Purchasing Decision - TGE vs. TGEA
by Trin · in Torque Game Engine · 08/08/2007 (9:13 am) · 6 replies
Hi all,
I'm a UT2003 modder looking to expand to for-profit game creation and it appears that Torque meets my needs. I've answered many of my questions by searching the forums, but the fact is I still have a billion questions. I'm going to try to limit my questions to those necessary to making a buying decision.
My game concept is a simple multiplayer FPS zombie shooter. Nothing too fancy.
I expect to have/need:
- Lots of AI opponents (zombies - slow moving, low poly, limited AI) - ideally 200+
- Primarily Interior based maps with lots of detail, some terrain maps with several explorable interiors
- Lots of interactable objects (doors, elevators, cabinets, radios, tools, etc.)
- Dynamic moving lights - i.e. flashlight, weapons with light attachments, mounted searchlights, etc.
- Multiplayer capability with 4-8 players
The obvious question is - Will Torque do what I want to do? I'm experienced with C++ and UnrealScript so I'm not afraid of coding the source or script aspects.
Assuming it will I need to figure out whether to buy TGE or TGEA. It appears that TGE 1.5 has the features I need but I'm not opposed to paying the extra just to have the latest and greatest or to squeeze out a little faster rendering.
Additional questions:
- Is TGE going to continue as a product or will TGEA eventually be the only supported engine?
- Is TGEA better in every way to TGE or are there tradeoffs? In other words, does TGE have advantages over TGEA?
- It appears there are loads of TGE community resources (scripts etc.) and very few TGEA resources. True?
- If I buy TGE and find I really needed TGEA is there an upgrade path that doesn't require a full cost TGEA license?
Thank you!!
I'm a UT2003 modder looking to expand to for-profit game creation and it appears that Torque meets my needs. I've answered many of my questions by searching the forums, but the fact is I still have a billion questions. I'm going to try to limit my questions to those necessary to making a buying decision.
My game concept is a simple multiplayer FPS zombie shooter. Nothing too fancy.
I expect to have/need:
- Lots of AI opponents (zombies - slow moving, low poly, limited AI) - ideally 200+
- Primarily Interior based maps with lots of detail, some terrain maps with several explorable interiors
- Lots of interactable objects (doors, elevators, cabinets, radios, tools, etc.)
- Dynamic moving lights - i.e. flashlight, weapons with light attachments, mounted searchlights, etc.
- Multiplayer capability with 4-8 players
The obvious question is - Will Torque do what I want to do? I'm experienced with C++ and UnrealScript so I'm not afraid of coding the source or script aspects.
Assuming it will I need to figure out whether to buy TGE or TGEA. It appears that TGE 1.5 has the features I need but I'm not opposed to paying the extra just to have the latest and greatest or to squeeze out a little faster rendering.
Additional questions:
- Is TGE going to continue as a product or will TGEA eventually be the only supported engine?
- Is TGEA better in every way to TGE or are there tradeoffs? In other words, does TGE have advantages over TGEA?
- It appears there are loads of TGE community resources (scripts etc.) and very few TGEA resources. True?
- If I buy TGE and find I really needed TGEA is there an upgrade path that doesn't require a full cost TGEA license?
Thank you!!
#2
I've got similar requirements, 500+ is my end target though.
That would be exceptionally harsh on TGE. The TGEA batching would make it alot friendlier. 200+ may still be pushing it though because in TGEA still does skinning on the CPU. I've gone and set up an optional short circuit that skips the skin transformations and offloads them to the GPU, this reduces the maximum number of bone influences to 4, but that's more than necessary on a low-poly character. In the end, only my collision hit boxes get transformed on the CPU (as is necessary).
This saves alot of CPU time when you've got a ton of characters. I'm still not breaking 200+ animated characters with logic yet.
Other general shader tricks will help to make batching more friendly and let you do things you wouldn't be able to do as easily with TGE, such as palette colorization or UV coordinate offsets on an Atlas texture letting you have 4+ textures for the same model.
You'll have to consider what data you really need to have for these characters. If they don't need energy then modify the code to get rid of it and any other unused values eating up memory.
The only comments I have on AI are that it's a bitch. Roughly 200 is doable though. More than that and AI is going to go out the window
(In all of the Total War games, the replays suffer from the fact that the individual characters have AI so low that they're placements and actions are mostly procedural. They're complicated enough as well that in a game where you had a slight run in with a group of enemy skirmishers who inflicted only 4 casualties on whatever unit will in the replay be slaughtered down to 1/4 of their original number, causing the rest of the battle to play out even more inaccurately where they're invovled. I've seen victories turn into defeats in Replay.)
Then you've got to figure out how to manage those objects most effectively, only updating the logic for those that are necessary, removing entities from the scenegraph yet still tracking them (perhaps by a hierarchial division e.g. Soldiers are members of -> Squads who are members of -> platoons). Allowing you still track basic soldier statistics but not fuss with their positions, animation states, etc unless there are players nearby. Then you start spawning "smart soldiers."
This is my hangup at the moment. Done properly (such as the recent Total Wars [which I checked, they only use shader constant instancing on higher end hardware and only for non-animated placeholder LOD models [on lower end hardware they both billboard and remove sooner), everywhere else they batch, and gpu skin) it should up maximum on screen to 200+ (afterall rarely in Medieval II do you ever see more than 200 characters at any distance that can be considered close.
Note: I consider close to be anywhere that the character cannot be swapped for an instanced non-animated (in the limbs sense) model [not necessarily truly swapped, just not doing anything other than changing position] or a blurry way off billboard.
The big Question
Are you capable of coding any necessary changes in the engine and not in script? Eventually almost everything pertaining to a large number of characters will need to be transferred to the engine code at some point to squeeze every CPU cycle you can get.
Asking because your description as being a modder lends me to the possibility that you're not a programmer.
Furthermore, on the topic of GPU Skinning, it's not a trivial undertaking. I've still bugs and quirks everywhere, nor do I have it at a point where it can be toggled at runtime.
Good News
All of the other requirements you have are trivial compared to rendering and managing 200+ entities.
09/11/2007 (5:42 am)
The biggest issue I see isQuote:Lots of AI opponents (zombies - slow moving, low poly, limited AI) - ideally 200+
I've got similar requirements, 500+ is my end target though.
That would be exceptionally harsh on TGE. The TGEA batching would make it alot friendlier. 200+ may still be pushing it though because in TGEA still does skinning on the CPU. I've gone and set up an optional short circuit that skips the skin transformations and offloads them to the GPU, this reduces the maximum number of bone influences to 4, but that's more than necessary on a low-poly character. In the end, only my collision hit boxes get transformed on the CPU (as is necessary).
This saves alot of CPU time when you've got a ton of characters. I'm still not breaking 200+ animated characters with logic yet.
Other general shader tricks will help to make batching more friendly and let you do things you wouldn't be able to do as easily with TGE, such as palette colorization or UV coordinate offsets on an Atlas texture letting you have 4+ textures for the same model.
You'll have to consider what data you really need to have for these characters. If they don't need energy then modify the code to get rid of it and any other unused values eating up memory.
The only comments I have on AI are that it's a bitch. Roughly 200 is doable though. More than that and AI is going to go out the window
(In all of the Total War games, the replays suffer from the fact that the individual characters have AI so low that they're placements and actions are mostly procedural. They're complicated enough as well that in a game where you had a slight run in with a group of enemy skirmishers who inflicted only 4 casualties on whatever unit will in the replay be slaughtered down to 1/4 of their original number, causing the rest of the battle to play out even more inaccurately where they're invovled. I've seen victories turn into defeats in Replay.)
Then you've got to figure out how to manage those objects most effectively, only updating the logic for those that are necessary, removing entities from the scenegraph yet still tracking them (perhaps by a hierarchial division e.g. Soldiers are members of -> Squads who are members of -> platoons). Allowing you still track basic soldier statistics but not fuss with their positions, animation states, etc unless there are players nearby. Then you start spawning "smart soldiers."
This is my hangup at the moment. Done properly (such as the recent Total Wars [which I checked, they only use shader constant instancing on higher end hardware and only for non-animated placeholder LOD models [on lower end hardware they both billboard and remove sooner), everywhere else they batch, and gpu skin) it should up maximum on screen to 200+ (afterall rarely in Medieval II do you ever see more than 200 characters at any distance that can be considered close.
Note: I consider close to be anywhere that the character cannot be swapped for an instanced non-animated (in the limbs sense) model [not necessarily truly swapped, just not doing anything other than changing position] or a blurry way off billboard.
The big Question
Are you capable of coding any necessary changes in the engine and not in script? Eventually almost everything pertaining to a large number of characters will need to be transferred to the engine code at some point to squeeze every CPU cycle you can get.
Asking because your description as being a modder lends me to the possibility that you're not a programmer.
Furthermore, on the topic of GPU Skinning, it's not a trivial undertaking. I've still bugs and quirks everywhere, nor do I have it at a point where it can be toggled at runtime.
Good News
All of the other requirements you have are trivial compared to rendering and managing 200+ entities.
#3
Sounds like TGEA has more liklihood of success in meeting my needs. That's what I needed to know to move forward.
I find the AI comments interesting. I wouldn't have thought the AI for 200+ opponents would be as limiting a factor as the polygon rendering. I'll consider myself warned.
To answer "the big question" yes, I am a programmer. I have 10+ years of corporate C++, several years of which dealt with an OpenGL app designed to allow engineers to model power plants in 3D. I have extensive experience with STL and MFC and I'm no stranger to squeezing cycles out of the CPU. One of the big draws for TGE/TGEA for me is the idea of getting my game programmer to a C++/scripting mix rather than scripting only.
I call myself a modder because in the gaming arena my experience is limited to modding.
Any comments on flashlights and/or doors? Are those part of the trivial group?
Again, thanks for the info. I really appreciate it!!
09/11/2007 (9:45 am)
Great info, guys. Thanks!! I'll admit that I had given up on getting any feedback here. Nice to see some informed discussion. If I buy the engine I can guarantee you that these replies have helped make that decision.Sounds like TGEA has more liklihood of success in meeting my needs. That's what I needed to know to move forward.
I find the AI comments interesting. I wouldn't have thought the AI for 200+ opponents would be as limiting a factor as the polygon rendering. I'll consider myself warned.
To answer "the big question" yes, I am a programmer. I have 10+ years of corporate C++, several years of which dealt with an OpenGL app designed to allow engineers to model power plants in 3D. I have extensive experience with STL and MFC and I'm no stranger to squeezing cycles out of the CPU. One of the big draws for TGE/TGEA for me is the idea of getting my game programmer to a C++/scripting mix rather than scripting only.
I call myself a modder because in the gaming arena my experience is limited to modding.
Any comments on flashlights and/or doors? Are those part of the trivial group?
Again, thanks for the info. I really appreciate it!!
#4
There is already a flashlight resource for TGE but I'm not sure how up-to-date it is.
09/11/2007 (1:29 pm)
Doors are trivial I would think, there are several resources created already. Flashlights may be a bit more difficult, but someone has used the AFX to great effect as a flashlight, but that is not yet out for TGEA. It doesn't look like it's miles off though.There is already a flashlight resource for TGE but I'm not sure how up-to-date it is.
#5
Gary (-;
09/11/2007 (2:09 pm)
Also don't forget that if you want to write your game for platforms other than just Windows, [eg mac or linux], TGEA doesn't work for that right now, and that seems unlikely to change.Gary (-;
#6
Doors = pretty trivial
Flashlights = shouldn't be too bad if you can settle for old school Alien vs. Predator 2 style flashlights (mount a spotlight to an item [flashlight model]). Doom 3 style would probably take a bit more work (you could use one of the fullscreen shading resources to do the visual aspect of the light [I don't know how to describe it] while still using a spotlight to produce the actual light).
Just be sure that you try out the demos of TGE and TGEA first and mess around with them.
09/15/2007 (7:56 am)
Well, I don't think you'll have too much trouble. Sounds like you've probably got far better odds than most lone-wolf developers.Quote:Any comments on flashlights and/or doors? Are those part of the trivial group?
Doors = pretty trivial
Flashlights = shouldn't be too bad if you can settle for old school Alien vs. Predator 2 style flashlights (mount a spotlight to an item [flashlight model]). Doom 3 style would probably take a bit more work (you could use one of the fullscreen shading resources to do the visual aspect of the light [I don't know how to describe it] while still using a spotlight to produce the actual light).
Just be sure that you try out the demos of TGE and TGEA first and mess around with them.
Torque Owner Andy Rollins
ZDay Game
To try and answer your questions:
Is TGE going to continue as a product or will TGEA eventually be the only supported engine?
TGE is an older product than TGEA, as such there are fewer bugs but it is most suitable for older hardware, there is a free community resource known as the Modernization kit that does add higher spec graphics to TGE (shader support for example).
Whilst it's not clear how much life there is in TGE as there is no published information about GG's strategy for the products I would suggest there is still a fair few years life in the product as there will still be older technology around for a long while, not to mention the version for the Nintendo Wii (which doesn't support shaders). Version 1.5 of Torque was a major upgrade from previous versions incorporting the lighting kit and a whole lot more so it doesn't appear that GG is ready to shelve it.
- Is TGEA better in every way to TGE or are there tradeoffs? In other words, does TGE have advantages over TGEA?
As I said before there is a lot of common areas with the two so they are the same there, TGEA is superior graphics layer and terrain basically. TGE has been around a lot longer so is more established, with fewer bugs , more resources, documentation and people with knowledge about it on the forums.
- It appears there are loads of TGE community resources (scripts etc.) and very few TGEA resources. True?
Yes true
- If I buy TGE and find I really needed TGEA is there an upgrade path that doesn't require a full cost TGEA license?
There is an upgrade offered to owners of TGE for purchasing TGEA, it's on the TGEA page