That probably relates to the fact that I work for an official Partner of SketchUp. But they don’t own us, for that matter, we also sell Modo and the whole Altair suite besides different Renders like Twinmotion (Epic!), V-Ray and so on.
But I originally was a building contracter and when asking why certain software did not take full advantage of my new state-of-the-art computer they tend to compare it with the building process, you can’t start on the second floor if the 1st floor walls aren’t there. Prefab would speed up some thing in the building process itself, but takes more time to prepare and logistics. This time can be spread, just hire a few more subcontractors and they could all be working on all the elements the same time, but putting things together relies on a sequential order.
Another thing is, if you ask two contractors about how they work and why one would be better than the other, you might choose contractor A to build your house and maybe even hire contractor B for parts of it, but when something goes wrong and ask contractor C for an expertise, you wonder how you even considered hiring A and B in the first place!
Parallel computing sometimes actually slows down the process, imagine you would go to work with two cars instead of one, you drive car A a few hundred yards, walk back to drive car B a few yards ahead, etc.
Sometimes when asked by a client when a job was suppose to be finished I would joke about starting with the skirting , first, because that’s usually done at the end (Client: ‘But there are no walls to put them on?!?’ reply: ‘But the wall contractor does know where to put them, though, between those skirts!’)
On the subject of reanimating, I agree that there are parts of the rendering that could be improved, especially in relation with the inference system. Check this thread for instance:
This is exactly where SketchUp could improve, it treats all the leave components as if it were constructible elements in a building design and so the database in Outliner get’s bloated, while if treated as an simple (or more complex) asset that just needs to get visualised while orbiting, it would only use the multithreaded GPU instead of the serial computed data in the CPU.
Or, like in the web-based 3D viewer of Trimble Connect, provide a way to decrease the load of info while zooming in on an object (like in a gaming engine, the level of detail is dependant on the distance of the object)
That’s why it’s easy to load multiple models inside the viewer. But when a model is altered, it needs to reload.
I’m glad I was able to direct this thread back to Trimble connect 