Begin anywhere...
Today is a quiet day. I am the team lead in an industrial strength digital Archive implementation at a major Canadian daily newspaper - The Globe and Mail. I also represent the Photodesk's interests in a new pagination system we are beginning to install. I am able to search just 2000 hits (returns) in the Archive system - SCC's MediaGrid - yet I need to withdraw from circulation 3, 940 pictures. The problem was how to extract all images without yet deleting them off the server? I inserted ("infused") a dummy keyword - "Larvik" the town where my father was born in Norway - with reasonable assurance it would not appear in any keyworded image anywhere in our library of 300 000 images. So this was done to the first 2000 returns of my search parametre (all filenames beginning with "F-*", where "8" represents a wildcard and so hits will return any file/image with a filename starting with "F-". I then extracted those to my desktop. Next I performed the same search, but excepted any files with "Larvik" in Keywords. This returned the last 1 940 images and I then extracted those to my dektop as well. I then went back and removed "Larvik" from the Keywords of the first 2000 files, and the states of all "F-*" files returned to their original state. I now have all 3 940 files on my computer. Ta Da.
I am burning our entire legacy database (4th dimension's AXS image database) to DVD today. I copied all HiRes images to my computer over a few days a couple of months ago and can burn the disks more easily and reliably here than across the network. So far I have filled 7 single-sided DVDs. It is a great shortcoming of the 4D product that the IPTC records are kept separate from the HiRes image files. This explains, in part, our migration to a new digital archive for the paper's Image Library.
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Okay.
Its several hours later and I have done a thing or two since then.
I spoke with an assistant photo editor at Sports Illustrated about their image library and the robotics they use to retrieve images from it when user-requested. We are considering a similar setup (albeit with a newer jukebox), and I wanted a user perspective before we spend a whack (and I do mean a whack) of cash on a jukebox retrieval system.
For those who might wonder, the cost of keeping an image library online and instantly available across a network is large and can be huge. We have something here downstairs called a "SAN" - a Storage Area Network. Its basically one gi-normous set of hardrive inside a big box. Ours is made by E=mc2. You don't buy theses things, generally, you have them installed and rent space on them from the installer. That's where the expensiveness comes in. The alternative, a DVD jukebox served with a robotic arm is called "Near-line storage" and is limited somewhat by the lag-time between a user request for an image up here in Editorial and the robotic's ability to find the DVD the image is on, grab it, spin it up and send it across the network to the user. So instead of the HiRes image appearing nearly instantly from the clicked thumbnail the user must wait for a few seconds for the robotic arm to do its thang. We have been quoted 3-5 seconds for the robotics to perform the manoever described above. I would be happy with 6 or 7 seconds.
I also spoke with the Chief Librarian at the Seattle Times to discuss a usage issue in the common Archive software we each employ. Our popup list for inputting Keywords is corrupting at about 3 000 line items. It turns out this is a shortcoming of the Macintosh OS that has a strict memory buffer of 64K for popup lists. Once you get beyond this your newly added items to the list appear as gibberish and can cause the list to fail in its entirety. Mac OS X apparently has upped this buffer to 128K, but I think this really is unacceptable. I love Macs, but PCs have no memory buffer at all for popup lists... The solution in Seattle has been to use more generic master keyword terms, where ours can be very specific indeed: Television cameras, Television News, Television SetsTelevison Shows, Television Studios - instead of just "Television." So no solution to that problem.
As well, I have been in touch with Adobe tech support for one of our graphic artists in the business section who is having trouble with an apparent bug in Illustrator 9.02.
Time to go home soon.
Today is a quiet day. I am the team lead in an industrial strength digital Archive implementation at a major Canadian daily newspaper - The Globe and Mail. I also represent the Photodesk's interests in a new pagination system we are beginning to install. I am able to search just 2000 hits (returns) in the Archive system - SCC's MediaGrid - yet I need to withdraw from circulation 3, 940 pictures. The problem was how to extract all images without yet deleting them off the server? I inserted ("infused") a dummy keyword - "Larvik" the town where my father was born in Norway - with reasonable assurance it would not appear in any keyworded image anywhere in our library of 300 000 images. So this was done to the first 2000 returns of my search parametre (all filenames beginning with "F-*", where "8" represents a wildcard and so hits will return any file/image with a filename starting with "F-". I then extracted those to my desktop. Next I performed the same search, but excepted any files with "Larvik" in Keywords. This returned the last 1 940 images and I then extracted those to my dektop as well. I then went back and removed "Larvik" from the Keywords of the first 2000 files, and the states of all "F-*" files returned to their original state. I now have all 3 940 files on my computer. Ta Da.
I am burning our entire legacy database (4th dimension's AXS image database) to DVD today. I copied all HiRes images to my computer over a few days a couple of months ago and can burn the disks more easily and reliably here than across the network. So far I have filled 7 single-sided DVDs. It is a great shortcoming of the 4D product that the IPTC records are kept separate from the HiRes image files. This explains, in part, our migration to a new digital archive for the paper's Image Library.
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Okay.
Its several hours later and I have done a thing or two since then.
I spoke with an assistant photo editor at Sports Illustrated about their image library and the robotics they use to retrieve images from it when user-requested. We are considering a similar setup (albeit with a newer jukebox), and I wanted a user perspective before we spend a whack (and I do mean a whack) of cash on a jukebox retrieval system.
For those who might wonder, the cost of keeping an image library online and instantly available across a network is large and can be huge. We have something here downstairs called a "SAN" - a Storage Area Network. Its basically one gi-normous set of hardrive inside a big box. Ours is made by E=mc2. You don't buy theses things, generally, you have them installed and rent space on them from the installer. That's where the expensiveness comes in. The alternative, a DVD jukebox served with a robotic arm is called "Near-line storage" and is limited somewhat by the lag-time between a user request for an image up here in Editorial and the robotic's ability to find the DVD the image is on, grab it, spin it up and send it across the network to the user. So instead of the HiRes image appearing nearly instantly from the clicked thumbnail the user must wait for a few seconds for the robotic arm to do its thang. We have been quoted 3-5 seconds for the robotics to perform the manoever described above. I would be happy with 6 or 7 seconds.
I also spoke with the Chief Librarian at the Seattle Times to discuss a usage issue in the common Archive software we each employ. Our popup list for inputting Keywords is corrupting at about 3 000 line items. It turns out this is a shortcoming of the Macintosh OS that has a strict memory buffer of 64K for popup lists. Once you get beyond this your newly added items to the list appear as gibberish and can cause the list to fail in its entirety. Mac OS X apparently has upped this buffer to 128K, but I think this really is unacceptable. I love Macs, but PCs have no memory buffer at all for popup lists... The solution in Seattle has been to use more generic master keyword terms, where ours can be very specific indeed: Television cameras, Television News, Television SetsTelevison Shows, Television Studios - instead of just "Television." So no solution to that problem.
As well, I have been in touch with Adobe tech support for one of our graphic artists in the business section who is having trouble with an apparent bug in Illustrator 9.02.
Time to go home soon.
posted by fenris64 at
1:44 PM
0 Comments:
Post a Comment
<< Home