Talk about the high end!!!
My post on this topic didn't seem to raise any interest, so I'm respinning.
There is this great article in The New Yorker which you can read online here.
It seems that in 1998, the Cloisters restored the Unicorn Tapestries. In the process they replaced the old linen backing. When the old backing was removed, a mirror image was revealed that had never been exposed to light and thus had not faded nearly as much as the front. These tapestries were so carefully woven that the backs are essentially as nice as the fronts so this was an amazing moment.
The Met decided to photograph the tapestiies, front and rear, during restoration. They did this by shooting lots of overlapping closeups each with a FOV of about 3x4 feet, planning to stitch them together later on. These tapestries are very big, the largest is 12x14 feet. The resulting data for 10 tapestries filled 200 cds. (I think my nubers are about right here. My memory is a little fuzzy and the article is obviously oversimplifying.) So there was about 1GB per tapestry.
Suffice it to say, that when the Met phogotraphers loaded all this stuff into PS and tried to stitch it, nothing good happened. So the data sat on the shelf, a problem waiting for a solution. Last year, the Chudnovsky brothers, a pair of famous, smart, and very excentric New York mathemiticians who like to build their own supercomputers from mail order parts, took up the challenge. The problem turned out to be much harder than anticipated because the tapestries were actually moving in subtle ways while they were being shot.
So this is a really fun article. Read it, you won't be sorry. Once you do read it, answer me this question -- what digital Leica were the Met photographers useing in 1998? (See I even made it fit into this forum.)
There is this great article in The New Yorker which you can read online here.
It seems that in 1998, the Cloisters restored the Unicorn Tapestries. In the process they replaced the old linen backing. When the old backing was removed, a mirror image was revealed that had never been exposed to light and thus had not faded nearly as much as the front. These tapestries were so carefully woven that the backs are essentially as nice as the fronts so this was an amazing moment.
The Met decided to photograph the tapestiies, front and rear, during restoration. They did this by shooting lots of overlapping closeups each with a FOV of about 3x4 feet, planning to stitch them together later on. These tapestries are very big, the largest is 12x14 feet. The resulting data for 10 tapestries filled 200 cds. (I think my nubers are about right here. My memory is a little fuzzy and the article is obviously oversimplifying.) So there was about 1GB per tapestry.
Suffice it to say, that when the Met phogotraphers loaded all this stuff into PS and tried to stitch it, nothing good happened. So the data sat on the shelf, a problem waiting for a solution. Last year, the Chudnovsky brothers, a pair of famous, smart, and very excentric New York mathemiticians who like to build their own supercomputers from mail order parts, took up the challenge. The problem turned out to be much harder than anticipated because the tapestries were actually moving in subtle ways while they were being shot.
So this is a really fun article. Read it, you won't be sorry. Once you do read it, answer me this question -- what digital Leica were the Met photographers useing in 1998? (See I even made it fit into this forum.)
If not now, when?
0
Comments
I certainly know of one place I want to visit if I ever make it to NY.
Thanks for posting!
Brad
www.digismile.ca
My math was way off. Guess it was just too early in the AM. More like 15GB per tapistry, right?
What I was trying to figure out was the camera resolution. Based on a tapistry size of 12' by 14' with each image covering only 3' x 3', you would need approx. 18 photos for one tapistry. That means each image is in the order of 1GB. Is that possible?
Brad
www.digismile.ca
I was trying to figure this out, too. Just based on the number of CDs, I got to this number, roughly 15GB give or take (what's 3 GB here or there at this point.) On the other hand the description of the camera and doesn't seem to make sense with these numbers. My guess is that the FOV was actually a lot smaller than the author quotes and/or the overlapp among images was pretty large. I think if I were shooting this without any real idea of when or how it would be stitched or used and knowing that there probably wouldn't be another opportunity to shoot for centuries, I'd take a lot of very close shots.
Another speculation. They are talking about a Leica digital camera in 1998. Would pros at this level have used anything like that? Was there anything like that? I guess they shot film and scanned it. My experience with professional film scanning geeks is that they can basically a file of any size you like from a scan and they belive there is always more resolution and quality to be gained as they go up in size. So film scans, a lot of overlap, and smaller FOV would all work together to make the numbers more plausible.
Yeah, two places. I've been to the Cloisters a number of times and know and love the the Unicorn Tapestries. Now I want to go to the Chudnovskys' lab in Brooklyn as well.
I don't know about the Cloisters, but both the Modern and the Boston Museum of Fine Art allow photography. The Cloisters would be an incredibly fun place to take pictures. But follow Andy's advice. Don't try to compete with the fine art photography specialists. This article shows how impossible that would be. Find your own spin on it.
About the Unicorn Tapestries. There is another set of at the Cluny Museum in Paris. The ones in Paris have a red background while the ones in NY have a blue background. I personally like the ones in NY better, I think they are more mysterious and somehow deeper in theme. But the ones in Paris are just as fine and the Cluny is at least as interesting and fun to visit as the Cloisters.
I've always been a unicorn fan, since I was a very young child, and always enthralled by midevil beasteries and the like. And I like stitching and photography and the techniques that make this work. So I guess you could say this article might have been written for me.
But then I read an article this evening in the May 2005 issue of Outdoor Photographer where the writer had a high res scan of a 35mm film on a Tango drum scanner with a final file size of 470MB.
So I think you're right. They probably used film or transparency and did a high res scan.
www.digismile.ca
It could easily have been a medium format camra, actually. I think the author didn't have all his technical facts 100% right and could easily have substituted Leica for Hasselblad or Rollie. It seems a lot more likely that the Met's Fine Art photographers would have used one of these than a Leica for this particular task, no? Wouldn't you? It's not like they wouldn't have had them or be able to justify going down to Fotocare and buying one.
I suspect they were shot with medium format or possibly 4x5 film and then scanned on a high end drum scanner.
I doubt museum did the digitizing itself, but probably outsourced this. How many commercial art/fashion studios could handle digital images in 1998? Darned few.
I did a brief WEB search for Leica digital 1998 and came up empty.
Very interesting article, John.
Moderator of the Technique Forum and Finishing School on Dgrin
I know everyone in the article except the Mathematicians. The Leica Camera is not the one you're familiar with but one that we sold to Museums and Conservators called the S-1. It is a scanning camera not instant capture. Very unlike a "Leica". It is no longer manufactured.
The article is pretty exciting. Barbara Bridgers called me two weeks ago to tell me about the article. Scott and Howard are consultants that we have used for 10 years. The Met,MOMA and the Guggenheim are clients of FotoCare.
Jeff
A little googling and I found some references to this camera:
http://www.lib.utah.edu/digital/leica.html
http://www.shutterbug.com/show_reports/0599sb_professional/
Now it all makes a lot more sense. One of these would have been an appropriate tool for this assignment. The size of the output is roughly right, especially if you allow for some overlapp and imperfect packing of the data onto the CDs. It even clears up one more thing from the article. Scanning cameras have very long exposure times (that's why they are distinguished from "instant capture"). The article mentions that a couple of tiles were discolored because a door was opened during the exposure and allowed in some flourscent light. That fits.
I've posted about Jeff and Foto-Care several times on dgrin:
http://www.dgrin.com/showthread.php?t=2811
http://www.dgrin.com/showpost.php?p=7686&postcount=5
If you are in NYC, it is well worth a trip to visit his store, even if you don't buy anything.
www.digismile.ca
Moderator of the Technique Forum and Finishing School on Dgrin
apart was the quality of the capture that proved that the total was more
than the sum it's parts. Very high quality capture.
Yeah, I can only just barely imagine.
Had an opportunity to visit the Cloisters yesterday, so I thought I'd post a snapshot of one of the Unicorn tapestries. It was rather dark in the room where they are displayed and I didn't have a tripod, so this was shot at iso 1600 with a D70 at 1/15 second. Not a great photograph but shows what the discussion is about...
Did you get any other good shots at the Cloisters? I think it would be a really fun place to have a camera. It's great they let you bring it. I think I'll go up there next time I'm in NYC. It's been way too long.
Also
I visited the Cloisters not long ago with my Sony V1 - here's a shot of one of the unicorn tapestiries from my vantage leaning on the floor (already seen above) and a gallery link to some more...
It's a neat place sitting way above the hudson river on the north end of Manhattan Island.
rEgards...Keith
http://kt.smugmug.com/gallery/381402
How did you get the sides vertical?
Perhaps some of you saw the PBS NOVA segment tonight on photographing and piecing together the portions of the Unicorn tapestry, featuring the Chudnovsky brothers. If you missed it you can view it online tomorrow here.
Very little about digital photography was discussed, but it was an interesting presentation anyway.
Reading the early part of this thread and seeing the period in which these were shot, I guessed that it probably wasn't the around-town portable digital camera we are now accustomed to but that it had to be one of those old, slow digital studio cameras that scanned one line at a time, requiring a subject that does not move. Thanks for confirming this.
And just one of those NASA sized sensors attached to a giant lens cam could have done the detail in one shot.
30 1Ds mkII cams shot in the same instant could have also done the task. No canvas shifting if it's all taken at the same moment.
I also though, how much overlap did they give each shot? Good stitching comes from a 30% overcoverage of each component shot.
Yes, there's always math involved, but I think they took the longest path around a relatively simple problem.
"You miss 100% of the shots you don't take" - Wayne Gretzky