And this is exactly what a good engine does: occlusion, automatic removing of content. A good scene management renders almost only visible polygons and tries to generate as less render batches as possible and tries to minimize the amount of state changes in the graphics card.

But it is not only the scene management of models. The management of textures, memory and the quality and optimization of shaders has to be taken into account to achieve good performance.

Because of this a huge scene (indoor or outdoor, depends on your project) can be a good scenario to explore the power of such a technology.

@Losticmate:
In the end you are assuming and I would say: seeing is believing.
I saw impressive scenes in T3D, in Unity and C4. They made their points, they proved what they can deliver.


That's years we wait for an impressive demo : nothing !
And i think we can wait some years again
Losticmate you like a lot A7 , we like the engine also , but unliek you , we are not fanboys, we are realistic and admit worfklow is behind some others, power also.

Some of us , are not people that want to play with tools or code , we don't want puzzle engines or toy engines
We think lot more in terms of making a game , with good features , without loosing time on coding or trying to integrate missing features.

Some of us see lot more making a game something more serious than playing all year with code , and we see it more in terms of time/delay and global project to reach ot a final game
Lot of other people like coding : it's OK, but other of us more Artists than coders ,
don't mind about playing with code or implementing features.

If you would think in terms of delivering a game (before in 1 or 2 years) :
you'll think differently and won't wait months for features you find in anothers engine ready/easy ot use for similar price.

We think lot more in tersm of easy to use features and workflow:
Just click on object properties to bring shadow mapping ;
just choose shaders on a list , just choose textures directly in object properties etc ... etc ...

That's the difference between artists bringing scenes and seeing problems, and programmers saying they can code all missing features and make A7 as powerfull as Unreal Engine 3 !

Yes, I'm definitely guilty of that one. I'll do some next week after my exams, just don't expect nice screenshots because I'm not about to make decent assets for the sake of a stress test, but thankfully it's things like the poly-count and texture resolutions that count, and they'll be there. Then I'll write some shaders for it, which WILL be good quality but take me longer (because there's no such thing as a "place-holder" shader -- we obviously want to see SSAO like that and shadows like those).
True, but those later on are in the comments. He appeared to post the blog under the impression that the first test was ground-breaking.

In the A7 manual there's a general material defined for particles, which has "future use" written next to it. I'm going to go hound jcl for that feature so I can write soft-particles -- until we get that, I'll try write soft-sprites to prove I can do it

Jibb

EDIT: And yes, I know I'm guilty in this post of making a post that does not relate to Torque 3D at all.

EDIT2:
I don't think that's a great argument. I, as a programmer, don't prefer A7 because I get to program missing features. I prefer it because it's an environment in which I can program innovative features without touching or looking at engine source code. By innovative, I again refer to my example of me fiddling around with SSAO before I'd even heard of it. There are other avenues (more shader stuff, but also other areas such as physics) that I can write stuff as I imagine it, or write it very soon after someone else invents it and opens it to the public, instead of waiting around for the engine developer to implement it or diving into source code.

Why haven't I released anything innovative into the 3DGS community? Quite frankly I haven't touched A7 in more than a year because of other commitments/priorities/issues. And honestly it's also because I'm lazy

wow. Ok thats bordering on personal insult. fanboy? I have already said, if i had the cash around I'd get torque3d or unity pro in a split second. The main two things I have said here is that people keep bitching and comparing those two engines workflow when its not a fair comparison, and saying that a7 can't do something just because they cant do it.

as far as both the "having never tested it" and the "bad framerate" comment: with the exception of volumetric god rays, a precipitation shader (which I've already described an even better process in the shader forum) and dynamic shadow mapping I have done with a7 everything that has been showing, I've implemented ssao, bloom, normal mapping, huge complex levels with entity management, dof, ive made editors for materials and landscapes. I just dont feel the need to recreate them all to prove something to you or why_do_i_die.

and as far as it being a toy engine, well apparently a toy engine gets my rent, food on my table, my utility bills, my savings, my taxes, and all my spending money. and tell ichiro lambe or david landcaster or dan_silverman, or andvari.... That its a toy engine.

ok, pm me when you finish a nice one. I can understand the placeholders, I'm a programmer. If you could also put a stats panel in there (fps, visible polys, ents, etc) it would be nice too.



Well it is a nice test still, even one 167k poly fortress with a desert scene (trees, buildings, huge terrain, etc) is still a nice test. When testing with A6 I couldn't come close to having that many polys or entities without the engine throwing up on itself. The terrain for A6 was rather slow.



Mkay

Although my experience with A6 has shown that using sprites over particles is faster.



pffff, what's a little off topic-ness between arguments



Well, I for one don't have Torque, so I'm not gonna say that the engine is groundbreaking and will destroy all engines out there. I will say that the test was really good nonetheless, the facts of the sizes and polycounts alone gives me a jump in wanting to get it.

I'll admit that I don't have A7, but I also don't plan on wasting my money on it(no offense). The reason I do not care for 3DGS much is this; Every time I do a decent test of the engine, a simple test that most other engines can handle the engine dies. I mean a simple test of just some entities and a small amount of code. My experience with A6 has shown me this:

1. The engine renders entities as a whole, meaning that if the textures are changed it doesn't make that huge of a difference. 1024 texs take alot of memory compared to 512, which is right, but anything around the 256 and below area doesn't really change the framerate at all.
2. LOD doesn't make as big a difference as it should. The engine doesn't seem to depend on polys as much as the fact that there is another entity in the game. For instance: In PreVa, we did an Island map that had trees around the island. IF we made the trees individual trees the engine crapped on itself. The framerate was horrible, so we had to make the trees all one model, they still had the same amount of polygons, but once they were made one model the difference was in about 20 fps or more. Making trees all one model like that is a big no-no, but we had to in order to get the framerate we wanted.
3. Sprites render waaaaaay faster than particles. You can pull off alot more sprites than you can particles. You can do some dense smoke and stuff with sprites when if you try to do it with particles, it dies. That's rather odd considering that a particle should have a little less code on it than a sprite, a sprite should be rendered similar to a model.
4. Terrains are not a good idea, even small ones with low poly counts cause extreme framerate drops compared to Torques MegaTerrains.
5. Even making a simple shooter with A6 is slow. I mean a scrolling shooter. Kino One's framerate is a bit slow on certain levels and we can't help it. It's not collision, it's not AI, it's not the player. Just from having entities with low poly counts.

I know that A7 is probably different and I've been told to upgrade quite a few times, but frankly, I don't want to. It's a waste of time and money.

The reason I want torque is that with some of the games I've seen are more than we pulled off in PreVa which leads me to believe that We could've pulled off PreVa with a very high fps and could've done way more than we did with A6. And that's just TGE.

I'm off topic but meh...

uh, you realize that if you want to test that fairly you need to have the same machine they have. you afaik were testing with a geforce 6100 which is now-a-days, basically crap compared to even the cheap cards out now.

with the texture resolution you have to remember, reducing from 1024x to 512 is reducing it by 4x, but the amount of pixels is a ton, were going from 512 to 256 your only taking of about 16th as much off.

as far as the problems with the trees its like that with a lot of engines, and a7 is missing one feature that they do really need, and thats object instancing.

as far as what you pulled off with preva vs what you coulda got with tge.... believe me i have love for preva, but you coulda still pulled a lot more off with it than you did and i never understood how you repeatedly said you had all these performance issues.



And i keep telling myself my next post will be on topic

well... @mpdeveloper, idk what you went through developing preva so i cant say much, but based on MY tests, i really think i couldve remade preva to run on a7 to look and play better... dont know if you'll ever make part 2, but that is a goal of mine... maybe you could have me develop part 2 or atleast a demo using the preva models anyways, thats not the point here... i understand where ur coming from but i do think a7 is a leap over a6.. based on work i've done so far...

the thing is just this, the ARTISTIC work flow of torque seems really easy... but i kno it can be done with 3dgs, even if i cant pull it off



lol.. i kno wat u mean

Ok, I respect you LC, and you're an ok guy, but you really don't know what you're talking about. I didn't just test on one pc that had a 6100, we have tested on more than 5 or 6 pcs and got the same kind of performance.

As for what we could pull off, no, there couldn't have been more. The fact that the game as is couldn't run on an older system with graphics that were not much more than a 2001 game, when the games from that era ran at blistering speed on a 6100, is pathetic.

If you think that we just placed alot of entities in the levels and left them as-is, you're completely wrong. The reason the game looks the way it does is because after countless (2 years) of testing multiple ways of doing the levels we had to go with a very simple setup or else the framerate was below 20, and yes that involves systems that are updated (or were updated for that time). It's pathetic that to run the game you have to have a modern setup considering that the game was very simple in coding and design.

You have the Game, you got it for beta testing. If you think the code is the problem, look at it. If you think the levels are the problem open them up and edit them.

I had to make my own scanning code because the engine's was too slow. We had to cut corners on multiple things.

As for David Lancaster, I know a decent bit about his work, we're kinda buddies. He quit working on 3DGS a while back, and his project that he was working on A6 with was turning out too slow. He then got A7 and it was running better but I assume he ran into more problems for them to switch engines. We had been talking about Torque and 3DGS too.



I can't say much on PreVa, but the actual name is a joke. We named it PreVa as a predecessor to a game that we wanted to make. Originally PreVa was just going to be a tech demo. Yes we plan on more games in the timeline of PreVa, as a matter of fact we are going to do a version of it on torque. We originally planned for multiplayer and ladder matches, but we decided not to with A6. That's all I can say on that. :P

As for a PreVa 2...you can't make one. We wouldn't take too kindly to our game being released by someone else. However, if you want to work with us we have a job contact from our site and we could always use someone else. I'll put you up for someone to hire.

If you want to do some tech tests with it though you can show me what you could do, but probably no public releases. I would like to see what you can do though. If you pull off something pretty good then it might be a little more push to hire you.

@Both: If you guys want to do a level test of PreVa, do so, but you must use A6 considering that the argument is over whether A6 could pull off more or not. I'm sure that A7 has more and that it could pull the game off a bit better, but we have made the decision to not use 3DGS again for our mech games.

As for being off-topic....maybe we should pm this stuff to each other?

umm.. i wasnt arguing over a6's capabilities, but i'd love to do the tech demo, though i would use a7... if i get the go ahead i'll work on smn and show you the vids, wouldnt need the world to see em... up to you

I have seen this on so many forums, people saying this and that about different engines,middleware and software developement kits, but the truth is how many actual games are made by anyone on any of the game engine forums,the fact is indie games are never going to be next gen,thats why we use engines that cost less than $5,000, if we wanted to make a next gen game we would need a budget running into millions. But we can make games of doom 3 quality with any of the indie game engines.You all talk about workflow but its the end product that matters.Its like some people say blender is a hard program to learn and the ui is crap, but it still can give same quality as 3ds max or maya.