Get a Metaclue.....
I continue to be extremely frustrated by how difficult it is to create and organize a cohesive collection of personal digital media (my photos, clips and videos). It has been my quest for several years to go back and fill-in my digital lifebook with the belief that someday this content will be of value to me or my kids. My dad’s photo albums are fading as is his memory – my folks moved to a condo a few years ago and he threw out all of his negatives – poof – gone with the trash………
For my kids, all of their lives I now have scanned from photos, converted from PhotoCD’s or were captured digitally to start out with. I’ve tried dozens of software packages and online services, maybe more ..... I’m kind of an anthropologist when it comes to this stuff. So here’s my plea to the companies and organizations that are creating the software, platforms and devices what we use in the capture and sharing of personal digital media –
Get A Metaclue!
The value of personal digital media is only partially in the bits and bytes that describe a scene or capture a memory. Digital memories are only valuable if you can find them. Findability of personal digital media is a problem that sounds like it requires teams of PhD’s in research labs to solve. That’s not true, while there are some very interesting technical challenges to try to create breakthroughs in the automated understanding of digital media, there are some very simple things having to do with capturing metadata that could be done that would improve things substantially. It’s the 80% solution. Here are some of those things:
1. Don’t trap MY metadata in some proprietary index or database. If I have taken the time to annotate and categorize my images, I own that data and want it to be accessible and portable. Google’s Picasa has three different ways to annotate you photos – labels, comments and keywords. Two of which I believe are only usable in Picasa.
2. Read and respect metadata from others and pass it on. Many programs fail to even look to see if you may have added metadata in some other application and it is stored in an EXIF tag or as IPTC data and marking it even worse, if you edited the picture in that unfriendly program, it will most likely strip out the previous metadata when you save the file. Poof, gone, like my Dad’s negatives.
3. Automate when possible. My Nokia Smartphone has a camera and my calendar. What about giving me the option to let my camera app talk to my calendar to pick up the fact that I’m at PMA in Orlando on the day the picture was taken? How hard is that! No need for computer vision over the network to recognize that I’m at Disneworld like some are developing, my calendar already knows.
4. Make it easy to add metadata at time of capture. So here’s an easy one. Let me automatically tag the pictures I take with my camera with my name as the photographer (if I turn on that feature on the camera). You could even make it easy to upload tagging info to your camera while it is connected to your PC that you could select on your camera before you started shooting (hey, this could come from your calendar too!). Services like Riya.com are already trying to make use of text recognition to extract information from your images and software like GPS Photo-link from GeoSpatial Experts reads a photo of your GPS screen to sync location and time with your photos. What about taking a photo of a piece of paper where you have written (in nice printing) the descriptive information/metadata you would like added to a sequence of images. When you transfer those images to your PC or upload them to a networked service, the image would be recognized as a metatag marker (maybe it has the word Start at the beginning), the metadata would be extracted and applied (I just ordered a ScanSoft product to see if text recognition works on my handwriting). You would then have another picture that said Stop and the end of the sequence. You could do the same exact thing with voice annotation and speech recognition (so OK, maybe you could use a few PhD’s!). You could even embed a thumbnail of the metatag marker image with each of the image files in case you wanted to reference it in the future.
5. Make it easy to set the “date a picture was taken” when it is a photograph scanned on a scanner. You have a print, you know the picture was taken on Dec. 25th, 1988, your scanner and scanning software creates a file with the date the picture was scanned. All downstream software uses the scan date to sort and organize that photo. It would be trivial for scanning software to allow you to also set the “date picture taken” date when you are scanning or even make the file creation date = the date the picture was taken. If you don’t do this at scan time, many downstream apps will only read the file creating date and log that file in their index – you can go edit it in other programs all you like and it is darn near impossible to get it to show up in the right place in a timeline. I tried talking to Epson who's scanner I use - they do not have a metaclue....
I’d love to hear people’s reactions to these ideas and others that we could pass on to the vendors who are building the products and services we’re trying to use.
For my kids, all of their lives I now have scanned from photos, converted from PhotoCD’s or were captured digitally to start out with. I’ve tried dozens of software packages and online services, maybe more ..... I’m kind of an anthropologist when it comes to this stuff. So here’s my plea to the companies and organizations that are creating the software, platforms and devices what we use in the capture and sharing of personal digital media –
Get A Metaclue!
The value of personal digital media is only partially in the bits and bytes that describe a scene or capture a memory. Digital memories are only valuable if you can find them. Findability of personal digital media is a problem that sounds like it requires teams of PhD’s in research labs to solve. That’s not true, while there are some very interesting technical challenges to try to create breakthroughs in the automated understanding of digital media, there are some very simple things having to do with capturing metadata that could be done that would improve things substantially. It’s the 80% solution. Here are some of those things:
1. Don’t trap MY metadata in some proprietary index or database. If I have taken the time to annotate and categorize my images, I own that data and want it to be accessible and portable. Google’s Picasa has three different ways to annotate you photos – labels, comments and keywords. Two of which I believe are only usable in Picasa.
2. Read and respect metadata from others and pass it on. Many programs fail to even look to see if you may have added metadata in some other application and it is stored in an EXIF tag or as IPTC data and marking it even worse, if you edited the picture in that unfriendly program, it will most likely strip out the previous metadata when you save the file. Poof, gone, like my Dad’s negatives.
3. Automate when possible. My Nokia Smartphone has a camera and my calendar. What about giving me the option to let my camera app talk to my calendar to pick up the fact that I’m at PMA in Orlando on the day the picture was taken? How hard is that! No need for computer vision over the network to recognize that I’m at Disneworld like some are developing, my calendar already knows.
4. Make it easy to add metadata at time of capture. So here’s an easy one. Let me automatically tag the pictures I take with my camera with my name as the photographer (if I turn on that feature on the camera). You could even make it easy to upload tagging info to your camera while it is connected to your PC that you could select on your camera before you started shooting (hey, this could come from your calendar too!). Services like Riya.com are already trying to make use of text recognition to extract information from your images and software like GPS Photo-link from GeoSpatial Experts reads a photo of your GPS screen to sync location and time with your photos. What about taking a photo of a piece of paper where you have written (in nice printing) the descriptive information/metadata you would like added to a sequence of images. When you transfer those images to your PC or upload them to a networked service, the image would be recognized as a metatag marker (maybe it has the word Start at the beginning), the metadata would be extracted and applied (I just ordered a ScanSoft product to see if text recognition works on my handwriting). You would then have another picture that said Stop and the end of the sequence. You could do the same exact thing with voice annotation and speech recognition (so OK, maybe you could use a few PhD’s!). You could even embed a thumbnail of the metatag marker image with each of the image files in case you wanted to reference it in the future.
5. Make it easy to set the “date a picture was taken” when it is a photograph scanned on a scanner. You have a print, you know the picture was taken on Dec. 25th, 1988, your scanner and scanning software creates a file with the date the picture was scanned. All downstream software uses the scan date to sort and organize that photo. It would be trivial for scanning software to allow you to also set the “date picture taken” date when you are scanning or even make the file creation date = the date the picture was taken. If you don’t do this at scan time, many downstream apps will only read the file creating date and log that file in their index – you can go edit it in other programs all you like and it is darn near impossible to get it to show up in the right place in a timeline. I tried talking to Epson who's scanner I use - they do not have a metaclue....
I’d love to hear people’s reactions to these ideas and others that we could pass on to the vendors who are building the products and services we’re trying to use.
posted by MyronK @ 1:31 PM
1 Comments:
At 5:22 PM,
Anonymous said…
Very valuable article. I hope somebody listens to it.
Especially point (2) can't be stressed enough. Every metadata in a media file should be kept! Only to strip the file down to the binaries is a kind of digital theft (as you mentioned it).
Thanks!
Post a Comment
<< Home