Three Image Requirement - Dumb Software!!!

  • 1
  • Idea
  • Updated 11 years ago
Improve the matching to be able to work with fewer images - the requirement of three images with common content is ridiculous.
Photo of Ray

Ray

  • 1 Post
  • 0 Reply Likes

Posted 11 years ago

  • 1
Photo of David Gedye

David Gedye, Principal Group Manager

  • 176 Posts
  • 78 Reply Likes
Ray -- please write your own bundle adjuster that can produce consistently decent 3D models with less than 3 matches on a point. We'd love to license it from you!
Photo of Drew Steedly

Drew Steedly, Employee

  • 15 Posts
  • 3 Reply Likes
Actually, the synther can generate a 2 image synth :-). The 3 image limit is artificial and will be lowered to 2 in a future release. For now, a workaround is to put in a random extra picture (not very elegant, I know).
Photo of Nathanael Lawrence

Nathanael Lawrence

  • 795 Posts
  • 55 Reply Likes
Actually, I prefer to add the same photo twice for two image synths. :)
Photo of David Gedye

David Gedye, Principal Group Manager

  • 176 Posts
  • 78 Reply Likes
I think Drew and I are answering two different questions. I was talking about the recommendation in our Photography Guide to capture each point in the scene in three photos. Drew was talking about the fact that the "synther" won't start until you've given it three photos. Not sure which issue you were questioning, but you've got answers to both of them now!
Photo of Joshua Podolak

Joshua Podolak, Employee

  • 63 Posts
  • 13 Reply Likes
You mean to say we'll never get it to synth with 1 image? Is that like.... .....trivial?
Photo of awalling

awalling

  • 1 Post
  • 0 Reply Likes
Is there anyway to suggest a synth? I uploaded 4 pics of the same room, but it couldn't pick up the meld....
Photo of Joshua Podolak

Joshua Podolak, Employee

  • 63 Posts
  • 13 Reply Likes
Hi awalling:

To answer your question: Not currently.

If you want Photosynth to pick up the meld, it needs enough information in the pictures to figure out where the photos were taken from.

It's basically this big mathematical optimization that requires enough constraints (in the form of "this point is the same as that point, so they must be in the place in space") to figure out where all the pictures were taken with respect to each other, so we need at least three points in common over three images, or things won't meld.