thanks for using our forum to put up questions.
We've also discussed the same question with our developers, so far from our mapping device natively it's not easy to have a structured pointcloud, as the scan position is constantly varying. Our current software is also not supporting the conversion.
But I've also emphasized your request with our developers again, it's good to know on how universal and the impact of the use cases, we'd also provide the information to our devs and see if it could be developed into a new feature.
Hi Ben & Peter,
Great to know that there're more and more scenarios that you're interested and that VLX has the potential reach to all these difficult situations.
Just found that I missed the last question from Ben, about clearing noisy pointclouds. From IVION we can do such croping of desired range of pointclouds, but only to export them:
To clean desired area of pointclouds and present it as default in IVION is still not available, a workaround is to clean the pointcloud.ply in 3rd party software before web-processing, and after web-processing it would be a cleaned version. (Of course this local step needs desktop processing SW and not our cloud processing.)
sorry to hear that it failed repeatedly, I also saw your team put up a support case. Let's move further discussions there, one of my colleague is working on it.
We're putting this issue to a higher priority since it happens repeatedly, currently we're searching for cause and solution with our development team and will update you in the support case.
Regarding the error code, sorry currently we don't have a public reference for the codes. In this case it's coming from other services like AWS with AWS specific descriptions. But I think it makes sense that we summarise and draft a user friendly error codes reference, with some solution steps so users can react on those errors. Will bring this up within our team on this general topic.
thanks for using the forum. To your problem, this looks like an error thrown from our cloud infrastructure from Amazon Web Services.
For some reference, here's a page from AWS on describing the spot instance interruption: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/spot-interruptions.html
For your case, could you try to restart the processing once again and see if the processing can finish normally?
If it doesn't, please create a support case via our portal or by sending one email to firstname.lastname@example.org.
Usually our forum is used for discussion of broader topics that is visible to the public audience, and for issue that requires logs, data etc. we switch to the support case where we could handle faster and easier.
thanks for the screenshot and description, I think I understood better of your scenario.
My previous reply was imagining you were on a boat and floating up & down under a wharf but without moving around much -> boat floor could be scanned well but the "ceiling" would be blurry.
From your last mail I guess you're on a small boat but moving quite a lot under the wharf, if the main structure you're capturing is the wharf, then actually what we can imagine is the wharf is static (which is true) and the VLX/mapper/boat is moving.
In an ideal case you'll get a relatively good scanning of the wharf structure, but the boat captured by the VLX will be the noise and will be like a trailing pointclouds in the final result. Similar to e.g. if you're stretching your hands during an indoor scanning you'll get a trailing pointclouds of your hand.
Sounds like a quite challenging environment and let me know if I misunderstood anything of your scenario.
thanks for using our forum to put up questions :)
I'm not too sure about the exact structure in your scenario, but yes, if in the environment, where the floor is static and the ceiling is "moving", then the SLAM and the accurate capturing of the structure will be suffered.
Could be that the scanning can still be finished and the result doesn't have a broken structure, but I guess the "moving" structures like the ceiling might be cloudy or inaccurate. It's similar to e.g. mapping a corridor with a constantly moving door, in the end the could be the corridor is still correctly scanned but the door is not a trustworthy object.
Hope it helped a bit,
thanks for the question, this sounds like a normal usage support issue.
I'm opening a support case for you and will reply you there.
thanks for reaching out to us.
If you didn't see obvious errors popped up during mapping, and also no obvious error in the log, then yes I think you should be able to process the data correctly.
Still could you send over the 2021-04-19_NavVis.log to us? I can forward the behaviour to our devs to make further improvements.
What we've observed is, if the VLX mapping gets very large / long, the qualitymap generation, which is at the end of the finalisation stage might fail, due to resource bottleneck, mainly on the RAMs. However, the essential data for processing are already saved prior to qualitymap generation during mapping.
Of course, if the processing failed for any reason, let us know we need a separate support case to investigate further.
thanks for the request, I think it makes sense to have such feature in the UI.
Just for now a workaround that can reach the desired result:
- after the mapping, jump out of the mappnig UI first
- Open Folders, then open datasets_rec/<Your_Last_Mapping>/dataset.json file
- Manually change the "name" field in the second line
I'm also attaching a screenshot that describes the above steps, hope that's clearer.
On M6 I think it's easier to do, on VLX could be harder because of the smaller screen.
In the meantime I'll provide this feedback to our devs on the feature.
thanks for putting up the question!
A short anwser to your question is no. It is not recommended to change the size of your adjustment plate during your mapping.
Especially when you have already taken some control points and already selected the size of your plate, in this scenario changing the plate size will corrupt the processing accuracy.
If you haven't taken any control points, then theoretically you can still change the size of the plate, and set the size on your first control point recording. But again, it's always recommended to have a set size in the beginning and not changing during mapping.