Loading Details - Blend4Web API

We're having issues with loading our project on OS devices, one is an iPad and the other is an iPhone 5 both in safari browser. Each reaches 40% when loading the file, then the webpage crashes and reloads. Is there a way I can trace out the loading stack and see what is being loaded at that time? The API guide only shows that I can get the loading percentage, but I need the details.

To add to this:

I have found that at 40% loading on a Mac desktop that when I show the console window a B4W Warning Appears, this warning is "B4W warning: depth-only issue was found". Which when I searched into the base code the framebuffer isn't supported. Is this because a setting in my Blender file or is it because the browser is not supporting Blend4Web? I have a feeling this is the reason why my webpage crashes on OS mobile devices and reloads.

Looks like an out-of-memory issue. It is not related to the warning you mentioned. It can happen on mobile devices than you got too many polygons or too big textures in your scene. May be some hidden high-poly geometry presents in your blend file that is still being loaded. I'd recommend to open your exported JSON file in our Scene Viewer in order to check actual polycount and memory consumption. You can also email us your model if you are not restricted in doing this.

If this is the case, the model will have to be reduced. Currently it's approximately 500k worth of polys. Again this high poly issue stems from the face that the models were not created by us, and we have little time to edit the models. I will have to let management know that more time will be needed to reduce and clean up the models before we can use them in most browsers. Currently it is only happening in older Apple devices. We're going look into determining a device's memory before determining the load. If it's too low we will prompt the user.

Yes, 500k polys is too much. Not only this eats system resources but also is a reason for slow loading, and therefore bad experience for end users. I agree that investing more effort in model optimization is a good idea.