Does RC make use of the spirit level of the camera?
I am just noticing that usually vertical alignment (e.g. of different storeys or floor levels in a building) are usually on the spot whereas horizontal alignment is not…
Hi Götz Echtenacher
Does RC make use of the spirit level of the camera?
I am just noticing that usually vertical alignment (e.g. of different storeys or floor levels in a building) are usually on the spot whereas horizontal alignment is not…
?? can you post some screenshots and description of this ?
Yes sure.
I posted this elswhere:
WOW.jpg
As you can see, the floorboards in the building are perfectly parallel, even though the model is quite difficult with the ends not connected. It is not possible, because the flooring is right at the wall and also in the left corner at the roof.
2.5 cm over a distance of almost 5 m seems almost too good to be true.
When I added the next level at the bottom, the floor was again absolutely accurate, but the rafters did not meet.
That means that the separate arms are stretched differently in horizontal direction, but the vertical direction is 100% correct, also the general alignment.
That triggered my thought if the digital spirit level is used - would explain the parallel arms.
If RC knows what’s vertical, it would help a lot.
There is no problem (yet), I’m just curious, so that I can estimate problems, or rather their solutions, better in the future.
Unless you specify the camera positions then no. Everything in photogrammetry is relative unless you specify some fixed camera positions or use CPs to define positions/distances. You could try setting one of the camera with a fixed location/rotation and see how well everything anchors to it. I usually use an L-shaped target with printed CPs to set the orientation of the model.
Hey Ben,
thanks for the reply.
Since my models are large scale, the L wouldn’t fit in the car anymore! 
I use theodolite GCPs for that…
The thing is, that the accuracy in horizontal dimension is far greater than in vertical.
At least in this case.
And there is no way for RC to know that.
In my opinion the difference is greater that the rectangular geometry of the images would explain it (I shot mostly in landscape).
Götz Echtenacher wrote:
the separate arms are stretched differently
That alarms me - suggests that RC stretches things to make them look joined-up? What does that do to reliability of accuracy e.g. in a building survey?
That’s normal in my eyes. Tiny errors add up.
Also, my equipment is not the fanciest.
But most importantly, those are very difficult circumstances, basically one plane of masonry consisting of several “arms” only connected in the middle. If you have a whole building that you can circle completely, then the errors go down significantly automatically.
Once I followed Vladlens workflow outlined here How to optimize Reconstruction (use lens grouping by exif), the result was almost perfect.
I used something like 30 odd GCPs obtained by theodolite and the worst deviation was 3 (in words: three !) cm in the bottom right corner. That is very well within acceptable tolerances already. But with another alignment with the GCPs present, the error went down to a few mm.
Satisfied? 
Provisionally!
thanks