Pointcloud merging problem

[robert langh]

Hi,

I would like to combine pointclouds with different accuracy, and merge them together, (like focusing on areas)
but Im not sure with the settings... Here are two different sized pointclouds, with the same settings, which appears to be different:

https://www.youtube.com/watch?v=42-tnpdzCWY

Can You please tell me, what may be the problem?

Thanks, Robert

[Andrea Alessi]

Hi Robert,

you can merge any two objects normally, but if the two point clouds are coming from the same workspace, it's generally better to re-run the reconstruction with different parameters. If instead these are coming from a laser scan for example, I'd manually cut only the parts of interest if possible.

I'm not sure if this answer your question, were you more interested in a reply like above, or were you looking for directions in the software on how to do that? Thank you!

[robert langh]

Hi Andrea!

Thanks,
my setup is a laptop with not enough memory for big pointclouds, so I try to find a way,
to reduce the number of points somehow... Is it possible to keep the overall quality of surfaces
with these kind of settings?
I tried to change the first few parameters, but not as fine as the default...
Then I will hopefully insert the high res parts to the big, low res space, with merge (high res priority).
Which parameters reduce the number of points without loosing too much details, like sharp corners, flat surfaces etc.?

https://www.youtube.com/watch?v=KZOb7rrLqrE

thanks forwards, robert

[Andrea Alessi]

You can try tuning the resolution to match your memory footprint requirement. It will be a tradeoff also with quality, like with all settings, but if memory is a constraint, that's where you want to work. That and in the meshing phase (if required) also on the photoconsistency step.

[robert langh]

... and yes, my first question was that, why the big model quality in the first video was better than the partial modell,
although the setting are the same?

bests, r

[robert langh]

So the depthmapSpeedUpOnLastLevel setting to higher is reducing the number of points, right?
Should I set the image res lower instead?

[Andrea Alessi]

Hard to tell about any difference in result without taking a proper look at the dataset and project, sorry!

Yes, I'd probably simply start from the default preset and if that's still too much for your system, try lowering the resolution, unless they're already very low-res (which seems to be from the video, but I might be wrong.).

Ultimately, you might want to increase the hardware resources as unfortunatley there's only so much you can do to maintain quality by lowering parameters. If your dataset requires a certain amount of ram, either entirely remove photos or increase the available ram.

[robert langh]

Thanks, I understood this,
In the university we are focusing on low budget technology, like mobile phones, etc
and try to increase productivity with the number of photos
I find the iPhone wide angle very good!

thanks, robert