It seems like there is unnecessary many images. You can watch this tutorial: https://www.youtube.com/watch?v=9e_NLp_FaUk how to capture images. In your case you can do perpendicular and tilted images with around 75% overlap from two different distances. Your images look a little bit noisy for me. Yes, ruller and targets could also help here and the cubes, as you won’t be capturing just flat area.