Issues viewing 3D model on Hololens 2

[Tony]

We loaded model into the Manifest portal and it imported without issues. When attempting to view the model on the Hololens 2 (less than a year old) as the model loaded, the 3D Manifest application would blink throughout the loading process. Once the model was loaded, we attempted to take a photograph. The application then exited abruptly. We tried this numerous times. Assistance would be appreciated with any ideas.

 

Thanks in advance

Tony

[Maureen]

Hi Tony,

It sounds like the 3D Model is impacting the performance of the device which often indicates the model is too large – perhaps there are too many poly’s or the textures are too large. The best way to check is to first refer to the 3D Model specifications in this article.

https://experts.taqtile.com/knowledgebase/using-3d-models-in-manifest/

If you do not see any obvious issues with your model based on the recommended specifications – please reach out to manifestsupport@taqtile.com and we can troubleshoot with you directly.

 

– Maureen

[Tony]

Maureen,

Thank you for the reply. We went through the 3D requriements and are confident that we met them the model had 47,237 vertices and that was apparently just too much. We uploaded a smaller model and had no issues with it loading other than the entire Manifest 3D application blinked on and off as the model loaded.

So, we think for now we are Ok and it was probably that the model was just too much for the hololens. In another effort previously we had used a Magic Leap for the same model and it loaded fine, but for this effort we weren’t able to get the ML to test with.

Lastly, a nice feature request might be that there is a progress indicator on model loading. On that large model from last night, it took ~5.5 minutes to load and with the dialogs blinking it wasn’t clear that the model was actually rendering.

 

Thanks again

[Maureen]

Tony,

With large 3D Models, it is not uncommon to experiencing some blinking in the application during the loading process. Unfortunately there are some operations that hitch the device, such as binding textures and creating the actual mesh objects (and more importantly, the colliders applied to the mesh). In addition, the library we use for loading FBX and GLTF/GLB models does some unthreaded operations which also can hitch the device temporarily during loading. We are constantly working to optimize and improve this, but it is not surprising to hear of that experience.

In terms of the model, 47k verts should not be an issue. We have loaded much much more complex models than that without an issue. The issue with having models more than 50k or 60k vertices is that they can cause performance degradation upon viewing (having a frame rate lower than 30, although the device is actually targeting 60 fps) which can cause the UI and model to “swim” as the device can not update it’s rendering to match the user’s position in the world.

If the application is not fully crashing you are interested in having us diagnose further – we always benefit from seeing the Logs. If the Manifest Session hasn’t been restarted – Go to Settings -> Bug Report and enter a brief title / description and submit. This will grab your system logs and help us diagnose the issue further. If the application does close from a crash, unfortunately the logs will be wiped when you restart the application. This is something that we are looking into adding solutions for in an upcoming release.

I’m surprised a 3D Model was taking ~ 5.5 min to load and wonder what might be going on. Please feel free to share the model file with us or share access to your domain if you would like us to take a closer look and advise.

As far as a loading indicator, I believe newer versions of the model loader that we use may expose that information and we agree that is a valuable indicator to the user, so we can look at adding a number indicator in a future release.

 

Maureen

[Tony]

Maureen,

Thank you for following up on this. I moved beyond it by having our modeler trim down the model just to the part I needed for the graphics I am working on. Once we get our heads above water I’ll load up the complete model and see if it happens again and send the bug report. It was exiting pretty consistently, though, as we are very new to the platform it may have been something in our model. I would be greatly interested in finding out. Once I do that, will it automatically open a bug ticket and then after I can share our model with you?

Thanks again

[Maureen]

Perfect! We are definitely curious what could be happening there, but happy to hear you’ve been able to move past that issue with the adjustments made. Keep us posted and yes – bug report and model file will be enough for us to investigate. Thanks!

[Author]

Posts