Can TGEA or T3D meet these requirements?

#1 1) Very much depends on your particular game. Network-wise, I've heard of people getting up to 300 clients on a server in network load testing, but you'll have to optimize the netcode for that (there's plenty of information about doing that on the forums). Once your player views 180 clients, their performance will likely slow considerably, but viewing 180 players at once is a big deal.

2) No idea.

3) Yes, I routinely have my view distance set to 1500, but the farther out you set that distance, the more of a performance hit you'll take. You can probably offset the hit by doing things like implementing distance-fading (a resource, I believe), and LOD.

4) That was covered in your other thread.

5) No idea.

6) How good are your artists? TGEA can look great- there's a demo nicknamed "Gears of Torque" out there you can check out with great AAA visuals, but you need to remember that art is the reason why all those AAA games cost $50MM to make. It has less to do with the engine (large texture sizes, shader support, and a thing or two else) and more to do with the quality of art that your team generates.

#2 Thanks Ted.

We currently use the lod system in the engine we work with now as a mod team. We'd need to assess the impact of distance fading (don't want a guy hiding behind an object that fades out).

My understanding is that TGE/TGEA are deficient in some optomizations areas such as the various types of culliing. I thought Torque engines do not perform any backface culling or occlusion culling, which I thought were pretty basic optomization techniques.

I think we'd probably be looking to T3D, so I am still not clear on the large terrain performance approaches. I thought I read somewhere in these forums that there was a "98% solution" for T3D (meaning it would meet 98% of the customers needs), but from the description it sounded like our project might be in the 2% not covered.

We have superb artists available on the team, so I am glad to hear that talent is a big factor in visual quality. While this is obvious to some extent, certainly no matter what you do soemthing like TGE is not going to look as modern as people have come to expect.

T3D appears to still be using a lightmapped approach. I know it's a budget engine (even at the possible $1000 price tag). But in 2009 lightmapping seems like a very dated approach. We'd be looking to have destructible buildings, sun and moon cycles, and *ideally* deformable terrain (to some extent, at least in the future). Those things just don't seem to fit with a lightmapped engine.

We currently use destructible buildings and such in a lightmapped engine and have had to do a lot of silly things for stuff to not look completely akward (we still get complete shadows rendered by walls that have had holes blown through them).

Thanks again for taking the time to respond - much appreciated.

#3

Quote:2. Realistic ballistics/projectile modeling so that targets (humans and vehicles) can be engaged at as far out as 1.5km

I've been experimenting a lot with the projectile system recently, and this is one question I have asked myself. What it comes down to is what exactly you mean by "realistic".
First off, the distance will not be a problem, but the out-of-the-box projectile behavior won't be satisfactory. There have been a few resources posted about how to add a simple implementation of ballistic drag. These are trivial additions, and could easily be adapted to become more realistic. That may be all you need.

I assume you are thinking of something similar in behavior to a .50 rifle, so you'll probably be wanting a muzzle velocity in the region of 800 to 900m/s. I have not yet tried anything that fast, but I think it would work OK. The projectile class works by recalculating its position every tick (32ms), and casting a ray from its current location to the next to detect collisions. So, it should hit things OK, even at high speeds.
What this means, however, is that while the overall trajectory may be correctly curved, it is essentially going in a straight line between each tick, so the bullet drop will not be super realistic. If your intention is to do very realistic ballistics, it will take a bit more effort to simulate. But I have had a few thoughts on that, and I see no reason why it couldn't be done.

Edit: when I say that bullet drop won't be realistic, it will at worst be no more than a few centimeters below the expected point of impact. Plenty good enough for most games, but visibly wrong if you are simulating something like a mildot reticle.

#4 @Eamonn, from what has been said in Brett Seyler's latest couple of blogs, I don't think that T3D will have baked shadowmaps as its primary shadowing solution.

Torque does indeed do various types of culling. I don't know where you heard it doesn't, but that's just not what happens. The LOD system is configurable by the artist, go look at the Art Documentation. At the lowest level of detail, you can have it render a billboard of the object at an appropriate angle to the viewer.

I'm assuming you want highly detailed characters and to have a large amount on screen at once. You might end up having to thread your skin mesh updates and up your game's requirements to do it well, but it's certainly possible.

#5 @ Tony: By realisitc, I mean pretty much what you have said, but I'd add in things like energy dissipation, gravitational forces, the parabolic curve of trajetory paths, etc. From what you are saying it sounds like there is a decent mechanism in TGEA to start with, but we'd need to modify the engine source to get a higher level of fidelity. Is that a correct interpretation?

We'd be using miulitary sights and would ideally be able to simulate bullet trajectories and behaviours to exact real world standards (to the best of our abilities and just before the point of diminishing returns in gameplay vs. effort).

Thanks for the answers.


@ Ross: I don't know what the development and release cycle for T3D is.. so it's difficult for me to gague where GG is at in the development cycle. But it seems pretty far along to be adding in something like dynamic lighting?

It's difficult (for me anyways) to discern comments about TGE from those about TGEA.. but I read on the DevMaster reviews that "Torque" does not support backface culling. Good to hear that is wrong!

==

Thanks for the answers. Am still not feeling as though I can comfortably say that TGEA or T3D is going to meet the requirements (which are steep for an indie game).

#6

Quote:From what you are saying it sounds like there is a decent mechanism in TGEA to start with, but we'd need to modify the engine source to get a higher level of fidelity. Is that a correct interpretation?

That's right. The standard projectile code provides a good base which does model gravity, but not drag or wind. Adding those things is a trivial change to the source code, and in fact you can very easily make it follow pretty much any curve you want. The maths of the trajectory is entirely up to you, and you could even replace the formula with a lookup into a ballistics table if you really wanted to.
The only issue you are likely to run into is the fact that the curve is effectively broken up into straight line segments, which, when looking through a calibrated scope, may start to become apparent at those high speeds and distances. But I believe that even that can be addressed without much effort.

So, yes you will need to modify the code, but you're looking at very little work to get what you want in this area. Plus, like I said, I have been playing with this sort of thing recently, and I was planning to release some of my work to the community. By the time any of this becomes an issue for you, you may find that there is a turn-key solution already waiting for you :)

#7 Thanks very much Tony, encouraging and look forward to any tips / resources.

#8

Quote:@ Ross: I don't know what the development and release cycle for T3D is.. so it's difficult for me to gague where GG is at in the development cycle. But it seems pretty far along to be adding in something like dynamic lighting?

That would be true if they were just now starting to add it. I imagine that's not the case, since I believe T3D has been in development from sometime before 1.8 shipped.

#9 Ahh understood, tyvm. I've only kept a sort of peripheral eye on Torque dev cycles, so I am not clear on what the process looks like.

But.. er... I'd still like to see a clearly articulated roadmap.

#10 Your best bet for T3D info is going to be to keep an eye on Brett Seyler's plan postings.