Image Processing | c u r i o

Stereoscopy Digitized

May 21, 2015 by lelgin | 1 Comment

Last fall, the Brown University Library acquired a set of 100 stereographs made of Palestine in 1901. Stereographs are made using stereoscopy: a technique that involves creating the illusion of three-dimensional space using two-dimensional imagery. In photography, this means making two images that are just slightly offset from each other, and using a special viewer (a stereopticon) to look at both images at once. The illusion of depth between different spatial elements emerges as our brains attempt to reconcile these two different images into one, creating a (seemingly) three-dimensional scene.

One of the stereographs of Palestine, showing the two offset images that will eventually form the single three-dimensional image.

As is the case with most stereographs of this kind, these photographs are albumen prints that have been mounted onto thick card stock. Over time, many of these cards develop a curvature towards the mounted-print side. This makes it very difficult to digitize the images using a flatbed scanner; you can’t get even focus or lighting on a non-flat object, and it’s easy to rip the images or crack the emulsion of the photographs if any attempt is made to flatten the cards. This set of stereographs was no exception, and many of the cards had a distinct curve to them. To account for this, I digitized them using our reprographic camera so I could light them effectively and avoid the need to flatten the object altogether.

3dslidermaker

The next step was to create a three-dimensional image from each stereograph. There are a number of software applications that enable the user to merge two images from a stereograph into a single image that can be viewed as three dimensional using 3d glasses. I chose 3d Slide Maker, a freeware Macintosh application made by Mike Cook. This software lets the user upload the left and right sides of the stereograph, and then adjust the level of offset on the horizontal and vertical axes. It’s easy to view changes are you make them, so if you have a set of 3-D glasses you can check your work as you edit. This software also allows for additional imaging adjustments to correct any problems that creating the final 3-D image, or anaglyph, has caused in the image quality.

Once each new image was complete and exported, it was added to the collection of images. Each stereograph has three images associated with it: the front of the card; the back of the card (many of these have writing about the scene pictured on the front); and the 3-D anaglyph. If you have 3-D glasses, you can view the image below as a representation of what viewing the stereograph using a stereopticon might be like. You can also click here for an animated gif that shows the offset of the images.

Palestine033_03d

This collection will be displayed at Brown through 100 stereoscopic viewers in Fall 2016, curated by Ariella Azoulay and Issam Nassar, under the title Time Travelers in Palestine – Stereoscopic Journey.

Reconstructing the Berrigan Airplane

May 6, 2015 by btylr | 2 Comments

Brown University Library’s Harris Collection of American Poetry and Plays, housed at the John Hay Library, contains a wealth of poetry-related ephemera. A promotional flyer from 1969 — designed to become a paper airplane glider — was recently acquired for the collection. The flyer advertised a February 5, 1969 reading at the Poetry Project at St. Mark’s Church (New York City) by Ted Berrigan (1934–1983; born in Providence, RI). As the previous owner of this item noted, the flyer itself was likely designed by Joe Brainard, who had collaborated with Berrigan in the past. (The original artwork resides in the Henry W. and Albert A. Berg Collection of English and American Literature at the New York Public Library.)

Since the flyer only becomes fully legible when folded, we decided to reconstruct the plane: our high-res. photos of the newly acquired printed copy, digitally processed to use as high-contrast line art, became a double-sided print ready for folding. Shown below is a view of one side of the flattened flyer, the reprinted flyer assembled as a paper airplane (underside view), and evidence of a successful test flight.

Flattened newly-acquired print

Preflight snapshot of folded reprint

Test flight

Portrait un-retouching

April 21, 2015 by lelgin | 1 Comment

A few weeks ago, I was given a photograph to digitize – a pretty standard request in DPS. The photograph is a portrait of Alexander Nesbitt, co-founder (with his wife Ilse Buchert Nesbitt) of the Third & Elm Press; the Brown University Library has a sizeable archive that includes “art and design work from the Third & Elm Press; in part, joint art and design work of Alexander Nesbitt and Ilse Buchert Nesbitt for other than Third & Elm; and in part the papers of Alexander Nesbitt, including correspondence, writings, miscellaneous papers, academic papers, non-academic lectures, biographical/personal materials, and museum objects” (from Brown University Library catalog record).

before It’s a lovely portrait, and one of the only portraits of him from that time period. The only problem? At some point, the image was retouched to mask out a large area around Mr. Nesbitt’s face – giving this unusual halo effect that proved very distracting. Normally, our archival digitization workflows aim only to capture the originals as accurately as possible, but this was a special case for this particular collection, so we decided that we would keep two copies of the scan that I’d made. We kept one version that included the original retouching, and a new version where I eliminated the halo, and attempted to recreate as best as possible the image underneath. We would add both versions to the Brown Digital Repository, and also wanted to be able to make new prints of the image – printed using archival processes and fine art papers – that we could use for display to protect the original from damage.

The photograph itself was also showing some aging and use problems: there were some minor color shifts in the photograph (the sepia-ish light brown tone was turning to yellow-orange in the highlights), and there were some bends and tears in the paper. I was able to correct those relatively easily using straightforward image processing techniques. I also removed some dust and scratches from the original printing. Then, I moved on to the main job: removing the while halo, and recreating all the areas underneath the white mask. This was relatively easy to do when it came to the background, but when it came to Mr. Nesbitt’s neck, suit, shirt, and tie, it was far more challenging to recreate an accurate and convincing image.

I was able to determine how to recreate the tie relatively easily – I did basic image searches for men’s ties and tried to find some good matches to work from. The shirt and coat lapels were trickier, because they’re a specific style. The coat lapels in particular required a little more investigation, especially to get the notch correct and to make sure the angles and spread looked right. Here, my research led me to Hollywood:

I used these images from Indiana Jones and the Temple of Doom (1984) and The King’s Speech (2010) as guides in recreating the suit jacket for Mr. Nesbitt. I did not use the images to add to the file I was working on; rather, I used them as reference as I recreated the lines, textures, and shadows of the jacket through various processing techniques. When retouching was complete, we had prints made of both images: the original with basic color adjustments and tear repairs, and the fully retouched version with the halo removed. Here’s the before and after, from raw scan to completed file:
before-after

Black, blue, and gray all over

March 18, 2015 by btylr | Comments Off on Black, blue, and gray all over

Last month a blue-and-black dress sparked a mass-scale debate about color theory. Even the New York Times and magazines like Wired eventually weighed in, explaining how color perception is contingent upon context and light sources.

In honor of the unexpected media attention to color theory, Curio features below a gray-and-black dress from c. 1866, reproduced from Fashion in Paris: the Various Phases of Feminine Taste and Aesthetics from 1797 to 1897 (New York: C. Scribner’s Sons, 1898) for the Library’s Paris: Capital of the 19th Century online project (the project’s items are also available via the Brown Digital Repository).

Page from Digital Production Services

This illustration was hand-colored for Uzanne’s book, and the tones used to represent the dress were not painted as scientifically neutral-gray hues. Traditional representational painters counsel “never use black paint” when rendering scenes, since in reality even black rarely visually appears as “pure” black, depending on the particular light sources illuminating an area. Scanning or digitally photographing hand-painted artwork often only exacerbates color uncertainty, if images are captured without a known neutral-gray or neutral-white reference point. For example, the same page illustration, scanned for Google Books’ digitized version of the volume, ends up looking even more blue-cast, at least in part due to the contrast-boosting post-processing applied to Google Books’ page images:

Page from Google Books

To test your color perception, try X-Rite and Pantone’s Online Color Challenge (taking the test on a calibrated and profiled monitor definitely helps). By chance, coinciding with the media frenzy about the black dress on Tumblr, the philosopher of perception and consciousness David Chalmers — who has written about color perception — was giving a series of Royce Lectures at Brown, February 24–27.

Capturing the Transit of Venus

March 2, 2015 by lelgin | 1 Comment

Recently, as part of Brown’s 250th celebration, my colleague and I were asked to photograph a selection of the many historical objects at Brown. Among these was the Transit of Venus telescope, a Gregorian reflecting telescope made by Watkins and Smith of London, and donated to the University by Joseph Brown. The telescope is named for the event it was purchased to observe: the movement of Venus across the sun. The transit of Venus was observed by Brown, Benjamin West, and others on June 3, 1769 – and more recently on June 5, 2012.

The final version of the Transit of Venus telescope, photographed on site in the Lownes Room.

Thankfully, my colleague and I did not need to address any planetary movements to photograph this object; however, we did need to photograph it on-site in the Lownes room at the John Hay Library. We brought our backdrop, lights, and camera, and it was a relatively simple setup. We were careful to set up the camera so that the lens and angle of view did not distort the telescope in any way; our main challenge was actually positioning the lights. The telescope is incredibly reflective, and while changes in tone help show the shape of the telescope, and some highlights are useful for showing surface texture, initially we were getting long streaks of blown-out highlights, causing a visual distraction within the overall image.

Our straightforward lighting setup gets the job mostly done – all we have to deal with is the large highlight streak in the center of the telescope.

After trying multiple approaches – from changing the position and distance of the lighting, to a system of reflectors – we decided that our best option was to make two captures to merge together. To keep image distortion to a minimum, we decided to keep the setup largely the same, but to reposition the angle of one of the lights for each shot. Once finished, we had one image that was properly exposed except for a blown-out streak on the right side; and another with the lighting problem on the left. Layered in Photoshop, it was easy to merge the images to create a single image that accurately captures the telescope without the distracting highlight.

The two vies of the telescope – the only differences are the positions of the highlights, created by repositioning and changing the angle of our lighting equipment to reduce the glare.

Photographing Clara

May 31, 2013 by lelgin | Comments Off on Photographing Clara

As a followup to last week’s post about John Hay, I thought it would be a good time to discuss photographing the portrait of Clara Stone Hay. The wife of Brown alumnus John Hay, Clara Stone Hay’s portrait hangs in the John Hay Library, where the university’s special collections are housed. I was charged to photograph the portrait for the 2010 edition of Special Collections of the Brown University Library: A History and Guide.

The John Hay Library, home to many of Brown’s Special Collections.

Photographing the portrait was no easy feat. The painting had to be photographed in place; it is too large and too heavy to move for photography. This made controlling many aspects of the lighting incredibly difficult. The painting is done in oils, which have a reflectance that can range anywhere from a slight slimmer to a strong glare. Additionally, the painting is behind glass – and decidedly not the anti-glare variety. Both these obstacles can be dealt with through careful placement of the lighting, including adjusting the angle and directionality of the lights to minimize reflections. However, this painting is in a marble hallway, with windows and reflective surfaces at every turn. While no windows shine directly onto the painting, the light that they bring into the hallway not only interferes with the lights I had brought to illuminate the painting, but they also create hot spots and highlights on the work itself. The other paintings and objects in the hallway also create their own reflections in the glass.

The hallway in the John Hay Library where the portrait of Clara Stone Hay was photographed. Clara’s portrait is top right.

A tilt-shift lens, showing the shift to the right.

I’m a big believer in preempting as many imaging issues as possible during the actual capture. Because I had scouted this painting beforehand, I knew about the glass and the lighting issue. The primary challenge with the painting being behind glass is that there’s no way to photograph the painting straight on, without myself or the camera (or both) being directly reflected in the image. I brought in my tilt-shift lens (also called perspective control, described here) so that I could photograph the painting from the side.

The t-s lens allows the photographer to photograph an object from the side, above, or below, with keeping the lens parallel to the object (otherwise, we would see great distortion). This is called shift: the lens and film plane/digital sensor are parallel to the object and to each other, but the image circle shifts within the camera, allowing you to capture a different area of the scene than that which is directly in front of you.

I took multiple shots with the tilt-shift lens, bracketing for exposure. I was able to capture a faithful reproduction of the painting, which required very little processing. I had to fix some minor distortion on the right side of the frame – an artifact of the lens not being 100% parallel with the painting. I also had to reduce some of the reflections on the painting which were unavoidable – the painting opposite that of Clara could not be moved, and could not be removed during capture. The progress of these changes and the final version are shown below.

The stages of processing the Clara Stone Hay portrait.

Garibaldi Returns

April 26, 2013 by btylr | Comments Off on Garibaldi Returns

: Garibaldi scroll (side 1)

In late-summer 2007, the Brown University Library contracted with Boston Photo to photograph both sides of a unique oversize panorama scroll depicting the life of Garibaldi. After the initial capture files were processed and digitally merged together by Digital Production Services staff in the Library, they were used to develop a website about the panorama. The process of reconstituting the panorama’s narrative scenes from the initial set of discrete digital photographs is described below:

Example from the end of side 2: Yellow line represents edges of initial digital files, which were then merged together in order to visually preserve the narrative continuity of scenes.

Example from the (supplementary) end of side 2: Yellow line represents edges of initial capture files (which included on average .5–1′ of overlap). Files were then digitally merged together in order to visually preserve the continuity of narrative scenes.

The sheer dimensions of the Garibaldi panorama — 4.75′ tall and 260′ wide on each side — presented unique digitization challenges. Boston Photo Imaging, a digital imaging company, was contracted to capture digital images of the panorama as it was unrolled across a custom-built wooden platform. Using a vertically mounted Better Light 4″ x 5″ digital scan back, capturing both sides of the panorama took three days and resulted in 91 digital image files, each file ~244 MB and representing ~6.5′ of horizontal width (including on average .5-1′ of overlap, in order to facilitate subsequent image merging). The scan back captured 300 dpi RGB TIF files; given the height of the device this resulted in an effective real-world resolution of ~137 dpi at the actual size of the panorama.

Because scenes within the panorama’s visual narrative do not correspond to the uniform width used in the capture process, sets of capture files were digitally merged together five at a time by the Brown Library’s Center for Digital Initiatives [now Digital Production Services] staff, at full capture resolution within Photoshop CS3/v10, and then individual scenes were isolated and saved from these roughly 30-feet merged sections. A continuous image of each side of the panorama was produced by subsequently merging sequences of these five-section composites at a reduced resolution. As part of this process the plastic-over-board background initially visible along the top and bottom edges was digitally removed, and each merged group of five was slightly rotated in order to compensate for some inevitable alignment drift produced during the unrolling process. Tonal levels and saturation values were slightly adjusted, and files were moderately sharpened for full-resolution and reduced-resolution delivery sizes. [Read more from “Behind the Scenes”]

In early 2013, the Center for Digital Scholarship and its student employees substantially upgraded the Web presentation of the Garibaldi scroll. In particular, one can now view the scroll’s narrative scenes directly alongside their accompanying descriptive texts. Flash-based animated views of the scroll are also still available on the site, which allow click-throughs into zoomable views of each narrative scene.

In addition to the redesigned website, the scroll has also been (briefly) mentioned in a new book from MIT Press, Illusions in Motion by Erkki Huhtamo, and has been featured as a teaching tool in the Patrick Ma Digital Scholarship Lab, by way of Massimo Riva’s Fall 2012 course on Garibaldi and the Risorgimento.

c u r i o

Imaging rare, unusual, and intriguing objects at the Brown University Library

Category Archives: Image Processing