment of a Botanic Garden in Washington, D.C.;
  • 1830, with the creation of Mount Auburn Cemetery in Massachusetts;
  • 1830-1840, which fostered the redevelopment and redesign of natural landscapes;
  • 1841, when A Treatise on the Theory and Practice of Landscape Gardening by A. J. Downing helped establish a new cultural and horticultural vocabulary for landscape design, influential on both private gardens and public spaces like Oakland;
  • 1860-1890, when Victorian gardens flourished throughout the United States, and our frequently visited Oakland expanded from its original six acres to its present size of 48 acres (in 1872).

When I photograph and write about a single plant species like Camellia japonica here, I typically start around 1860 as an anchor point, then branch out to any time or any place from there. This kind of anchoring and interpretation of geographic spaces is often referred to as genius loci, a concept that — as described in A World of Gardens by John Dixon Hunt — evolved from early Roman assessment of the spiritual significance of a place, to a more secular connotation where we consider the characteristics of a space and create meaning from connections among the elements we observe.

Within this kind of framework, we don’t just see a Camellia japonica in isolation as a lovely plant (even though it is), but instead can spin out a variety of narratives about it that may cross into disciplines like botany, architectural and landscape design, history, cultural history, or — in my work — start from examining its photographic and visual characteristics (as I did in the previous post) then bring other threads in from there. This is, of course, why I don’t post photos on Instagram (not that there’s anything wrong with that) where I can’t unroll a thousand words to go with them, but do it here where I can tell whatever story captures my fancy while I’m working on the photos and bouncing around the Internet.

The Japanese Gardens book also contains the following description of the botanical characteristics of Camellia japonica and related species:

“For centuries the Japanese have grown camellias in their gardens, admiring them for their year-round glossy dark green leaves and profusion of large waxy flowers. Sometimes Japanese gardeners use them as hedges, sometimes as ornamental trees, pruning them to accentuate their shapes. Far less delicate than generally believed, these shade-loving evergreens withstand salt air and polluted city conditions. Their 2- to 5-inch flowers come in pink, red or white, or in mixtures of these shades; the blooms may be single or multi-petaled. The thick, glossy leaves are 2 to 4 inches long. 

“In Japan, Korea and China, common camellia may reach a height of 45 feet or more, but in North America it usually grows 6 to 10 feet tall in 15 years and rarely exceeds 15 feet after 20 to 25 years. Some varieties, called early-blooming, flower in midfall; late-blooming types flower in midspring. Sasanqua camellia, which may become 15 to 20 feet tall in its native Japan, becomes only 6 to 8 feet tall in most American gardens, taking 10 years to reach that size. It blooms from early to late fall.”

This description is consistent with how my photographed Camellias look — notably, the glossy leaves and large waxy flowers — but also tells me something about the yet-to-be photographed Camellia trees I mentioned earlier. From the detail in that second paragraph, I could likely estimate the trees’ minimum age; and if they’re taller than fifteen feet, they’re not only older than fifteen to twenty years, but may be capable of exceeding fifteen feet tall simply because southeastern environmental conditions could induce them to reach their native proportions. Now I just have to figure out how to stand on my own shoulders three or four times to determine their height and photograph their flowers….

Thanks for reading and taking a look!

Camellia japonica (1 of 2)

 , , , , , 6

From “Camellia” in Garden Flora: The Natural and Cultural History of the Plants In Your Garden by Noel Kingsbury:

“Camellia contains 125 species: about 104 in China, with more than 80% concentrated south of the Yangtze; some in Korea and Japan; and then more in Southeast Asia. The genus is thought to have originated in southern China in the Cretaceous. In the name, Linnaeus commemorates 17th-century Czech botanist Georg Kamel….

“The current range of species in cultivation encompasses three main species, each with a long history of east Asian cultivation: Camellia japonica (2,000 cultivars), C. reticulata (400 cultivars), and C. sasanqua (300 cultivars).

“Camellias are typical evergreen understorey trees, being an important part of the forest community, usually in regions with moderate to high year-round rainfall. They are relatively slow-growing, appear late in the successional process, and can live for centuries….

“Literary and pictorial evidence suggests that camellias have been cultivated in China as ornamentals for at least 1,800 years. The Song dynasty saw a lot of creative breeding, probably mostly with Camellia japonica, and the use of grafting to produce plants that combined multiple varieties; the Song capital of Hangzhou became a centre for growing and trading the flowers. Camellia sasanquawas also cultivated during this period. The next major stable dynasty, the Ming, saw the first books published on camellias.

“In Japan, Camellia japonicawas taken up by the samurai in the 12th century and was much further developed in the Edo period. The Higo clan were very interested in them, and it was they who introduced C. sasanquainto cultivation in the 17th century…. “

Hello!

This is the first of two posts with photographs of a lovely small shrub that I took a few weeks ago at Oakland Cemetery. The plant is Camellia japonica, one of the many Camellia species and hybrids that are as frequently gardened in the southeast as azaleas and magnolias — especially popular since there are varieties that produce flowers from late fall through early spring when other flowering plants are dormant. In A Garden of One’s Own, Elizabeth Lawrence describes how Camellia japonica and its relative Camellia sasanqua (mentioned in the excerpt above) make excellent garden companions, as C. sasanqua blooms around Christmas in the southeast and C. japonica blooms not too long after, creating a succession of Camellia blooms across several months. Camellia japonica may flower here as early as January and as late as April, and it’s often called Japanese Camellia (though its origins are both Japanese and Chinese), or Common Camellia.

From a photographer’s perspective, though, this plant and its flowers are anything but common. The flowers in full bloom show a striking blend of magenta and red, with red increasing in intensity as you get close to the center of the flower. I was fortunate to have photographed this plant on a relatively bright but overcast day, where the gradations between magenta and red are much more evident. Bright yellow sunlight tends to “blow out” red colors and diminish the appearance of magenta (which is actually its own distinctive blend of blue and red). On a sunny day, you would likely perceive the flower as red, but in filtered light the contrast between the two colors becomes much more apparent.

If we zoom in as far as possible in Lightroom, we get a look at the individual pixels (each square in this screenshot is one colored pixel captured by the camera) from a quarter-inch section of the flower petals. This shows both the intensity of the colors and the extent to which they’re mixed together to produce the blend of red and magenta.

While they appear to be single, solid colors, these pixels are actually the camera’s interpretation of how much red, green, and blue (RGB colors and their relationships) are present in each one. Pixels toward the top of the screenshot (closer to the flower petal’s center) are shades of red, while those toward the bottom (near petal edges) are mixes of red and blue that produce magenta. How much red or how much blue is present determines how close the lower colors get to pure magenta (which contains equal amounts of red and blue). And if you imagine this image separated into horizontal thirds, the middle section contains the largest collection of mostly-red and mostly-magenta pixels adjacent to each other, which our eyes will interpret as the gradual boundary where the colors shift between magenta and red.

For complex reasons involving camera sensors, their capabilities, and their jobs as visual interpreters, the color red can be difficult to capture accurately, often rendered as oversaturated when photographed. With biological subjects like flowers — where each single pixel may represent hundreds of colored flower cells — the camera averages or balances the results within the range of colors it can reproduce. With flowers whose colors are closer to pure red, reducing saturation is often sufficient to restore their natural appearance. But with flowers whose colors are blends of red and magenta, the camera’s limits become more apparent, where red adjacent to magenta (which contains a lot of blue) produces an effect similar to chromostereopsis, the visual “vibrating” characteristic of red letters on blue backgrounds, or vice versa.

Here we see the original version of the photograph I shared above, where red is oversaturated and the boundaries between red and magenta demonstrate chromostereopsis…

… while simultaneously causing a red color cast across the yellow stamens at the center of the flower. Correcting colors in a flower with these tonal combinations becomes a bit experimental in Lightroom, where it’s necessary to reduce red saturation overall but also adjust blue hues so that the smooth gradients we saw in real life are faithfully reproduced in the image. Here I’ve posted the two photos side-by-side (before then after)…

…where (I hope) it’s apparent that the adjustments reduced the excess red, restored the gradual visual transition between red and magenta colors, and clarified the yellow color and detail at the flower’s center.

This is the first time I’ve produced a series of Camellia japonica photographs from my Oakland trips, because most of them are the same red/magenta blend for which I’ve only recently developed a color correcting method I’m satisfied with. They’ll likely get additional space here in the future; but I’m intrigued about Noel Kingsbury’s statement above that Camellias grow slowly but can live for centuries. I often pass by a pair of enormous Camellia trees — one with white flowers and one with dark red/magenta flowers — that are fifteen to twenty feet tall with equivalent width. I’m wondering now, since they’re located in the cemetery’s original six acres, if they may date back to the cemetery’s 1850 founding or at least have been growing there for several decades. I’ll see what I can find out; stay tuned!

Thanks for reading and taking a look!