I'm currently going to school for programming, and my 3d graphics teacher absolutely hates unity. He complains about how he can always tell when a game was made with unity because the load times will be huge for small things and he will think to himself, "That shouldn't have taken that long." Another complaint of his is that it does everything for you and thus leads to incredibly generic games. I rather like unity, as it allows me to wip up a game fairly quickly without having to worry about first making a framework, but I must admit hearing his hate for it does spark a bit of fear in me, in the long run if my productions do get bigger, despite unity's ease of use, will it harm my product in the end, or is that entirely dependent on the user? If my fundamentals of coding are strong enough does it's easy workflow still come with a price or is it just as efficient as any other engine or is my teachers hate for unity justified?
Is hatred for unity justified?
I must say, Unity really has a good design. Saying Unity is slow is for most non-bad designed games ****shit. C# is indeed slower than raw C(++), but using JIT and such, some things can even be faster. Telling if a game is made using Unity or not is mostly impossible... C# is however generally a bit slower than C(++), but your development times shrink a lot by using Unity. If you are not going to create some game that needs to push the system to its limits, you probably won't have any problems. I suspect your teacher is from an older generation (pre-DOS)? I come across more people that seem to hate new technologies without good reason.
He does have a point about the initial loading time, even for small scenes. A lot of companies create their own scene exporters in order to create their own loading pipeline.
The type of games made in Unity is often a simple physics based platformer. This is because Unity is an amazing tool for creating that kind of games, but no. Just because Unity is inviting (and to some extend does all the work for you) to make that kind of games it doesn't mean you have to create the game like that. A "mistake" a lot of people do is to make their games too physics based, because it is so easy. So to some extend I understands what he is saying, but there are tons of games not like that.
No I don't think his hate for Unity is justifiable. If you care about tech then maybe yes, but if you care about games then no.
Related: http://blogs.unity3d.com/2012/04/01/semantic-game-builder-interface-to-be-integrated-into-unity/
He does have a point about the initial loading time, even for small scenes. A lot of companies create their own scene exporters in order to create their own loading pipeline.
The type of games made in Unity is often a simple physics based platformer. This is because Unity is an amazing tool for creating that kind of games, but no. Just because Unity is inviting (and to some extend does all the work for you) to make that kind of games it doesn't mean you have to create the game like that. A "mistake" a lot of people do is to make their games too physics based, because it is so easy. So to some extend I understands what he is saying, but there are tons of games not like that.
No I don't think his hate for Unity is justifiable. If you care about tech then maybe yes, but if you care about games then no.
Related: http://blogs.unity3d.com/2012/04/01/semantic-game-builder-interface-to-be-integrated-into-unity/
Are people actually using the built in editor and scene system for anything beyond prototyping or small games ?
For pretty much any reasonably large project you'll be far more productive with a more specialized system.
It's all in how you use the tool; it's not really Unity's fault if it allows to make incredibly generic game logic or to use the generic built-in shaders / materials, as it also provides you full means to customize them. As positive sides, the component-based scene model, the workflow and overall usability are pretty much incredible. Some negatives are the inherent performance and memory use problems caused by the Mono runtime and garbage collection.
In the worst case you could get your project 99% done but then hit for example some bad memory leak you aren't quite able to work around, and then you would just have to wait for an update and hope it improves things. I've seen this in context of a multiplayer server built from the same Unity project as the game client; in this case the server would always exhaust the memory and crash after running for some hours, but this was more than one year ago and I'm quite sure improvements have happened since.
To summarize, I don't think hate toward Unity isn't justified, but especially in a bigger project you have to know what you're getting into and experiment with the performance and stability before committing too deeply.
Saying Unity is slow is for most non-bad designed games ****shit.
I'm trying (and failing) to figure out what you censored with **** that was so bad you had to nix it.
Parts of Unity are incredibly frustrating. Serialization is poorly documented and is inconsistently implemented, the GUI sucks, and the editor can get out of hand quickly with large projects.
However, It is rather strange that your game dev teacher is talking about unity from the perspective of a player when your teacher is a developer yes?
Also, before GDC with UE4 and CryEngine, consider what options you had before unity? They were far and few between, and Unity sure beats writing your OWN engine.
Now of course this could limit you if your game doesn't fit well with what Unity provides. There is a statement that is made often on these forums "Make games, not engines". products like unity exist so that small teams or possibly 1 person that works their ass off can actually make a decent 3d game.
The only problem I have with Unity is actually in mono - the AOT compiler for iOS is utter shit when you're trying to do anything interesting with generic interfaces (like, you know, IEnumerable
Other than that and their crappy documentation, Unity is the best experience I've had in my entire professional game development career.
That said, unity does have a bunch of weirdness and quirks, which I attribute to the fact that they don't pass the "dog food" test - they don't consume their own product, they've never made a game with their own engine. CryTek, Id, Epic, Valve, all actually ship games on their engines, which forces them to care about any rough edges.
So, what games has your professor made?
So, what games has your professor made?
This, a thousand times.
"Faster" can have more than one meaning. Faster to write (productivity) and faster to finish (time to market) are very important, even more so when initial load times and runtime performance aren't critical issues. Shipping is a feature.
In the worst case you could get your project 99% done but then hit for example some bad memory leak you aren't quite able to work around, and then you would just have to wait for an update and hope it improves things. I've seen this in context of a multiplayer server built from the same Unity project as the game client; in this case the server would always exhaust the memory and crash after running for some hours, but this was more than one year ago and I'm quite sure improvements have happened since.
Yes, I've seen this before. I got pulled in to help with a project. He had completed the game with a Unity style IDE and tried to publish it for the 360. The game ran for four minutes, then crashed.
He had bought a none source code licence for Torque, so he was completely stuck. If all you have is the IDE and the engine as DLL's, it is very easy to end up in this situation. Dog Food is a very good test.
In the end he had to pay 10 times more to get the source code licence, (plus the cost for me to work on it) and I sat down and debugged Torque.
They were creating a new canvas for every screen update. The system very quickly used up all the memory and triggered garbage collections. The garbage collector couldn't cope and soon went bang.
Easy fix, but it just proves that the developers had never played a game for more than 4 minutes. Dog Food, I really like that as a concept.
In the worst case you could get your project 99% done but then hit for example some bad memory leak you aren't quite able to work around, and then you would just have to wait for an update and hope it improves things. I've seen this in context of a multiplayer server built from the same Unity project as the game client; in this case the server would always exhaust the memory and crash after running for some hours, but this was more than one year ago and I'm quite sure improvements have happened since.Yes, I've seen this before. I got pulled in to help with a project. He had completed the game with a Unity style IDE and tried to publish it for the 360. The game ran for four minutes, then crashed.
He had bought a none source code licence for Torque, so he was completely stuck. If all you have is the IDE and the engine as DLL's, it is very easy to end up in this situation. Dog Food is a very good test.
In the end he had to pay 10 times more to get the source code licence, (plus the cost for me to work on it) and I sat down and debugged Torque.
They were creating a new canvas for every screen update. The system very quickly used up all the memory and triggered garbage collections. The garbage collector couldn't cope and soon went bang.
Easy fix, but it just proves that the developers had never played a game for more than 4 minutes. Dog Food, I really like that as a concept.
Good one c:
Though in this story there is unclear to me why garbage collector "couldn't cope" could you maybe explain?
I've had a similar problem with Lua on consoles. The frame-time budget was 33ms (30 frames per second). The game systems were taking about 15ms and the render systems about 15ms. The garbage collection all by itself would sometimes take 8ms! That's a total of 38ms (26fps, or 20fps if vsync is on).
Though in this story there is unclear to me why garbage collector "couldn't cope" could you maybe explain?
We were in control of our engine, so we first fixed it by putting a limit on the amount of time that the garbage collector could run each frame. We limit it to 2ms per frame, and now the framerate is always perfect 30fps!
However, now, sometimes garbage builds up... If the GC would've run for 8ms, but we only let it run for 2ms, which means that we only cleaned up 25% of that frame's garbage -- 75% of that frame's garbage is left for the next frame to deal with!
Next frame, it generates another 100% of a frame's garbage, cleans up 25% of it, but leaves another 75% for the next frame to deal with, for a total of 150% of a frame's worth of garbage lying around.
Very quickly, your run out of RAM and crash...
The solution here was to rewrite all the Lua code to be extremely careful about when it generated garbage...
Generally on a PC, you can pretend that you've got "unlimited" RAM (especially on x64), because virtual memory kicks in, and only the memory that you're actually using is stored in RAM. Any bits of memory that you haven't used for a long time get moved to your hard-drive.
Consoles don't do this -- so if you run out of RAM, your game crashes.
Topic Locked
This topic has been locked by a moderator. New replies are not allowed.