-
What Ever Happened to Metadata?
-
Coming back to the original question.
[Simon Ubsdell] “So my question is: will we always still be using containers of one kind or another, alongside these metadata filtering options, or can you see a time when we might move over entirely to the metadata-driven model?”
What if the user doesn’t have to choose? I think this is a thing that developers can drive, especially if they make it transparent. I think that people tend to do whatever the software makes easy.
[Simon Ubsdell] “It’s worth pointing out that when we drop a file into a container of any kind, we are applying a kind of virtual metadata to it – we are describing the file in a particular way such that we will be able to recall it at will. I can make a folder called “Boston Shoot, 05-15-2016″ and drop all my relevant camera dailies in there such that I can later recall when and where they were shot. However, if I take any of those files out of that container, the virtual metadata doesn’t follow along with it – it’s volatile and ephemeral. If instead I embed the place and time of the shoot into the file, I can use a filtering method to access it regardless of where I happen to have put it. And that seems to me to be a major advantage.”
Going back to your idea of a smart search (and mine of analytics), what if the computer could use people’s ephemeral classifications to derive more permanent metadata and associate it with the file via XMP?
Google and Bing both know that Boston is a place. Why can’t a cloud-enabled NLE?
Just putting clips in a folder named “Boston Shoot, 05-15-2016” should be enough for the system to know to tag the media with Boston as its location. (Of course, you’d be hosed if you were doing a documentary on rock and roll in the 1970s and 1980s, but surely “location” would be the correct assumption in the overwhelming majority of cases…)
Walter Soyka
Designer & Mad Scientist at Keen Live [link]
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
@keenlive | RenderBreak [blog] | Profile [LinkedIn] -
[Walter Soyka] “Just putting clips in a folder named “Boston Shoot, 05-15-2016” should be enough for the system to know to tag the media with Boston as its location. (Of course, you’d be hosed if you were doing a documentary on rock and roll in the 1970s and 1980s, but surely “location” would be the correct assumption in the overwhelming majority of cases…)
“There’s actually a lot of work being done in computing circles on this sort of stuff under the banner of machine learning. This could be a hybrid of human and NLE sorting. For example, a human does the initial sort. The software analyzes what’s been done and “learns” the patterns. As you get deeper into it more becomes automatic or at least “pre-filtered” by the NLE. The software understands that you want more of X and therefore does it for you.
– Oliver
Oliver Peters Post Production Services, LLC
Orlando, FL
http://www.oliverpeters.com -
[Oliver Peters] “There’s actually a lot of work being done in computing circles on this sort of stuff under the banner of machine learning. This could be a hybrid of human and NLE sorting. For example, a human does the initial sort. The software analyzes what’s been done and “learns” the patterns. As you get deeper into it more becomes automatic or at least “pre-filtered” by the NLE. The software understands that you want more of X and therefore does it for you.”
I would be very wary of my NLE using the Cloud to “learn” patterns and take data from my footage and sorting, client confidentiality is an increasingly large issue.
-
[Steve Connor] “I would be very wary of my NLE using the Cloud to “learn” patterns and take data from my footage and sorting, client confidentiality is an increasingly large issue.”
I don’t think the technology is cloud-dependent.
Oliver
Oliver Peters Post Production Services, LLC
Orlando, FL
http://www.oliverpeters.com -
[Oliver Peters] “I don’t think the technology is cloud-dependent.
“Most of the machine learning tech I’ve seen so far is based in the cloud and not local
-
Here’s a real-world example of the need for post-delivery workflow metadata, from elsewhere here on the COW this week:
“Best practices for tracking an original file name after client delivery? Metadata?”
https://forums.creativecow.net/thread/2/1074041Walter Soyka
Designer & Mad Scientist at Keen Live [link]
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
@keenlive | RenderBreak [blog] | Profile [LinkedIn]
Reply to this Discussion! Login or Sign Up