So we're trying to develop an app which shows you a 3d model based on the scaned floor plan (marker).
The whole thing is supposed to be modular. A develop once maintain while never touching the code again kind of deal.
We achieve this by having the models on a seperate server while managing the marker through the wikitude studio and using the wikitude "cloud recognition - single recognition" to display the models.
When the app scan the marker it gets the corisponding models name through the metadata, downloads it from the server, imports it during runtime and assigns the CloudTracker -> Trackable as a parent.
The Problem starts here. The models don't match the marker size. I'm testing at the moment with two different models/markers and while one needs a scaling of 0.065 the other needs a bigger one.
It gets more puzzling if I use both models on the same marker as they appear correctly scaled in correlation to one another.
The issue is that the models have about the same scale in the 3D software about 9m heigth while the marker are about 9cm high, In other words they have a 1:100 scale. So my conclusion would be that I'd need a scale of 0.01 in Unity but as I said I need a scaling of 0.065 for the one model and (i think it was about) 0.076 on the other one so they fit their markers respectively.
The target size in Unity is always 1m, regardless of how big the image actually is. We're working on introducing a new workflow what will behave as you're expecting, but for now you'll have to work around the limitation. One thing you can do is specify the scaling factor in the metadata of the cloud target, and then use that in Unity to scale the augmentation appropriately.
I'm not sure I fully understand the issue, but the size of the augmentations scales with the size of the marker. So in Unity, if you create a cube of size 1, it will span the entire width or height of the target, depending on which one is larger, regardless of how the images are then printed. So depending on how your models are initially sized, it might be that different scaling factors are required for each model.