Wikipedia: Data Compression
Hey, something I know a little about! JPEGS! In college I did a lot of layout work for a student magazine, so we were always trying to figure out how little we could compress image files so that they'd look as good as possible, while not making the entire document too big to fit on a disk for the printer.
The thing that gives me pause about this article is that it gives the impression that lossless and lossy systems are these two black-and-white polar opposite choices. With most image-editing programs (and some digital cameras), the degree of lossiness is on a sliding bar. So there are actually many different ways of striking the balance between fidelity to original and file size.
The example of run-length encoding is nice and straightforward (storing 25.888888888 as 25.[9]8). The "Theory" section is a little confusing to me, though. What are the 4 stages of the 4-stage model of lossless compression? Perhaps it depends on the content of the document?
Data Compression Basics
Very well-written for the layperson. Analogies are so helpful (such as thinking of acronyms as a sort of mental data compression).
Predictive compression works best with "natural" images. Based on the examples given it seems that these are also the hardest to reduce in lossless compression.
I think the description of lossy compression as preserving meaning rather than data is a big improvement on the wikipedia article's definition. I have to revise my earlier statement about photo compression -- we weren't trying to compress the photo as little as possible, but to lose as little meaning as possible.
Imaging Pittsburgh
As a big proponent of online archival photo databases, it's interesting to see the challenges in putting together such a collection. One of them being metadata of course.
"The team has spent tremendous energy cataloging the images; what better way to tell some of Pittsburgh’s stories than by leveraging the controlled metadata? This will allow us to provide curatorial context and guidance to the images in the collections. Our avenues for exploring the collections will encompass time, place and theme. Some of our theme ideas include 'Pittsburgh at Work,' 'Pittsburgh at Play,' 'Pittsburgh at Home,' and 'Pittsburgh Personalities.'"
Metadata = better access/searchability for users. I wonder if they have thought about letting users themselves help tag photos? Maybe someone looking at the photos will recognize a person or street previously unidentified.
Youtube and Libraries
youtube is a great teaching tool! Especially for kinetic things that are hard to understand through verbal description or still pictures.
like sign language:
http://www.youtube.com/watch?v=7cayMkVidpc&feature=related
or cooking roti:
http://www.youtube.com/watch?v=Zg0p5TXi_Cc
I don't know if instructional videos should be a top priority for libraries at the moment-- it doesn't seem worth taking time/staff away from other duties. But if the resources are available it could be a nice idea, especially for online students, or those working from off-campus (Lord knows I could have used help the first time I was trying to access the online journals from home). Sometimes I feel like college librarians in particular spend a little too much mental energy trying to seem hip and with-it for the benefit of the undergrad population.
Subscribe to:
Post Comments (Atom)
2 comments:
The idea of letting users tag the photos wasn't something I had thought of. I'm always really torn on that sort of thing because on the one hand it would be really helpful (like you said, identifying streets or people or houses), but I'm always wary of mistaken tagging or inconsistent tagging.
I like the idea of letting users tag, and I hadn't thought of that either. I think that our readings and our experiences show that people will be interacting more with the resources than ever before, and providers need to accept it, and shape it in the right direction.
Post a Comment