ContextCapture Problem

Hello! My name is Roman. I'm an engineer and use ContextCapture. Last week we took a photos of bridge by drone Phantom IV. We  took a photos by two session. At the first session we filmed one side of a bridge at the second session we filmed another side of a bridge. Then we have taken together photos and created 3D model of bridge in ContextCapture using all require parameters. The model was wrong (see attach photos). Can you help us solve the problem? Or give some recomendation? 

Photo 1 with correct side

Photo 2 with not correct side