The Rose Window at St. Mary’s Cathedral, Austin, Texas

rose-window-600pxUpdate: This photo is available for purchase. Send inquiries to the photographer at

— — — — —

This picture is a composite of 33 separate exposures, assembled by hand in Photoshop. It shows the complete rose window at St. Mary’s in downtown Austin, and is likely the first photo ever to show the whole of the stained-glass design. You can see in the photo below how much of it is normally hidden from view behind wooden beams.

During last year’s restoration of the church’s towers, cracks were found in the stone spokes of the rose window. These spokes, I am told, are not just “window dressing” (har har), but are structural elements. So, in order to avert catastrophe, restoration of the rose window was added to the “to do” list . . .

Here is a picture of the window before any restoration. This was July 22, 2011. The wood is all cracking and rotting, the glass is dirty, and the leading is corroding. Window had issues, aside from the cracks in the exterior stonework.

Rose window at St. Mary's Cathedral

Close-up of some nastiness. Do I see pigeon-droppings? (It could happen. Also from the “unexpected happenings” file: a couple of weeks ago a homeless guy got into the church at night through the rose window—climbed the scaffolding and hopped in through a gap.)

Rose window at St. Mary's CathedralNow the window has been completely cleaned, and all the lead replaced. Not sure if window-makers and conservationists still use lead, or if something more human-life-friendly has been substituted . . . anyway, the stuff looks like grey metal.

The pieces were all stacked up waiting for me this morning.


I hung a white sheet from a background stand and shot an Einstein through it, through the back of the window pieces, for the tiny ones. Added one AlienBees for the 7-foot-wide (or -tall) pieces, for more even coverage. Almost none of these pieces was even or regular enough to be able to rest on its side without holding it, so I had the camera on a tripod and triggered it via infrared remote that I’ve had for 15 years and (so thankfully) works with my 5D Mk 3. I rested each piece on top of a shortish bookcase; the bookcase was centered over a line in the floor, over which line my camera was also aligned. I put masking tape along the edge of the bookcase’s top and marked the center, so each piece would be put in roughly the exact same place (big fan of rough exactness). I called for help with the 7-footers and the center medallion. Rested the center roundel on a small table and both of us held it upright.

Rose window-052

This could have compromised the focus on this piece, but you know what? If you want to know the truth, I know sharp and I know a near-miss, and this one is a lot closer to sharp, no kidding, so I’m not going to go commit suicide or anything (I just finished rereading The Catcher in the Rye and all).

Lightroom Screen Shot 2013-07-25 at 9.54.43 PM


Rose window Screen Shot 2013-07-25 at 4.57.45 PM

The above picture shows part of my Lightroom 4 catalog. I opened each picture in Photoshop and downsized everything to match the center roundel. Then opened a big document (15,000 x 15,000), threw down some guidelines, and started putting pieces in place. I would rotate and position each piece, and then add a layer mask to get rid of the white background, bookcase top, etc. This “progress” screenshot is from my first go—I had put all the “arms” in in the order I shot them, which ended up looking odd, so I started again from zero, with new guidelines and everything. If you look closely you can see that the arms all have either a blue or a green “dot” separating the main window section from the circular portion at the ends of the arms—except one arm has a red dot and one has an orange dot. So there are 8 greens, 6 blues, 1 red, 1 yellow. In my redone image (see below) I decided to put the orange dot on top, like it’s the sun, the red dot on bottom, like it’s the fires of hell (just making stuff up now . . . I just wanted it opposite the orange), the the blues and greens alternating evenly on each side. Each wooden frame has a number etched into its side, from 1 to 16, but the resulting color patterns appeared random when I arranged the pieces by number. So I just did what I wanted. This is not how the window was previously installed—but just because some workmen 100 years ago installed it that way doesn’t mean they were right. I mean, at the front of the church we’ve got a St. Joseph window that reads in ornate script, “sancjos tuseph.” Oops.

Hope you enjoy the pic and the brief discussion of its making. Here’s a slightly larger version of the pic above.


400-megapixel image of New Sweden Evangelical Lutheran Church

New Sweden Evangelical Lutheran Church

I’ve been experimenting a little bit with shooting panoramas with my 70-200mm lens, zoomed in to 200mm. The black and white church image above was taken in this manner, requiring 48 shots to obtain complete coverage of the area. (The area extends somewhat on all sides of the above crop.) The same shot done with a 50mm lens takes 12 shots—which is what one might expect, I suppose. 200/50 = 48/12.

It’s black and white because the day was overcast and ugly, and the color version merely serves to underline just how ugly. The final image, as shown above, measures 20,068 pixels square, or about 402 megapixels. But really I’m just glad that it stitched as well as it did. There’s only one spot I could see where it doesn’t seamlessly line up, but it is only visible at the largest size. And it’s on the central roofline, too—not in the trees or some obscure place.


Technical Concerns
The reason I’m surprised at the stitching’s not being worse is that I was just guessing as to the “nodal point” of the camera/lens apparatus. To make panoramas that stitch properly, one has to position the camera so that the optical center of the whole rig is placed directly above the vertical axis of the tripod, both left-to-right and also forward-to-back. It’s relatively easy to eyeball the left-to-right positioning—one can see whether the lens is over the tripod’s center. But determining the forward-to-back positioning is a matter of trial and error and close observation. You have to pan the camera back and forth with two fixed objects in view, one near and one far. When you pan left and right you observe the amount of parallax that occurs between the two objects. If they stay in the same place relative to each other, you have no parallax and you have found the right front-to-back positioning for that camera and lens and focal length. Problem is that zoomed in at 200mm, one can only pan about 7.5º before the objects leave the frame . . . and I can’t tell how much parallax is happening. Have only tried this in my apartment so far, though—maybe I need to go outside.

At any rate, I’m pretty confident that I’ll be able to get good results without buying more equipment. I was planning on buying the Really Right Stuff PG-02 VA with B2 LR II clamp, which is the newer and beefier version of their CRD rail (mine has a panning clamp on it). I thought I would have to hang the camera off the back of the vertical component, as with the standard 3-rail panoramic setup—which, with the 70-200, is too heavy for the old-skool panning clamp on my existing vertical rail—but as it happens, I seem to be able to clamp the 70-200 right into my existing vertical rail and obtain acceptable results. When I get my 600mm lens, I’ll revisit this question . . .


Here’s what the individual shots look like arranged in a grid. I assume that the misalignment of some of the pieces results from my guessing at the positioning of the camera, but don’t know for sure.