I’m trying to build a detailed model of a local landmark from photos taken with a Pentax DSLR, a GOPRE Hero 4 and a DJI FC350 camera.
There are reasons for using each one.
What’s the best way to align the images? All in one big lump, or make a separate mesh for each camera and export as a compenent then align the componenets…or…other?
Hi Nigel,
For the case the link didn’t answer your question:
RC can deal very well with different cameras in the same project, so there is no need for separating them…
Currently not my experience…it’s great at aligning my drone camera, my GoPro and my Pentax, but it’s creating three separate components and I’m having to spend a lot of time on Control Points…in RC’s defence it’s a very complex model and I’m trying to capture a ridiculous level of detail, I just wondered if there was one way that works better than others.
This might not be due to the different cameras or RCs way of dealing with them though. If the difference in angle from the drone images to the ground shots is too much, it will be difficult for any photogrammetry software. If it’s possible, you could upload some screenshots with the camera standpoints of all components (from the same viewpoint in RC). Then it’s a bit easier to evaluate the situation…
I’ll try later. I managed to stitch together two of the larger chunks so I’m doing a high detail build now, just to see if it’s delivering what I need and if it’s stitched accurately where I smothered it in control points…!
Good luck with that! 
To check the general quality of alignment a normal reconstruction is usually sufficient.
From my own experience I can say that it is quite hard to get a difficult model under control with many control points…
My, you’re a little ray of sunshine aren’t you Götz?
I’m worried that it LOOKS like it’s properly merged but might be slightly off on closer inspection…hence detailed…
if that’s the case what would you recommend; switching to using one camera, just taking more pictures to “fill in the blanks” - or would using some actual control point markers at the site help?
But…I started with four groups of about 400 shots each (more on the GoPro cos I just left it on 1 pic per second while I was walking around, but most were useless so heavily filtered before import.
my main group now has over a thousand…with one group of another four hundred or so still to “persuade” into place with more control points…
Sunshine? Sarcasm? 
If the different shots are too far apart from each other RC will have issues.
So my suggestion for you is to take some inbetween shots. From way up in the air down to the ground so that there is a transition.
It helps RC to connect the dots
So you know, this is the level of detail I’m trying to achieve…I’m quite happy with the surface definition in this image…I got “fooled” by the texture in my first one - although it looked quite detailed, it was all in the images and so had zero depth…which led to a very featureless 3D printed version.