Well finding images for one! Or having the ability to build virtual copies for multiple renderings. Or the ability to build collections of images that don't reside anywhere near each other. Its called DAM (Digital Assist Management).
Thanks. I'm pretty good with the shoe boxes myself and the kinds of searches and organization you describe aren't that tempting. But I can imagine something that would be very tempting. Name-place/faces matching. Queries like:
All the pictures of my daughter and her friend Jane taken before our son was born.
What's the name of this guy?
Where is this street corner?
Any more shots of this dancer?
It's going to happen, you know. But I'm bound to have lost the rest of my mind by that time and and probably won't appreciate it any more than the current cool new features.
Oh, and right now, I just can't figure out why the *-right-arrow key in LR doesn't always take me to the next selected image. It seems to get stuck in tight little circles sometimes.
Oh, and right now, I just can't figure out why the *-right-arrow key in LR doesn't always take me to the next selected image. It seems to get stuck in tight little circles sometimes.
I've noticed that too, but for me it always seems to be when I got into Develop mode instead of Library mode. In develop mode, the right arrow operates the current slider. In Library mode, the arrows seem to work consistently for me.
[*]All the pictures of my daughter and her friend Jane taken before our son was born.
This could be done in LR, but of course, you need to keyword the images: Daughter, Friend. Then you could do a compound search using the keyword (parent and child or just both keywords) with a date range.
[*]What's the name of this guy?
[*]Where is this street corner?
[*]Any more shots of this dancer?
Like any database, you have to tag the images with keywords to do this (or GPS data in the case of the street corner).
Right. Any decent catalog manager will support the kind of searches John mentioned. But the problem is, you have to do a lot of data entry to make this possible. And do it consistently, or your search results will not be complete. GPS enabled cameras will probably be the norm in a few years, so that will take care of location. In the further future we can probably expect software pattern recognition capabilities that will do some keywording automatically. IMatch has a "search for visually similar" images today, but its accuracy is unimpressive. Eventually, that will improve, as will face recognition algorithms that will support finding multiple shots of the same dancer. We all want the Star Trek computer, but my sense is that it's at least a decade away, maybe more. For now, databases mostly need manual input to be useful.
I've noticed that too, but for me it always seems to be when I got into Develop mode instead of Library mode. In develop mode, the right arrow operates the current slider. In Library mode, the arrows seem to work consistently for me.
Now, I'm in Libray mode. I am happily moving forward and backward though my shots with *-arrow. Then I get to a particular shot, which is part a version of one I have edited. After that I get stuck in a small cycle of shots I have edited.
This could be done in LR, but of course, you need to keyword the images: Daughter, Friend. Then you could do a compound search using the keyword (parent and child or just both keywords) with a date range.
Like any database, you have to tag the images with keywords to do this (or GPS data in the case of the street corner).
Right. That's today. I want computer visual recognition. Google among others is investing a lot in this and I think it will eventually come.
It's especially interesting as a google feature. "Are there any pictures of my wife with that man last Saturday, anywhere on line?"
Right. That's today. I want computer visual recognition. Google among others is investing a lot in this and I think it will eventually come.
It's especially interesting as a google feature. "Are there any pictures of my wife with [t]that man[/i] last Saturday, anywhere on line?"
Hey John, I have those pix you're looking for J/K.
There are two or three folks I know of that are actively working in this space
and who've made some pretty good progress too. I've seen demos that can
identify night and day, flowers, cities, scenics with people and stuff with a
very broad scope. The demos were hundreds of photos so very small.
If you think about it, the problem has, on a much smaller scale, been
"solved" for the gaming and law enforcement industries (facial recognition)
as well as for the parking lot management/toll authority (license plates).
Moderator Journeys/Sports/Big Picture :: Need some help with dgrin?
Comments
Thanks. I'm pretty good with the shoe boxes myself and the kinds of searches and organization you describe aren't that tempting. But I can imagine something that would be very tempting. Name-place/faces matching. Queries like:
- All the pictures of my daughter and her friend Jane taken before our son was born.
- What's the name of this guy?
- Where is this street corner?
- Any more shots of this dancer?
It's going to happen, you know. But I'm bound to have lost the rest of my mind by that time and and probably won't appreciate it any more than the current cool new features.I've noticed that too, but for me it always seems to be when I got into Develop mode instead of Library mode. In develop mode, the right arrow operates the current slider. In Library mode, the arrows seem to work consistently for me.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
This could be done in LR, but of course, you need to keyword the images: Daughter, Friend. Then you could do a compound search using the keyword (parent and child or just both keywords) with a date range.
Like any database, you have to tag the images with keywords to do this (or GPS data in the case of the street corner).
Author "Color Management for Photographers"
http://www.digitaldog.net/
Now, I'm in Libray mode. I am happily moving forward and backward though my shots with *-arrow. Then I get to a particular shot, which is part a version of one I have edited. After that I get stuck in a small cycle of shots I have edited.
Right. That's today. I want computer visual recognition. Google among others is investing a lot in this and I think it will eventually come.
It's especially interesting as a google feature. "Are there any pictures of my wife with that man last Saturday, anywhere on line?"
Hey John, I have those pix you're looking for
There are two or three folks I know of that are actively working in this space
and who've made some pretty good progress too. I've seen demos that can
identify night and day, flowers, cities, scenics with people and stuff with a
very broad scope. The demos were hundreds of photos so very small.
If you think about it, the problem has, on a much smaller scale, been
"solved" for the gaming and law enforcement industries (facial recognition)
as well as for the parking lot management/toll authority (license plates).