UDK - Opinions

[Feralidragon]

So, serious question: from who actually at least tried UDK, is it difficult to learn? Is UScript similar (well, I it seems so from the basic code questions I see in Epic forums specially related to replication, but would like to hear from who actually already tried it out)?

Anyway, I got a urge today to check out its FAQ, features and whatnot and it's convincing so far, and I have seen people with reasonable works in UT1 and then moving to UDK and do gorgeous stuff, which leads me to think that UDK may actually be optimized to be productive and easy to use and let only the imagination be the limit, and there's the huge advantage of being able to do a free standalone game out of it without any of the limitations mods have, and without the frustrating limitations of UEngine 1 as a whole (the kind of limitations that make me use MS3D to model and animate (=pain) and not being able to use the GPU for what it's meant to be used (lag with 2.000 poly models... lol wut?)).

I intend to start a quite ambitious concept on it (after I release a few things for UT1), and which I didn't see anywhere else, and then progress with it as time and my skill allows it (probably it's "more sand than my wagon can hold", but we'll see how far this may go, if started at all).
It's nothing set in stone yet, and it's likely I will start a much simpler and straightforward version in UT, the thing is, in UT is very tricky and has a high chance of failing miserably.

Anyway, what's your opinion in the UDK?

[Shadow]

Hm well, I tried it, tested it. Uscript has far more features than our all day UE1.x UScript (dynamic arrays, struct defaults etc.) but I was simply overthrown by the possibilities.. it's a nice set of tools, but I'm just a fucking lazy bastard. If you wanna go UDK after our UE1 stuff, I might follow

[papercoffee]

Anyway, what's your opinion in the UDK?

Well, to be honest ...We discussed it (Gen and me) and we want to transform the FF concept, after it is released for UT99, in an standalone game maybe on UDK or the Unity engine.

For UDK is this main problem, if you want to sell your game...

Licensing:
Next we get into the cost and licensing restrictions, here is where it all starts breaking down for me. UDK is free to use which is great, allows you and your team to start working immediately with no additional cost. When your project is complete you can release it free of charge and not have to pay Epic anything (again great for hobby / freeware games!). If you decide you want to sell the game you made, go ahead. You only need a $99 license to start selling your game. This will allow you to collect all of the profits from your game up to the first $50,000. Everything you make after the first $50,000 you must start paying 25% royalties on... (ouch!)

If you only make the game for the lulz ...then will there be no problem.

[Feralidragon]

@Shadow: Nice, so it's really that great.
With my real life job being a php coder (also designer 10% of the times when one of the bosses either wants or lets me do so), I miss greatly having the ability of creating dynamic arrays in uscript (either to store configurations, or create a simple DB or simply for processing purposes).
But tell me something, are these dynamic arrays also associative and multidimensional, or not? (I heard that they were still vectors, so you have to create arrays of arrays instead manually, but not sure if they are associative)

@papercoffee: Well, at least for me, I don't plan to make a paid game, it would be simply a free game at start and, just in the sole case it became really popular, I would make it a free to play with just aesthetics and some other non-gameplay influencing components to have to be paid for (so the game doesn't become pay to win in any way), and even perhaps making a spin-off later on with a single player campaign or something, or just a new paid game.
The license is not a problem provided that you get someone to pay it for you.
For example:
- paid games:
a) a publisher may take it and pay the license (or use one already owned by them since they're going to sell it);
b) you can ask someone to borrow (not advisable, but a possible solution);
c) you can ask for donations to pay the license before the game goes on sale and then give the game for free to who donated as a reward (these donations could be something like half or a third of the actual price you're planning of selling the game)
d) you can try an arrangement with Epic Games to make your first 99$ on the game selling to go directly to them to pay off the license (much less likely than any of the above, since they would need some legal adjustments to the license);

- free to play games (only certain things can get paid):
The same as above on a) and b), or also donations ( c) ) but a bit differently: instead they would have access to certain things or unique stuff in the game no one else would be able to have ever (not pay to win stuff, just aesthetics and other similar things), while in d) it would be the revenue of said services later on.

And I am sure there are other ways, but none of them is ideal and for full independence and zero annoyances is better to just pay it off once it goes on sale or once it becomes a tool to win money directly from a game.

But again, I am not really worried about that right now personally, since I may never actually do anything, although I am seriously considering the possibilities. In all honesty, in my case, going to UDK or not will be decided by something within a few weeks.
But yeah it sucks you have to pay a license right off the bat to sell something (they could always let you at least start selling it first and send any money got to them until the license is covered up, hell they could even add a tax to it to win even more money in the process and it wouldn't make a difference to a selling game (unless the game totally sucks).

[Shadow]

Ferali that's also a big reason for me to use C++ most of the time now, particle engine without dynamic array!? forget about it^^

Well you might have a look here: http://udn.epicgames.com/Three/UnrealSc ... l#DynArray

The third incarnation of UScript also provides, as you can see, a lot of functions and even an addition iterator! (which are present on Unreal Engine 1.x too.. but again only on the C++ level, I myself made only a few dynamic arrays functions for the SDK)

btw: Unreal Engine 4 comes without Unreal Script!

[Isotoxin]

So, serious question: from who actually at least tried UDK, is it difficult to learn? Is UScript similar (well, I it seems so from the basic code questions I see in Epic forums specially related to replication, but would like to hear from who actually already tried it out)?

Anyway, I got a urge today to check out its FAQ, features and whatnot and it's convincing so far, and I have seen people with reasonable works in UT1 and then moving to UDK and do gorgeous stuff, which leads me to think that UDK may actually be optimized to be productive and easy to use and let only the imagination be the limit, and there's the huge advantage of being able to do a free standalone game out of it without any of the limitations mods have, and without the frustrating limitations of UEngine 1 as a whole (the kind of limitations that make me use MS3D to model and animate (=pain) and not being able to use the GPU for what it's meant to be used (lag with 2.000 poly models... lol wut?)).

I intend to start a quite ambitious concept on it (after I release a few things for UT1), and which I didn't see anywhere else, and then progress with it as time and my skill allows it (probably it's "more sand than my wagon can hold", but we'll see how far this may go, if started at all).
It's nothing set in stone yet, and it's likely I will start a much simpler and straightforward version in UT, the thing is, in UT is very tricky and has a high chance of failing miserably.

Anyway, what's your opinion in the UDK?

Well, I like it, everyting is possible in it, i'm still learning it though. But, I do not like working with meshes.

[Creavion]

But, I do not like working with meshes.

This is like walking without moving your legs. With that way of thinking you will be stuck with outdated old bsp engines forever.
Basically within 3D model applications all geometry structures are meshes you know. It depends later on the engine and your decision what ends up as bsp, static meshes, skeletal meshes...

@UDK:
Well, what I somehow fear...
I am an amateuer with 3D modelling ... still. My UV mapping skils are damn low! I also have not animated anything so far. If it comes to realistic things, I would fail as well. It is somehow enough for basic foilage and landscape. For everything else I would have to study a lot of tutorials. Before I would have gained knowledge which is required for current level and game design, I would have missed the connection already. I couldn`t be on my own. While with UT a done map is a success, I would be satisfied with a complex mesh or foilage or whatever. And THEN I have to deal with the UDK UnrealEd itself... and I do not have that much free time anymore. I would wonder how long it would take to finish a simple project with everything done by permitted resources.

Edit: Several (former) skilled ut2004 and/or ut3 mappers from important mappacks are in the industry now. That way they can proceed with their overall skills much easier. They are able to connect their hobbies with their work. But if you do not deal there with a 3d modelling application, photoshop ... I do not think it is really possible to succed in the near future. You wil just lack of available/free time. it is nearly impossible to manage that. Also I think if you are new with the UDK and you have only dealed with the older UnrealEds before with some modelling, you would need some "guide". I think, it is very stupid to start just learning, by simply picking up anything. After some time you would be totally lost.

[Feralidragon]

@Shadow: Thanks for the reference. Unfortunately it seems doesn't support do associative arrays, they're still vectorial only and no boolean arrays (I never got why they can't do it... the only logic explanation is that they are optimizing memory usage on booleans by storing 8 booleans in a single byte at a time, and then accessing them bitwise, but it's just a wild guess though).

Also, I don't know if you know this, but at least in Java, when you add 1 more element to a dynamic array, the entire array is copied to a new location in memory and the new element added to the end of it. So basically dynamic arrays are memory hogs in Java (and a problem for mobile devices with very limited memory), so I wonder if UE3 makes the same thing, or if they optimized it in some way... I would hate to think that if I have a dynamic array of 1000 elements, and if I add 1 element, that it gets copied to new location as 1001 elements and leaves that old memory position as garbage.

But anyway, the sheer possibility of having them is absolutely great, for a lot of things.

Btw, I know that about UE4, it seems they are going to replace it by Kismet alone, and leave all the actual coding to C++ alone. Perhaps is better that way, as long as their new editor does what they promised without issues.


@Isotoxin: if you grab a good modeling application and apply to it for 1 month or 2, you will thank the day you did so. Meshes = more detailed stuff without any BSP headaches.
At max you may get some lighting issues, but nothing major.

@Creavion: Spent some time seeing their FAQ, and some videos around the web, from "basic" to "advanced" stuff. Afaik, I still didn't see anything other than some very specific features that an average UT1 mapper doesn't know already.
I have still yet to try it though, but it seems as easy as UEd2.0, but with many more mind blowing features, and it's not that hard to setup stuff at all, so is mostly down to the level designer ability to create meshes, and skill comes with practice/experience.

[iloveut99]

Also, I don't know if you know this, but at least in Java, when you add 1 more element to a dynamic array, the entire array is copied to a new location in memory and the new element added to the end of it. So basically dynamic arrays are memory hogs in Java (and a problem for mobile devices with very limited memory), so I wonder if UE3 makes the same thing, or if they optimized it in some way... I would hate to think that if I have a dynamic array of 1000 elements, and if I add 1 element, that it gets copied to new location as 1001 elements and leaves that old memory position as garbage.

Hi Ferali,

That's not exactly true, Java dynamic arrays (C++ seems to use it too) don't get copied every time you add 1 element. This objects contain a variable called capacity that tells to which position there were allocated memory. This means you usually have arrays that aren't filling all the allocated space. Once the capacity gets exceed the memory is then reallocated. (not sure if it calls garbage collection automatically to free the previous allocated one)

For c++ here is suggested to double the memory capacity each time an array is filled:
http://www.cplusplus.com/forum/general/11111/

A better explanation how capacity is used at java is found at this link:
http://stackoverflow.com/questions/3564 ... -arraylist

So you don't need to worry much about the memory hogs with dynamic arrays (you can even set the capacity for yourself, to reduce the memory allocations or even call garbage collection manually?).

[Shadow]

Yeah Ferali, there're a alot of things and decision Epic did we do not understand in the first place xDD

For me I can say I really like the idea of C++/Kismet Combination as new programming and code base, beacuse the UScript/C++ combination is sometimes horrible!! *cough* UScript variable list VS. C++ variable list and unexpected problems *cough*

[Feralidragon]

@iloveut99: Thanks for the info. I never went to confirm it myself, a java (and Android) expert here actually told me that and thus he avoids using dynamic arrays (he would rather know the max length and create a static one or do another algorithm altogether), since memory and processing in mobile devices is extremely low, specially older ones which he still tries to support.

@Shadow: Well, it seems cool to who is already used to C++, but to the ones that don't know it it will be hard to adapt. I mean, UScript is easy, very very easy to code with, as for C++ it's harder and it's a more hardcore language.
Well, perhaps it will still be easy in UE4 since it will perhaps be used to mostly change existing stuff or overriding functions as it's already done in UScript, and that extra coding layer is non-existent thus avoiding those problems, but still I think people will stick to UE3 or just leave to another less demanding engine (not only because of the UScript extinction, but also UE4 is certainly much much heavier, and there are few machines which can handle it nowadays, so perhaps it's still better to do games with UE3 as that way a wider range of people will be able to play it, unless they optimize their engine to run in the same kind of machines as long as you do an equally less demanding game).

But anyway, checked some videos on UDK, and it's like tingling a carrot in front of a horse for me... I am strongly considering it and finally give a good usage to my i7 cpu.

[Rakiayn]

im curious about your idea
or is it a top secret project? :p

[Feralidragon]

Actually, yeah, it is somehow "top secret", of the kind that you don't even share with your friends until you have something already done to show up.
The industry is full of people and entire (big) companies stealing ideas from others, and this is the kind of idea (along with a few others I have) which I want to preserve to myself (it may sound selfish, but I would hate starting an idea and someone else release something sooner and then telling I copied it... -.-)

But I can give a small grasp in the "concept": imagine a game where you build the game yourself while in-game. Confusing? Basically I mean almost full freedom and usage of any of your capabilities (aim skill, intelligence, strategy, architecture) with the possibility of having any kind of gametype (fps team games, deathmatch, coop, pve or just plain exploration and fun with your friends) you wish as any kind of map your wish and built literally with your (and potentially your friends) hands.

In short, the concept is to try to get all I ever missed in a game: freedom, creation and destruction, with a bit of fps gameplay in it, all in one package, and everything people like to do. And with UT this is nearly impossible to do, while with the UDK it is perfectly doable (since it has all the features such a game would ever need).
And the games that ever got remotely close to this concept, ended up being extremely limited and downright boring after a while, where this one if well executed could be really really addictive.

I already thought out a few things in how they would work and all, but it's just theory until I manage to start it (if I ever do). It's quite an ambitious and complex concept though.

[Creavion]

This is not selfish, this is pretty damn normal! Even by not earning any money with it, it is all about glory and fame!

But I can give a small grasp in the "concept": imagine a game where you build the game yourself while in-game.

Seriously, this was intended, wasn`t it?

[Rakiayn]

I understand you keep it to yourself.
the concept sounds interesting though.