It’s not a game related topic but more about industry design.
A friend of mine are searching for a solution to create an interactive 3d demo of a mechanical system for certain industry. The project can be huge including thousands of 3d animating mechanical parts (not necessary in the same scene). The demo will be used to design and promote the system, so the models and animations are aiming for industry standard.
Here’s my question:
Is unity capable of making an industry standard 3d interactive mechanical system demo? If so is there any problem with licensing?
What does “industry standard” mean? CAD models? If so, no, CAD models are not suitable for realtime use, and there is no engine that will work well with them.
The problem isn’t the format, the problem is that CAD models typically result in gazillions of polygons. If you can use models appropriate for real-time use, sure, that’s fine.
Even then there are limitations due to the precision of physics and the potential problems a more complex mechanical system brings up where you need to create code to bridge different components and aspects to create new ones like chains that transfer a torque.
It can work very well but it can also fail, so you optimally evaluate it with the free version (Pro does not offer more physics, so no difference there) in a “common usage” scenario to see if it can work in the type of situation you require it to work.
License wise I don’t see any possibility for “being a problem” actually as you own what you create with unity.
You should take a look at Algor, it is more for what you describe. The fact is Unity isn’t really up for the sort of thing. I certainly wouldn’t consider for mechanical prototyping.
When you export data from a CAD program you have many options for file format and polycount. That said, I am sure OP knows his way around the CAD software. Yes, you can use Unity for this, I think it’s a good choice as you can deploy on many platforms, setup would be quick I’d export the entire scene animated (looped of course) from 3DS Max/Maya.
A lot depends on who is providing the CAD model. IMO if it’s coming from off site (ie. a client) it’s likely that they know CAD but haven’t the foggiest what the real-time developer needs.
Irregardless of how much time and effort it takes to get the CAD model into Unity, Dreamora’s right, the one issue you can’t work around (at least not without a huge man hour investment) is the limitations of Unity’s implementation of PhysX. Yes, you can code your own physics behavior (been there, done that) but it’s not a project for the faint of heart.
I think it’s perfectly possible.
You would have to model it low poly though. You might be able to rig some of it as animations, either in Unity or other software, and the programming in Unity could complex depending on the level of physics required.
It would be an interesting project.
Thanks a lot all of you! Actually I’m now more clear about my question after reading all your inputs
I think Lamont’s guess about polycount is more or less the case: the models will be made or exported from CAD package, but the focus of the demo is input and output (animations). As long as the dimensions are correct, polycount can be optimized to fit the budget.
Physics is a big issue. The simulation like SolidWorks provides looks too much for game engine and developers. But after showing my friend Solidworks, he thinks the design work can be shifted to other package. When it comes to game engine, the demo can focus more on interactive stuff (think about some management system). I think the project can tolerate fully predefined animation at the beginning.
Another issue IMO is the training required by different softwares. My friend and his team can model and make mechanical animation but know nothing about any game engine. My best guess would be finding a couple of game developers for his team so they can work closely together.
Let’s assume it can be done without realistic Physics, how about pre-defined animations? Can CAD packages export animation in fbx format?
We did something similar last year; a mechanical model of a system, but it was used more for logic verification/testing of PLC parts and we used simplified models that were nowhere near physically accurate, and we had much less than 100 moving parts per scene.
It was difficult to get the precision for simulating the physics well; we had to make logical equivalents for clamps and conveyors. It worked well enough for our purposes, but I have a feeling it won’t work well for yours considering you’re looking at a thousand parts.
