Timeframe for update allowing 'Add Photos/Edit' options?

  • 3
  • Question
  • Updated 11 years ago
Assuming that you are going to add the ability to add photos to a synth. Do we have a timeframe as to when this might get updated? Even a broad guesstimate? And I would pay money for an offline version as well. I have a beautiful machine I'd like to run this stuff on without being confined to a Web Browser.
Photo of Musicofangels

Musicofangels

  • 2 Posts
  • 0 Reply Likes

Posted 11 years ago

  • 3
Photo of David Gedye

David Gedye, Principal Group Manager

  • 176 Posts
  • 78 Reply Likes
Here's as much as I can say... I'd say more if I could be sure of it :-)

We have no current plans to sell an offline version of Photosynth for home use.

We DO have plans to add some form of private version of the online experience, so that you can control who sees your synths. This won't happen within the next month, but it's fairly high up on our must-have features list.

The ability to improve your synths by adding or removing photos is something that many people ask for when they first try Photosynth. The (relatively) easy thing for us to offer -- as some people have noted -- is an ability for the author to suppress photos. In other words -- use 100 photos in the reconstruction to get a great point cloud, but only offer 50 to the viewer because the other 50 are simply repetitive. I'm hopeful that we'll get that one by the end of the year, but don't hold me to it. Our aim is to be responsive to our community, and there are many many other must-have features that rank higher than this one.

Adding photos is both easy and hard. Even now, if you see that two parts of your room didn't connect, you can shoot a few more photos and re-synth them. Yes -- it takes a while, but the result will be worth it. Much more difficult, but often requested is an interactive tool that would allow the super-enthusiast to "hang" unmatched photos in space, and join the world together manually. Our team is unlikely to do this, but we definitely want to enable others to take a crack at this problem. More details about this soon.
Photo of Nathanael Lawrence

Nathanael Lawrence

  • 795 Posts
  • 55 Reply Likes
David, in regards to the manual adjustment of a photo to match another, I am not keen to have a tool that would rely completely on user-positioning of a photo.

What I would like is to maintain all the current links that the synther has determined but be able to, in the case of a missed link, request that the synther take a second pass. I'm sure that from your end, you will quickly say that the algorithm will not do anything a second time that it failed to do the first time, but I was thinking along the lines of a tool that would allow the user to 'tell' the synther, "Point A in photo 1 corresponds to point A in photos 7 and 34; Point B likewise, etc., etc." I would still trust Photosynth far more than trying to correctly position a photo in 3 dimensional space - armed with only a mouse. Is there really no likelihood of such a tool? I do realise that even this method would have to allow for some degree of user error as the chances of placing Point A perfectly on two or more photos is slim.

Essentially I still want Photosynth to do the calculation. I just want a way to insist to it that a certain minumum number of points in several photos correspond and then have it take a second pass using the user assigned points as a guide and proceed to drag all connected photos into position based on any newly discovered points of reference.

Am I as naive as I feel? I realise that I am asking a question about a method so common in computer vision and so fundamental to the way that Photosynth, itself, automates the positioning of things that I run the risk of making an absolute fool of myself, but I just had to ask.
Photo of David Gedye

David Gedye, Principal Group Manager

  • 176 Posts
  • 78 Reply Likes
Nathanael -- Your question isn't naive at all. In principle, what you are suggesting would probably work. It would be a little tedious though... you would probably need to establish 3 or 4 correspondences to tie each unregistered photo in. It's a good idea.
Photo of Nathanael Lawrence

Nathanael Lawrence

  • 795 Posts
  • 55 Reply Likes
If I'm going to wait for a multi-hundred photo synth to complete, I have no issue taking 15 minutes (or any amount of time) to tag however many correspondences are necessary. :)

Whenever adding photos/editing a synth becomes possible, please give this feature serious consideration! I realise that the goal it to have the process be as automated as possible, but as you pointed out above, this would be a very desirable feature for the "super-enthusiast".

P.S. Since, in this case, one is trying to make a connection between photos that the algorithm missed the first time around, would it be beneficial to use (slightly) larger versions of the few photos that one is attempting to link, or would the resulting calculations be too radically different from the data gathered from the usual small versions currently used in the big batch synth process?
Photo of Johan Benesch

Johan Benesch

  • 15 Posts
  • 6 Reply Likes
I agree with Nathanael Lawrence. there is more on this i this thread: http://getsatisfaction.com/livelabs/t...

just noting that when going back for some complementary shoots things might have changed, like cars parked in front of a building or people walking/sitting near a statue etc

I aspire to get a 100% synth one day :-)
Photo of Dane Jasper

Dane Jasper

  • 8 Posts
  • 1 Reply Like
David,

One thought. The challenge is two areas that don't meet because there were not quite enough transitional images between them. Your suggestion to take more photos and re-synth is fine, IF the scene is one that can be revisited.

However, often it cannot. For example, a vacation, or something temporary like an event. Or, something which has changed.

The ability to give the synther some hints to help it connect points it can't quite work out sure would be nice.

-Dane