[FVG]: Choosing an Engine: Final Poll

[Tristan Lall]

Originally Posted by Astronouth7303

The unreal engine has a whole host of legal issues involved. Also, a 'mod' is considered significantly less than 'game'. (Which would you rather tell your friends? "I helped make the FIRST Video Game." "I helped make the FIRST Unreal Mod.")

"Total conversion" mods can be quite extensive--look at some of the Unreal variants out there, or at the variety of things that have been done with Quake. Also, as for the legality, one must consider that Epic provides an editor with their software. That could constitute some sort of implied consent; at the very least, it's clear that they won't sue you if you don't sell it.

In any case, I've got to point out that some of this sounds suspiciously like an issue of pride (which is fine, understandable and all that), but which also means that if you fail to make it work, the consequences w/r/t your pride may be more significant, because of higher expectations.

Originally Posted by Astronouth7303

I excpect that the diference of dificulty between Unreal and CS APIs are slim to none. CS has many of the Features that we need.

True enough, but do you need API support beyond whatever scripting is provided in the Unreal editors? If this is truly so, then perhaps you're right about Crystal.

Originally Posted by Astronouth7303

While using Unreal would produce ultra-realistic scenes, there are 2 things that should be remembered: 1. that it can only do this on the best macines; 2. that it can only make it as good as the models.

1. I've managed to run the old Unreal Tournament at 30 fps on a PII 450 with 384 MB RAM, and a 32 MB GeForce 2 MX (serving as host for a network game). That's the pride of 1998, with 2000's video card. It runs at about 20-25 fps on a PII 333 laptop with 192 MB, and an 8 MB Rage Pro. The dedicated server works on a Pentium 166 with 32 MB of RAM, even with a Windows 2000 session running (not blindingly fast, but certainly playable with those other two computers as clients--and the Windows 2000 thing is admittedly a bad idea on a computer of 1996 vintage). In all cases, graphical quality was nothing to be ashamed of--1024×768×16; and certainly no worse-looking than Quake II. Like I said, Unreal really is spectacularly efficient.
2. There's a bit of CAD and 3-D graphical design talent in the FIRST community--why not make use of it? The models can be as good as you need them to be (i.e. start with a box, and progress from there).

By choosing Crystal, you do have access to a reasonably complete feature set (to which you referred); you don't, however, have the assurance that the engine is stable and bug-free. With Unreal, you have a very large support community, and a proven technology, but are limited in what you can do to the actual code.

As I understand it, you want to make a 3-D simulation of a field/stadium, and populate it with robots and objects. Couldn't this be accomplished with an Unreal map mod, several new bots, a few objects like balls or goals, and an adjusted set of physics parameters? I still don't see exactly what you need to get into the source code for. If it's not the learning experience that is primarily driving you, why not take the slightly more straightforward route? Also consider that you can test models, game dynamics and all sorts of other issues in Unreal, then (if the Crystal feature set is to be believed), you should be able to port it to native Crystal, if things work out. It's like rapid application development....

Also, things like announcers aren't as simple as they appear. In the other thread, D.J. Fluck accused the EA Sports announcers of being stilted and unimaginative in their phrasing. This is true--but not for lack of trying (EA has tried play-by play since at least 1997, and it still isn't perfected.) There are hundreds of different phrases and names in NHL 2003 (for example); they have to be accessed on a moment's notice, and played in some coherent order. When Kaberle receives the puck, it doesn't help for the announcer to still be talking about Sundin--and if Kaberle holds onto the puck, the announcer can't just sit silently and say nothing. On the other hand, the implementation of play by play might be more than an Unreal mod can handle--so this could be a backhanded vote of confidence for Crystal (just brace yourselves for a vast library of sounds that can be pulled up instantly, and an algorithm to tie them together in coherent patterns).

Creating the game from raw source code is clearly the more significant achievement, but without having a true measure of everyone's committment to this project, I can't help but wonder if you are biting off more than you can chew. I don't by any means consider the use of Crystal to be impossible, only difficult. So as project leaders, it is upon you to know your (and your associates') capabilities, and choose the course of action that benefits the project the most. (If you haven't already come to some conclusions about each other's abilities and potential for improvement, now is the time for an honest discussion. Leave pride out of it for now.)