With a laser (BLK360) and photo dataset, reconstruction of some parts of the model fails if the minimal sampling distance (under mesh calculation) is set to the default value (0).
Further description of the problem here:
Dear Cabral,
please excuse our late response, we do our best to reply as soon as we can,
I have not been able to reproduce the problem you are reporting with our data. Would it be possible to provide us with the dataset for testing?
You can use our contact form to send it to us privately. Please refer to this forum topic (the exact forum link). Thank you.
I’m not totally surprised.
I’ve made projects work in the past, that kept on failing during high reconstruction. by adjusting this to a larger number.
it may be that only a small part of your scene is causing this issue.
but there arn’t really any decent ways to diagnose it.
you can split the scene into many different regions. and process each of them until one fails.
but it may not be worth spending that amount of time trying to figure it out.
Hi chris,
That’s the behavior I’m experiencing. Some parts of the model work with the default values (0) but just by moving the region a little to the side the reconstruction fails unless the default is changed.
In any case, I’ll be sending the dataset for testing.
yeah sometime parts of the alignment just gets these issues.
so you can either lower res, it sometime works ok on normal, but fails on high. (or use the min sample distance)
try and realign it.
or just avoid the area in the reconstruction.
it would be great in rc handle this better. even just by skipping the problem area. rather than just getting stuck with a memory leak causing massive ram usage and very little cpu.
sometimes it can be left and it can just finish at some point. but often it just goes nowhere or crashes.