Object jumping around mesh, not latching - VPS

Include the following details (edit as applicable):

  • Issue category: VPS
  • Device type & OS version: iOS
  • Host machine & OS version: Mac
  • Issue Environment : On Device / Dev Portal
  • Xcode version:
  • ARDK version:
  • Unity version:

Description of the issue:
We have worked out how to place content relative to the mesh of the locations but when we set about triggering the content we are finding that we get a sort of ‘drift’ with the object where it will appear (although not reliably) and then it will seem to jump around between 2 or 3 locations in nearby space. The interesting thing is that the jumping seems to move between consistent different spots so it almost feels like VPS is finding different latch points periodically and jumping between them.

So I am trying to track down where the issue is coming from.

  1. Is it because the scans of the locations are not good enough? I don’t think it is this as from what I can see, some of the approved locations we are trying have nicely complete scans.

  2. Is it something to do with the GPS?

  3. Is it proximity or height? Are we scanning from a bad position?

  4. Is it something completely different that I haven’t thought about?

We are just struggling to track it down as it’s not really a code issue that we can debug but it seems something more contextual.

I have also attached a video. In this case we were using location “new-street-p” but we have tried a few and are finding similar behaviours at each spot.

Hey @Marcus_Dyer ,
alright lets go one by one:

  1. This could be a reason, specially for the wayspot you’ve shown on the sample. But if this is happening in multiple stable wayspot locations, then that makes this less likely.
  2. No, unless your code somehow introduces gps, VPS doesn’t use GPS.
  3. Proximity does affect localization and anchor placement, but this shouldn’t be a problem unless you are more than 20 meters away… From your sample I doubt this is the case (though I don’t really know where you are placing your anchor, it could be that the snowman is loaded far away or something silly).
  4. So I got 2 theories, the first one is realted to number 1 in that the wayspot location is not actually very stable (specially if the ones you’ve used are all private scans). The second might have to do more with the placement of the anchor, or if you are trying to move it or the object associated to it in some way.

All this is kind of hard to debug without taking a close look at the code and I don’t think I can go super deep with that so, a few questions.

How are you placing the anchor? (I think this has the highest chance of being the root cause, so the more detail the better)
How are you loading the snowman?
Do you, at any point, update the anchor or the snowman’s position in your code?
And finally, can you tell me the address of the wayspot you are using? Makes it easier for me to look for and have a better idea of the scan.

Gilberto

Thank you so much for coming back to us with so much detail, that is enormously kind.

So regarding the code we haven’t really done anything bespoke yet.

We are just using the demo template from here:

https://www.8thwall.com/8thwall/vps-bespoke

I haven’t added any bespoke code beyond changing the location and importing the gab of the new location.

The location I have been testing with (that you can see in the video) is:

new-street-p

This is not actually the final location, just close to where we work so ultimately I will change this but I am finding the jumping that I am seeing with most of the locations we try (we have seen the same with 3 or 4 different places). What is weird is that some of the scans look quite good. Maybe you will have a view on that and it may be that we still need better scans but those did look fairly good to me.

The ultimate destination I am going to use is:

welcome-to-c

But again I am seeing a similar jumping problem here.

More specifics here….

How are you placing the anchor? (I think this has the highest chance of being the root cause, so the more detail the better)

I have done the following.

  1. I have included the location glb.

  2. I have put the new location name in the code.

  3. I have included the GLB in the project.

  4. and also added it to the list in the project browser.

(Attached image shows details).

Other than that I have not adjusted the project from the template so I think it is quite ‘vanilla’.

The way you have phrased the question is interesting too because it makes me wonder if there is another separate ‘anchoring’ step that I have entirely overlooked? Is that possible?

How are you loading the snowman?

The Snowman is loading fine at the moment. I have just swapped the GLB for the squaloons original…

I think the demo code is doing the loading and visibility toggle. Again maybe I am overlooking a step here but I don’t think so?

Do you, at any point, update the anchor or the snowman’s position in your code?

I have not written anything bespoke so whatever movement is happening is just as a result of the template implementation. I haven’t tried do anything clever yet as I am just working on the basic latching first.

And finally, can you tell me the address of the wayspot you are using? Makes it easier for me to look for and have a better idea of the scan.

For sure. I am testing on

new-street-p

But my final live project will be at

welcome-to-c

Thank you for the help once again.

Marcus

Marcus Dyer
––––––––––––––––––––––––––––––––––––––––––––––––––––––––
FLIPSIDE Impossible. Done.
––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Mobile. +44(0)7956 949 566 | Skype. calyp50 | flipsidegroup.com
135 Bishopsgate, London, EC2M 3AN
––––––––––––––––––––––––––––––––––––––––––––––––––––––––

CONFIDENTIAL ---------------------------------------------------------------------------------------------------------------------------------------------------------
This email and the information it contains is confidential to the intended recipient. If received in error, please inform the sender
then delete it from your system; you should not copy the message or disclose its contents to anyone. Please note that neither
Flipside nor the sender accepts liability for viruses and it is your responsibility to scan the email and attachments (if any).

Hey @Marcus_Dyer ,
Just as a quick update, I have not been able to reproduce this issue. I have tried a number of locations and different objects in them too, so this are just shots in the dark right now but a couple of things I think might give me a bit more info.
Have you tried private scans and do you see the same moving around in them?
I see some of the meshes on those locations (but not all of them) have the warning “unaligned” in them, can you verify that those are not the ones you are using?
And finally, do these happen on the Remote Desktop Development Setup on the browser too or only on your phone build (presently there)?

Thanks again Gilberto

Some comments here….

Just as a quick update, I have not been able to reproduce this issue. I have tried a number of locations and different objects in them too, so this are just shots in the dark right now but a couple of things I think might give me a bit more info.

Thank you for this. This is the bit that is really confusing me at the moment. We are using the vanilla template project and we are trying it on quite a few local points and we are able to replicate the issue fairly faithfully.

So it seem to me that either we are missing something obvious or we are just very unlucky with a bad set of waypoints but given that it is quite consistent I don’t think that is what is going on.

Have you tried private scans and do you see the same moving around in them?

We have just submitted a private scan and are waiting for it to be approved and we will try that too.

I see some of the meshes on those locations (but not all of them) have the warning “unaligned” in them, can you verify that those are not the ones you are using?

Yes. That’s a good point and I have been careful not to select those as I had the same concern that it might have been bad scans but we have been choosing what look to me like quite complete scans.

And finally, do these happen on the Remote Desktop Development Setup on the browser too or only on your phone build (presently there)?

Interesting. I don’t know much about Remote Desktop. These scans are all ones that we have tested in person on a device and we have tried a range of devices and browsers so I don’t think it is device linked.

I hope that’s helpful.

I can collect some more videos too if helpful as I know this is quite challenging to debug / address as it is so contextual.

Thank you again.

Marcus

Marcus Dyer
––––––––––––––––––––––––––––––––––––––––––––––––––––––––
FLIPSIDE Impossible. Done.
––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Mobile. +44(0)7956 949 566 | Skype. calyp50 | flipsidegroup.com
135 Bishopsgate, London, EC2M 3AN
––––––––––––––––––––––––––––––––––––––––––––––––––––––––

CONFIDENTIAL ---------------------------------------------------------------------------------------------------------------------------------------------------------
This email and the information it contains is confidential to the intended recipient. If received in error, please inform the sender
then delete it from your system; you should not copy the message or disclose its contents to anyone. Please note that neither
Flipside nor the sender accepts liability for viruses and it is your responsibility to scan the email and attachments (if any).

Hey @Marcus_Dyer ,
It’s been a while so I’m just checking in on this issue.
I never got an update on if this was happening on the remote desktop development or the private scans, also kind of hoping you found your solution but did you?