Do alignment settings affect reconstruction?

Hello!

 

I can do alignment in 2 ways:

 

I choose fast alignment settings (like image downscale factor 4), and alignment is very quick.

If not all images were aligned during the first alignment, I simply delete the small models and keep only the biggest model and run the alignment again.

Then I run the alignment again and repeat this process until I’m satifisfied the number of aligned / used images.

I choose slow and very intense alignment settings.

This takes much longer than running method #1 but results in the same number of aligned images.

 

I would like to know if both methods result in 99,999% the same model, or is there a noticable difference?

 

Thank you.

 

Hello Alice, 

the quality of the reconstructed model depends on the precision of the alignment and the precision of the calculated camera positions. 

So can method 1 or method 2 produce equal “precision of the alignment and … precision of the calculated camera positions”? Which is likely to be best, or is there no significant difference? i think that’s what Alice was asking.

If there is a significant difference depends also on the image set and your expectations.

As usual, I suggest to just try it out and compare the models, for example in CloudCompare.

Don’t forget to post the results here!  :slight_smile: