How reliable is armory right now


Hi, i am working on my thesis, i was working on an development and testing oriented to robotics on bge but i feel limited by the single core usage of the bge and the visuals limitations, i have 1 month to decide wich game engine i will use and i dont want to have problems in half road, how finished is armory right now?, does the transformation and rotation matrix are working here?, i need to do a very realistic approach on the programation of a robot, thats why i need to use transformation and orientation matrix, castray and sensors, you think i will be able to do stuff like this here, this means that i will have 1 month to learn haxe, i only know python and some assembler


Armory is fine as a game engine, some rough edges but it is manageable. Though in your case when you are relying on good transformation I don’t think Armory is the right choice. Rotations are total madness imo. For example when you rotate something along the Z axis only it will first go up pi until the first half then the x or y rotation value gets set to pi while the z value goes down again. Once you are at the 3/4 waypoint sometimes one of them randomly jumps up to 2pi. So already making a basic script for one object to look at another which is moving trough space is hellishly complicated.


But is still possible?, if i take in account that weird transformation i can adjust the math to respond to those curves i guess


Right now, for me, freezing is an issue. I would test your setup by mocking up a simple scene with simple rotations of simple objects and let it run for 24 hours. I only managed to get a project to run for 24 hours with a very simple cube, one logic node tree, one light, and no shaders. Adding shaders and game logic cuts down the non-freeze time. Once I got freezes due to a single object indistinguishable from the other objects in the scene that I only found through systematic removal and retesting. I ended up spending days changing settings, removing shaders, logic, and objects, and retesting for freezes. You may have to do freeze tests every time you add a new “aspect” to your project if you intend it to run consistently for long periods. This can be a whopper of time consumption, as my longest wait for freeze time was just over two hours before locking up. Right now I get between one hour (interacting with project) and two hours (letting it sit untouched) before freezes.

That being said, if you don’t get any lockups (which may be just a fluke of my setup), Armory rocks, assuming it has what else you need. I can’t do what you do and am speculating by personal experience.

For rotation madness there is quaternion math, which is a madness of its own.


@LostInRotation can you please drop an issue about it at