Reproducing the same result

[Ainur]

Dear Agisoft team,

does the Photos Alignment process depend on a hardware properties?
I am wondering if images are processed on different computers, but the same parameters, would the result be different.

Thanks and regards,
Ainur

[Alexey Pasumansky]

Hello Ainur,

The alignment results may be different due to stochastic values are employed. However, if there are no problems with the overlap and coverage the results of the camera alignment shouldn't differ much between several tries.

The difference may be considerable for unstable alignment (lack of texture pattern, low overlap, low accuracy or low image quality).

[Ainur]

Hello Alexey,

many thanks for your quick response.

Now it is clear for me, in my case, probably the reason is a low image quality.

BR,
Ainur

[robeastham]

Extending this slightly, I noticed that I get very poor/failing alignment the first time I align when using images that have lost their EXIF data.  But if I run alignment a second time on the poorly aligned chunk I seem to get good alignment. I've done this about five times with the same dataset.

If I transplant the EXIF from my CR2s thing are much better. I get much closer and consistent results, with alignment working the first time and the number of points being very close.

It would be good to understand what is going on in the first case that requires two align steps. Is this just random luck that I got alignment the on second align or is the first align data used in some way to help in the second align when using images that have their EXIF data missing.