Non native models tend to be heavier

Hi all,
This is a question about imported models:

Most of the time, I create models in sketchup and export them to a renderer/slicer/etc. but from time to time, in order to save time, I attempt to import models (mainly obj. or .fbx) from the internet.
Almost every time that happens, the imported model is far heavier than its polycount would suggest.

As an example, I am uploading a minaret I downloaded from thingiverse as an stl.
It had half a million polygons so I used transmoutr to reduce its polycount to around 150k polys but still it is very heavy. I decided to recreate it within sketchup and as you see they have roughly the same polycount but behave very differently:
The one created within SU is light as a feather while the imported is almost too heavy to handle.
(I have a decent system, of 64Gb ram /1080 GTX so the effect is not so noticeable but if I make a few copies it really starts to show)

From my experience, the usual suspects that make an imported model heavy are these:

  1. sometimes while importing, a single texture ‘breaks up’ into as many textures as the polygons of the model.
  2. other times, depending on the units of the imported model, each polygon is not attached to its neighboring ones, so the model looks the same but you have a very big edge count.

In this particular example, neither issue applies, so I am buffled as to why this happens. And yes, the imported model has a slightly bigger polycount (170k vs130k) but even if I have two of the native model it is still more responsive than the imported one.
The reason I am asking is to understand better how things work so that I can improve my workflow even with models I create within sketchup.

P.S. I am attaching a link to the file with the two minarets (26Mb), and I have 3 scenes:
“created in sketchup”,“imported” and “both” so that you can notice the effect. (if you don’t notice it you can multiply them a few times until your system starts halting)
https://we.tl/t-LZHSjSa6h1

By handling the model I assume you mean rotate / zoom / pan.

First thing that comes to mind is (in your uploaded model example) that the native object consists of many groups/components. The other has not.
SketchUp displaces large amounts of grouped objects as simplified bloks (bounding boxes) with these aforementioned operations. While raw geometry, each and every single edge etc. has to be taken care of in the second (imported) object while rotating / zooming / panning.

This may be (part of) the answer to your question.

yes, I am talking about the navigation (but even copying the object takes a lot of time)

I forgot to mention that I tried that as well:
I exploded all groups/components in native, still performed better. Then I randomly grouped parts of the imported (though not near to the number of the native-it would take a lot of time) and still the same.

Texture (and their positions orientations) on edges that have to be taken into account during operations?
Exploding groups and components will have tags and materials being “dumped” on default geometry inside.
Just thinking of some kind of explanation?

Is the model purged?

yes, the model is purged (components and materials) but I don’t think that has anything to do with the performance