How to load the models one by one when there are many(more than 1) models reconstructed from Visual SFM?

[xinqi Wang]

    Excuse me, I would like to ask a few questions.

    I did Visual SFM with different datasets. Firstly, I got success from a small image dataset containing 67 images captured for a campus gate because eventually I saw the 3D reconstruction display as I expected and the run log file is the Success_68images.log.

    Then I got failed from a large dataset containing 1344 images(The Madrid Metropolis images Dataset from link: https://research.cs.cornell.edu/1dsfm/ ) because finally I only got a few chaotic points of 3D reconstruction and the run log information after the sparse reconstruction started is the Seem_Fail_1344images.log. 

    I want to ask questions about the sparse reconstruction part(after the long matching process is finished).

    In the successful example, the log file displayed in sparse reconstruction part that :

    

1 model(s) reconstructed from 68 images;

68 modeled; 0 reused; 68 EXIF; 

10MB(9) used to store feature location.

    

   It only generate 1 model from all images and the result is fine. Then in the dense reconstruction part the log file showed that:

Run dense reconstruction, finished
Totally 14.900 minutes used

    The successful example took me 15 minutes to finish the dense reconstruction part. 

    However, in the failed example, the log file displayed  in sparse reconstruction part that:

    34 model(s) reconstructed from 1344 images;
    729 modeled; 24 reused; 713 EXIF; 
    196MB(173) used to store feature location.

    It generated 34 models and finally it showed me the chaotic one.Then  in the dense reconstruction part the log file showed that:

    Run dense reconstruction, finished
    Totally 10.000 seconds used

    The failed example took me only 10 seconds to finish the dense reconstruction part.  It is a failed reconstruction in my opinion.

    My questions are: 

    1. In the failed example, Can I load the 34 models one by one? How to do that? Which button should I click after the sparse reconstruction finished?

    2. At present, I guess there is 1 available model can reach my respect in the 34 ones and if my assumption is right,  can I pick it up and generate the nvm file only for the 1 model?

    3. If the former 2 questions turn out with answers no, could you please give me help of the proper parameters of the sparse reconstruction part in nv.ini file in Visual SFM folder?

    

    Thank you.