Skip to main content
B
BenAtLmco13-Aquamarine
13-Aquamarine
July 15, 2021
Solved

Automated Model Target Generation

  • July 15, 2021
  • 1 reply
  • 2420 views

I was curious if it's possible to semi-automate the generation of model targets.  I have a lot of different models that may have some small differences, but are largely the same.  If I manually set the model target for a single model, is there any way to retrieve the camera position/orientation from the model target and then automatically generate model targets by providing another model and the desired camera position/orientation? 

Best answer by Suraj_Patil

Actually whatever you mentioned above is really wonderful idea to save a developers time. But I really don't know is it possible to achieve in todays scenario. May be PTC guys can help in this.

Temporary solution I would like to suggest is, We can use two same models in Vuforia Studio overlapping on each other. One model is having all sequences/Illustrations and second one is purely for model target generation and its in hide condition (visibility OFF).

Now, yes whatever I m suggesting is manipulation but it will really fulfill your requirement . Only thing you need to do is model target generation model should be generalized one , means like we can provide outer structure of model for target generation(which we know that it will not change in revision )

lets take example of car

now if I know overall body of car is not going change but its features like mirrors door shape n all may change so ,what we can do is ill take two models in vuforia studio one is having all detailed feature and another one is for target having no feature(outer shape is going to be used for model Target generation). once model target generation is done in Vuforia studio, ill import actual model overlapping on model target generated model. and ill hide the model targeted model.

I think so this will work for in your case too. please share the feedback on this 

 

Thanks in advance

1 reply

S
Suraj_Patil16-Pearl
16-Pearl
July 27, 2021

what you really want to achieve by doing this?. what's the idea behind it ? Can you please elaborate it?

 

B
BenAtLmco13-AquamarineAuthor
13-Aquamarine
July 27, 2021

Hey Suraj!  This question is motivated by the fact that the projects where we're using AR involve many different revisions of the same assembly and the same assembly at multiple points in the manufacturing process.  Regardless of exactly which revision or step in the manufacturing process, the function of the AR experience is more or less the same with the only changes being the model target.  Because generating the model target is a manual process, the time needed to generate new model targets can add up.

Let's say that a new revision of a part in my assembly has some slightly different hole placements that could affect my tracking.  The overall structure of the assembly hasn't really changed, so I'd like to just regenerate the model target from the same perspective as the previous revision but with the newer assembly with the updated part.

 

Or let's say that I want to examine the same assembly after some brackets or clamps have been installed.  Again, the overall structure of the assembly hasn't really changed, but I have added some components that I would like to update my tracking model with. 

 

Given the potential scope where AR could be used in this particular application, I would like to be able to automate the generation of potentially hundreds to thousands of different targets that only have some small minor changes between them.  This would save a lot of developer time and make sure that whoever was using the experience was leveraging the most accurate model for tracking available.

 

The ideal end state would be writing a script which would generate a model based on a list of part numbers, procedurally generate an experience from a given template, update the model target automatically based on a specified model target generation camera position/orientation, and then publish the experience automatically. 

 

Obviously most of that process would take place outside of Vuforia Studio, but a critical piece would be the ability to procedurally generate model targets. This isn't necessarily something we're planning for at the moment, but I was curious whether this could in theory be feasible.  Thanks!

    S
    Suraj_Patil16-PearlAnswer
    16-Pearl
    July 28, 2021

    Actually whatever you mentioned above is really wonderful idea to save a developers time. But I really don't know is it possible to achieve in todays scenario. May be PTC guys can help in this.

    Temporary solution I would like to suggest is, We can use two same models in Vuforia Studio overlapping on each other. One model is having all sequences/Illustrations and second one is purely for model target generation and its in hide condition (visibility OFF).

    Now, yes whatever I m suggesting is manipulation but it will really fulfill your requirement . Only thing you need to do is model target generation model should be generalized one , means like we can provide outer structure of model for target generation(which we know that it will not change in revision )

    lets take example of car

    now if I know overall body of car is not going change but its features like mirrors door shape n all may change so ,what we can do is ill take two models in vuforia studio one is having all detailed feature and another one is for target having no feature(outer shape is going to be used for model Target generation). once model target generation is done in Vuforia studio, ill import actual model overlapping on model target generated model. and ill hide the model targeted model.

    I think so this will work for in your case too. please share the feedback on this 

     

    Thanks in advance

    Terms & ConditionsCookie settingsAccessibility statement