[POLL] Feature Wishlist

[hiker]

So you plan to maintain both GUIs?

Yes, for now I'll leave both version in, and the user can select in the UI which one to use.

Cheers,
Joerg

[Funto]

In my opinion, STK needs:
- first and foremost, performance improvements. Is the game usually GPU-bound or CPU-bound? Multithreading could come in, but I suppose adding support for it would be a huge effort...(but well, porting to Irrlicht and Bullet are other huge efforts )
- less boring tracks, as Kinsu suggested. Take inspiration from Mario Kart and Crash Team Racing
- (maybe) better bonus items to the players that are ranked last, which can make them come to the front more easily, and maintain more competitiveness. This is what's done in Mario Kart, and although it's a bit unfair, it helps maintain the races interesting. This could be optional.
- some more bonus items (but this is not a priority). STK features original ones I think (yet I didn't understand why in hell are cakes explosive )
- a nicer theme for the GUI, but more importantly, something more "smooth": right now, everything is static. Going from one screen to another could benefit from some animation. This lack of animations also makes the choosing of tracks and the challenges screen much harder to understand (yes I know that this one is temporary).
- SSAO: this should be relatively simple to add (it's just a post-process after all) and could make the game look much nicer . Real shadow mapping could also come in, or just lightmaps if you want to avoid such computations at run-time.
- less locked items: when a new player starts STK, he really only has access to a small part of all the available content, which is really sad ^^. Plus, many people will just play a little, and I think the tracks that are available in the beginning are not the most entertaining ones. I would lock some more karts and less tracks (I think in the beginning more that 50% of the tracks are locked right?).
- the music is sometimes too "melodic": for example in the "farm" track ("Basse-cour cocorico" in French, I hope that's the one I'm talking about), the end of the song and the loop are clearly identified, and this has a negative impact on the player: it feels like you are going to "yet again" hear this song.
- a Beastie kart of course

[Auria]

I agree the GUI could be improved, but it's first time I hear that selecting tracks was difficult?

Regarding performance, it depends on the options you choose; if you enable animated characters then the bottleneck is there. When you disable that, IIRC then it's the rendering that is slow (but still can be improved by using stuff like VBOs and instancing)

Regarding loopable songs, I agree but they are difficult to find

[Funto]

Well, what I meant was not really that it was "difficult" to select tracks but rather that it doesn't feel natural because of the fact that it is too static.

And yes if you are not using VBOs, this will clearly give a gain!! But I'm quite surprised Irrlicht doesn't do this automatically...??
You should also make sure you're using GL_STATIC_DRAW everytime this is possible with the VBOs.

As for instancing I'm not sure, you would need to have many visible instances of the same model for it to be of some interest, and I don't think this is the case, is it?

Regarding music, maybe it would be possible to modify them a bit to make them more loops-friendly? Do you have the "sources" for the music tracks?

[hiker]

Well, what I meant was not really that it was "difficult" to select tracks but rather that it doesn't feel natural because of the fact that it is too static.

I don't really understand what you mean with static. I actually find the changing pictures of the GP already a bit 'distracting', but I left it this way since it does indeed show nicely what tracks are included (perhaps only scroll the GP if it is selected??)

And yes if you are not using VBOs, this will clearly give a gain!!

That might not be true in all cases. We are using Octtrees, and in some tests we have done we got more benefit from better culling than using VBOs. We can combine both, but that needs quite a bit of tuning, since VBOs need a certain amount of vertices to make a big difference. So it means finding the best size of the octtree leaves to still enable VBOs efficiently. So far not many people have complained about the performance, so at this stage performance optimisation is not high up on our todo list.

But I'm quite surprised Irrlicht doesn't do this automatically...??

It does, if the mesh buffers have a certain size (default 1500). And (iirc) if we are not using animations (which explains part of the kart problem). At this stage animations (which are done on CPU) are the biggest contributor to slow performance.

...

As for instancing I'm not sure, you would need to have many visible instances of the same model for it to be of some interest, and I don't think this is the case, is it?

It's not only the number of instances itself, it's also the reduction in scene nodes, which make the whole culling process potentially CPU bound. ATM we have too many scene nodes (each kart is like 12 scene nodes, and only some of them are children of the actual kart and would be culled implicitly if the kart is culled), each item is a separate scene node ... and irrlicht recommends not to use more than ~150 scene nodes. So using instances for items (bananas, ...) might give some benefit, just because the number of scene nodes is reduced.


Cheers,
Joerg

[Funto]

About that "static" thing, what I mean is that you don't have an animation that makes the preview of the track slide from one column to the next one. This simple thing would make the menus much clearer, especially on the challenges screen.

About the VBOs, I thought of combining those.
But I think we are not talking about the same thing: I'm talking about VBOs, which cover GL_STREAM_DRAW, GL_STATIC_DRAW and GL_DYNAMIC_DRAW, while I think you are talking about the fact of sending the data to the GPU only once (which corresponds to the behaviour wanted for GL_STATIC_DRAW). Am I right?

[hiker]

About that "static" thing, what I mean is that you don't have an animation that makes the preview of the track slide from one column to the next one. This simple thing would make the menus much clearer, especially on the challenges screen.

OK, that sounds indeed interesting. Not sure if this might make the GUI appear to be too slow (imho it sometimes is quite slow, but perhaps that's just for me on my rather old laptop).

About the VBOs, I thought of combining those.
But I think we are not talking about the same thing: I'm talking about VBOs, which cover GL_STREAM_DRAW, GL_STATIC_DRAW and GL_DYNAMIC_DRAW, while I think you are talking about the fact of sending the data to the GPU only once (which corresponds to the behaviour wanted for GL_STATIC_DRAW). Am I right?

Irrlicht supports all three drawing modes for what they call harddware buffers. I am not sure where they store the data in case of GL_DYNAMIC_DRAW etc., so far we haven't really looked at performance optimisations since not too many people have complained about it BTW, anyone with more OpenGL knowledge is welcome to help us out!

Cheers,
Joerg

[hiker]

- (maybe) better bonus items to the players that are ranked last, which can make them come to the front more easily, and maintain more competitiveness. This is what's done in Mario Kart, and although it's a bit unfair, it helps maintain the races interesting. This could be optional.

This is actually already the case. We had to soften the impact somewhat, since people complained that they only got the boring items (because they are first ). So this might actually indicate that the AI needs improvement (nothing new ).

- some more bonus items (but this is not a priority). STK features original ones I think (yet I didn't understand why in hell are cakes explosive )

Any suggestions? Esp. new suggestions? So far the plunger is probably the most original item we have, then perhaps the bomb. Perhaps open a new thread for suggestions (what about a remote control? For (say) 5 seconds you can't steer your own kart which will just go straight or be controlled by the AI, but you control the first kart? Or perhaps switch two players, i.e. those players would drive the kart of an opponent for a limited time?) Brainstorming please (in a separate thread).

- less locked items: when a new player starts STK, he really only has access to a small part of all the available content, which is really sad ^^. Plus, many people will just play a little, and I think the tracks that are available in the beginning are not the most entertaining ones. I would lock some more karts and less tracks (I think in the beginning more that 50% of the tracks are locked right?).

May people like the challenges, and while there could be better ways (trophies, career mode or so), for now that's not too bad. Esp. since we tend to lock the same tracks (due to the complexity of changing the challenges and their dependencies), so new and mostly better looking tracks tend to be unlocked from the beginning. But I agree, perhaps locking a few more karts would be better. Improving the challenges is in trac, that would be part of it.

Cheers,
Joerg

[Kinsu]

(maybe) better bonus items to the players that are ranked last, which can make them come to the front more easily, and maintain more competitiveness.

We had to soften the impact somewhat, since people complained that they only got the boring items (because they are first )

I've noticed that when I was first I often got bubble gums, and when I was last it is more zippers. Zippers are cool to get back in the race, but bubble gums are maybe a little boring in the end. I think if a new object has to be added, it should be one that is useful for the first kart (as temporary invincibility to avoid anchors, or maybe an object that enable the player to do something on other karts in exchange for a little nitro or speed... let's brainstorm in another topic, as you suggested ).

But I agree, perhaps locking a few more karts would be better.

Locking karts is not actually locking content (for the moment at least, as they all have the same physics properties), and it adds a goal for challenges : people want to achieve them to unlock the kart they want, not just to achieve them. I think it would be good to lock Beagle, Emule and Beastie with Gnu for the release, as they are those players will be the most curious about. To avoid getting players mad, karts unlocking should also be measured : one kart unlock for one fourth of challenges achieved should be fine.

[Funto]

Yup I think Kinsu is right: when locking a kart, you lock less "valuable" content. You could also just display the silhouette of the kart in the main menu, so the player has an idea of what it is and makes him even more curious about it

I persist in thinking that you should have less tracks locked in the beginning ^^.

You are mentionning a complexity in changing the challenges, maybe some simple graphical tool which would generate the XML content would help? I think making it with Qt would not be really difficult. Would it be of some valuable help?

[Auria]

I am not convinced this would help at this time, because in the future with career mode maybe the way to manage challenges will be completely changed
But, if you want to help, there are a lot of things to do

[hiker]

You are mentionning a complexity in changing the challenges, maybe some simple graphical tool which would generate the XML content would help? I think making it with Qt would not be really difficult. Would it be of some valuable help?

The problem is not the xml or so, just the dependencies between the challenges: e.g. to make sure that no GP is unlocked before all its tracks have been unlocked. I have some plans to improve the dependency system a bit - but then again Auria might want to get started on a better system anyway.

I persist in thinking that you should have less tracks locked in the beginning ^^.

We will see how we go, but locking tracks is also a got motivation for people to do the challenges. So in a way we increase the amount of gameplay people will get out of STK. Also we can (I am not saying that we do atm ) make more difficult tracks accessible later (e.g. star track is rather difficult since it's easy to fall off).

Thanks for the feedback!
Joerg

[acme_pjz]

The problem is not the xml or so, just the dependencies between the challenges: e.g. to make sure that no GP is unlocked before all its tracks have been unlocked. I have some plans to improve the dependency system a bit - but then again Auria might want to get started on a better system anyway.

I don't understand why the game needs a dependency system ... I think it's not necessary to make sure that no GP is unlocked before all its tracks have been unlocked. If a track is locked and the GP contains this track, than it means that the track can only be played in the GP ...

[hiker]

I don't understand why the game needs a dependency system ... I think it's not necessary to make sure that no GP is unlocked before all its tracks have been unlocked. If a track is locked and the GP contains this track, than it means that the track can only be played in the GP ...

It might not be necessary, but it feels better designed if this is the case: so people can work on a certain track so that it easier for them to win the GP, as opposed to having to run the GP over and over just to be able to train a certain track.

Cheers,
Joerg

[hiker]

Hi,

thanks a lot for this poll, and your votes. Quite interesting. Well, the fact that network multiplayer is high up is not that surprising, I am quite keen on implementing this as well But unfortunately this has to wait for a while (*sigh*). More interesting was that not thaaat many people considered a physics update important. But perhaps with only one vote to cast people had to prioritise.

Some points that should also have been asked:

• Do we need track improvements (i.e. what would be more important: making the existing tracks better, or getting more more tracks, even if the quality is not as good)? We are actually considering using the money donated to pay someone to make a track or two. But that should wait till the physics are more or less in the final state, so that the track designer knows what the karts can do, and so what can be used in this track. But this is also an obvious points were people can contribute. It needs some skills with blender though.
• What about game modes - should there be more? Any suggestions?
• And items? Well, the latter has already started a nice discussion in a separate thread, and hopefully the next release will already get one more item

I agree that STK has enough karts for now (considering that many more can be downloaded ... well, once they have all been ported and the addon webpage is properly setup). I guess people contribute karts since it's less work than contributing a whole track (though don't underestimate the work on a kart - e.g. the animations for the characters, ...).

We will keep on working according to our Milestones page - which means addon-manager in the next minor release (good news: this is mostly finished, we just need to get the web page fixed), then physics, then networking. 'Minor' goodies (items, game modes) will be thrown in as we have time

Thanks a lot for your opinions!
Joerg

[Kinsu]

Do we need track improvements (i.e. what would be more important: making the existing tracks better, or getting more more tracks, even if the quality is not as good)?

Maybe we should define what criteria do make a track a good quality track, this is not really obvious to me.
I am tempted to say it includes the difficulty (and alternative way possibilities), maybe also the decor details, graphic quality and of course the triangles count... I really like Canis Lupus tracks, they are pretty immersive. The bovine barnyard is also very cool.

I would say that maybe graphic quality is one of the things the current tracks need the most to be improved. Also, some acrobatic features as we said : loopings, jumps and cannons, a little as in some tracks in Trackmania (so improving tracks may also mean improving physics ), and more interactive content, as exploding barrels, floating objects on rivers (maybe on a river of some track we could make a little frogster-like game ) camels that cross the road, thwomps-like objects, Prince-of-persia-like traps (and drive on the walls !), could make the tracks much much funnier.

By the way, I really liked the lava tunnel in the 0.7 Fort Magma, I think the 0.7.1 version with a bridge offers a flatter gameplay.

We are actually considering using the money donated to pay someone to make a track or two. But that should wait till the physics are more or less in the final state, so that the track designer knows what the karts can do, and so what can be used in this track.

If this is done, I think one of the most important thing is to ask this person to provide documentation in the end about how he made the track. This will enable the community to get skills and to re-use his work in order to make good quality tracks / improve the existing ones.

What about game modes - should there be more? Any suggestions?

I'd like to play in the arena against the AI, but I think this is less important than improving tracks and the multiplayer mode. I think for the moment, additional game modes are pretty much optional content, as around 90% of the races are played in the default mode.
However, when the multiplayer arrives it could be very interesting to add game modes as CTF, "don't keep the bomb", maybe some team races (don't know what could be the rules...), etc. Then maybe port back these modes to single player, developing a good little AI ?

And items? Well, the latter has already started a nice discussion in a separate thread, and hopefully the next release will already get one more item

I vote for the punch and then the monkey mode !

I agree that STK has enough karts for now (considering that many more can be downloaded ... well, once they have all been ported and the addon webpage is properly setup). I guess people contribute karts since it's less work than contributing a whole track (though don't underestimate the work on a kart - e.g. the animations for the characters, ...).

I think so, this was my case. Maybe a page that lists and prioritizes what STK needs should be added to the wiki, so people would know making tracks is more needed than making karts for the moment ?