Idea for Expanding Online Presence of STK

[leftisright]

Currently on the STK website, the only way a person can experience the thrill,quality and capability of STK is by watching a video on YouTube of someone ELSE playing the game,rather than the potential gamer. I do not know if the idea I am stating is currently under development or not, but either way I would love to be a part of the implementation of this idea. To be precise, the main idea I want to put forth is ONLINE TRIAL. Many commercial and successful games include a significant part of the game playable as a trial, to just get a feel of the game and its interface. By introducing this concept into the STK, we can expand our users as well as developer base by giving the game more visibility on the Internet.

The online "trial" of the game will probably be the best option for people who don't have the bandwidth and/or the motivation to download the complete 200 mb(approx) game without first checking it out. To keep the online game light weight, we can load only,say, about five levels, multiplayer support and restricted variety of sound effects and so on.. Once the gamer completes the trial, he may be prompted to download the complete game, if interested.

Potentially, the problems with this approach are the same that led me to this idea- the overhead of loading the game into a browser online. I am absolutely clue less as to how heavy the game will actually be, if at all implemented on to the online platform, which programming/scripting language is to be used and so on.. All such problems with this suggestion are thoroughly welcomed!

[charlie]

200mb just isn't that much any more.

Really what you are advocating (the term "online trial" is misleading and a silly notion) is a cut down version with maybe 1 track and a couple of karts so it only ways maybe 10-20mb instead of a few hundred. How feasible that is, well, it's not for me to say.

[leftisright]

Yes I agree 200 mb could not be classified as too large, but the thing I'm trying to express here is that we can probably increase the visibility of the game and make it more interesting for the audience, if they play the game for a while. Personally, I feel an interactive, playable trailer of the game would be much more fun and intriguing to a person than simply watching the video. and as far as the feasibility goes, if properly implemented, the trailer ("cut down version") could also be fun and intuitive. Any thoughts on that?

[Totoplus62]

This game will be very popular when people will be able to play online imho. I'm not sure a "trial" will be necessary

In the future (when online mode will be done), this game should have a forum especially dedicated to the players (and not the developers like this one) to share ranking, feelings, trolling etc ...

[Arthur]

Hi and welcome.

Sorry, but this idea is not very good in practice. Making the game available through a browser would mean having to remake most if not the whole game, and that would take far too long and too much resources that could be better spent on the game as it is. However, offering less tracks and karts with the game but allow people to download them when needed from within the game is something we are considering. This way it would be a smaller initial download and hopefully make it more accessible that way.

[leftisright]

Sorry to argue my point further, but I feel this could really help, hence the persistence in this

Making the game available through a browser would mean having to remake most if not the whole game, and that would take far too long and too much resources that could be better spent on the game as it is.

Arthur sir, I'm not saying that we make the entire game all over again to the online platform I'm simply suggesting we regenerate only the following-
1. One challenge of each type(timed,soccer,follow the leader,etc)
2. Sound track for each of those(plus may be one for the interactive menu)
3. GUI to navigate through these races and to get the person's details before registering (may I add another point here- we prompt these people to opt in to news of our further releases,updates and add-ons).

And as Totoplus said, the game will be popular AFTER the online mode will be done. Till then, I feel this should be used as a replacement for the YouTube video being used right now for showing the gameplay

[Arthur]

Our 3D renderer doesn't do WebGL and even if it did technically it would not be able to render fast enough in a browser for most people. Again, doing this would be a major rewrite (at least everything graphics-related).

[leftisright]

Alright Thanks for the insight everyone

[gouessej]

@leftisright I agree with Arthur and I advise you to read my article about WebGL to see why using it would cause a lot of troubles especially under GNU Linux. I like Super Tux Kart, I find it even better than Super Mario Kart on Gamecube (which I find particularly disappointing and poor) but its performance is not that good on my main computer (AMD Sempron 2600+, ATI Radeon 9250 Pro supporting OpenGL 1.3) and it would be worse with WebGL for sure, it can only be worse when I get half a frame per second or nothing at all. Arthur makes an interesting suggestion which is a good compromise, providing a "lightweight" version. Maintaining both an "offline" version and an online version is challenging especially with technologies not designed for this purpose and as far as I know, Google NaCl (native client) can't work without Pepper API which is only available in Chrome and Chromium. I assume Super Tux Kart still uses Irrlicht C++ version.

[Arthur]

Yeah, well in fact we have now forked and worked our way out of the Irrlicht codebase so while you could say it's still Irrlicht-based we essentially have our own custom renderer now. Unfortunately for your hardware we are going in the direction of having somewhat heavier requirements going forward (though you will still be able to disable a lot of the heavy stuff). Frankly, that graphics card is now 10 years old so I'd certainly think of getting an upgrade if possible if I were you.

[vlj]

For browser support we probably need to port a lot more things than the renderer. Actually I think the renderer wouldn't be that hard to port given than most of our shaders are already compatible, but you'll need to port everything else like sound engine (I think Nacl doesn't support openal), transform all load related routine to download routine and so on.
And if you go to the js route, it's even worse.

[gouessej]

Yeah, well in fact we have now forked and worked our way out of the Irrlicht codebase so while you could say it's still Irrlicht-based we essentially have our own custom renderer now. Unfortunately for your hardware we are going in the direction of having somewhat heavier requirements going forward (though you will still be able to disable a lot of the heavy stuff). Frankly, that graphics card is now 10 years old so I'd certainly think of getting an upgrade if possible if I were you.

I already tried to fix the power supply of my computer in order to be able to install a new graphics card, I upgraded it several times but they worked only a few days (overheat because the cooling system didn't get enough power). I should look at the source code but I suspect that there is a lack of optimization, the whole geometry is probably sent as is to the graphics card, maybe you use the frustum culling but it isn't enough. It's not a problem for modern graphics cards but it means that you use them not very efficiently, don't you? I assume that maintaining STK already requires a lot of time, I won't suggest you to improve the performance but I think that as my graphics card supports Open Arena without trouble, I expect that it should be similar with STK as its geometries aren't extremely complicated.

For browser support we probably need to port a lot more things than the renderer. Actually I think the renderer wouldn't be that hard to port given than most of our shaders are already compatible, but you'll need to port everything else like sound engine (I think Nacl doesn't support openal), transform all load related routine to download routine and so on.
And if you go to the js route, it's even worse.

OpenAL-Soft works with NaCl, look at the project naclports. There is Ben Vanik's WebAL but of course I imagine that the JS route would force you to rewrite the whole game.

[Arthur]

I assume that maintaining STK already requires a lot of time, I won't suggest you to improve the performance but I think that as my graphics card supports Open Arena without trouble, I expect that it should be similar with STK as its geometries aren't extremely complicated.

Well I am personally not a graphics expert - see vlj and cand for that, but arena FPS games are typically "slower" games in terms of game speed and contains lots of corridors and not as much scenery as a racing game such as STK does, so I don't think they're really comparable. I suspect you'll have equal trouble running say 0 A.D.

[Auria]

Yeah, well in fact we have now forked and worked our way out of the Irrlicht codebase so while you could say it's still Irrlicht-based we essentially have our own custom renderer now. Unfortunately for your hardware we are going in the direction of having somewhat heavier requirements going forward (though you will still be able to disable a lot of the heavy stuff). Frankly, that graphics card is now 10 years old so I'd certainly think of getting an upgrade if possible if I were you.

I already tried to fix the power supply of my computer in order to be able to install a new graphics card, I upgraded it several times but they worked only a few days (overheat because the cooling system didn't get enough power). I should look at the source code but I suspect that there is a lack of optimization, the whole geometry is probably sent as is to the graphics card, maybe you use the frustum culling but it isn't enough. It's not a problem for modern graphics cards but it means that you use them not very efficiently, don't you? I assume that maintaining STK already requires a lot of time, I won't suggest you to improve the performance but I think that as my graphics card supports Open Arena without trouble, I expect that it should be similar with STK as its geometries aren't extremely complicated.

We know STK is not optimal in performance, though there is some progress. For recently, we recently added instancing, which promises to give a good performance boost. Texture compression is also likely to come and should help. So yeah we're clearly not AAA-level optimised, but vlj is doing some pretty nice work here. Also, we also plan to leave a playable fallback with most graphical effects disabled, that should run on HD 3000 and up

Also note that OpenArena is very old school, featuring graphics from over 10 years ago, so Open Arena working well is not really a comparison point

[vlj]

Actually we're closer to Tesseract (which is the best open source game engine atm in my opinion) than open arena engine in term of visual level in some of our hd track. I can run Tesseract on low at 15 fps, exactly like stk on low on chocolate.

[gouessej]

We know STK is not optimal in performance, though there is some progress. For recently, we recently added instancing, which promises to give a good performance boost. Texture compression is also likely to come and should help. So yeah we're clearly not AAA-level optimised, but vlj is doing some pretty nice work here. Also, we also plan to leave a playable fallback with most graphical effects disabled, that should run on HD 3000 and up

I know mesh instancing, it's useful but not supported by my hardware (and it's efficient if and only if you draw the same meshes a lot of times in the scene, you can still use the instance id in the shader to make them look a bit different). I'm glad to know that you plan to leave a playable fallback for low end graphics cards and integrated chipsets. In my case, the game takes ages to run, especially when loading the karts. I'll give it another try after switching to Mageia Linux 4.

Also note that OpenArena is very old school, featuring graphics from over 10 years ago, so Open Arena working well is not really a comparison point

Actually we're closer to Tesseract (which is the best open source game engine atm in my opinion) than open arena engine in term of visual level in some of our hd track. I can run Tesseract on low at 15 fps, exactly like stk on low on chocolate.

Tesseract is a fork of Cube 2 engine, I see what you mean. I can only run Cube 1 on my machine.

Well I am personally not a graphics expert - see vlj and cand for that, but arena FPS games are typically "slower" games in terms of game speed and contains lots of corridors and not as much scenery as a racing game such as STK does, so I don't think they're really comparable. I suspect you'll have equal trouble running say 0 A.D.

You're right. Those first person shooters can use both BSPs and portals in indoor environments whereas STK has to manage larger meshes in outdoor environments. However, frustum culling and contribution culling can be used, maybe it's already the case.