To our current situation. We work on finding a 3D-library in which we can load multiple .gltf or .glb files into. All those files together create one big model.
This is how we do it. First we get a json from the backend in which we can see how many objects are coming, their guid, their position and rotation and on which object they reffer to. After that we have a stream of objects which we will recieve from the backend. Based on the json we know how often we need to clone each object. Now we need to load with all this information the model into the scene.
We have already tried many different ways of loading.
With the AssetContainer we where able to collect all Elements and load them in one command. But inside the container each file was loaded single and took over 3 minutes to load all.
With the Task Loading we have the same problem as with the AssetContainer. Collecting all Elements in single Tasks and loading them in one go. But also here the loading process was the same.
This is our current solution and also the best one so far. We load all the objects async and after that clone it and position it as many times as we need it. It currently takes about 30-40 seconds which we look to improve.
Now loading the model as one file is no option. The only way we could do this if we would be able to merge all the files or meshes together as one and then load it. When we load the model as one file it will take about 3-4 seconds. Loading it in our current solution (LoadAysnc and Clone) is about 30-40 seconds.
What are our options to make the loading process faster?