-
-
Notifications
You must be signed in to change notification settings - Fork 1.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[question] Decreasing sequential outliers #1065
Comments
There will be a new sequential mode on the ImageMatching node in the next release, that can be combined with voctree. |
Should that be already in git? |
Is there anything that would use timing information between frames? We are currently missing in Meshroom:
@natowi as I put in the post I have used SequentialAndVocubaryTree, and this produced this problem :( |
Please keep open. |
@fabiencastan imagine a linear movement of a camera on a single axis to make it clear, this trajectory should be interpolatable, especially for a video. I would expect that the camera position between photo 2 and 5 should be detected as linear. I notice that meshroom does not take this into account, hence creates outliners such as 3 and 4, which does not make sense if I would consider this a trajectory. Therefore I wonder, if sequential is choose, interframe delta's should guide the trajectory, which could provide a heuristic such as a path. |
Ah ok, so temporal continuity constraints in SfM. |
Describe the problem

This is a drone flight, as you can see there is a gap in the flight, which later appears side by side.
This is likely to happen because of a similar repeated structure, two similar radio telescopes. The results are obtained using SequentialVocubularyTree with 40 neighbours.
The basis question is: is there any way to strongly prefer sequential behavior (which I would perceive as images placed on an ordered path) over out of order movements? My understanding of the sequential neighbours was that these are the photos which are being cross checked for similar features, so having sequential neighbours = 2 does not sound right to me.
The text was updated successfully, but these errors were encountered: