[SHOW] platform engine

I like making platform games... So now I'm taking a break from my rpg to work on my ultimate (=reusable) platform engine. Here's what I've got after a couple of hours of work :

http://www.crossconscious.com/platformer/

Not much game-wise, the camera movement is very basic at the moment, but you can jump, double jump, float (keep jump pressed), and jump down (press down on the gray platforms).

No vector math was used, because I suck at vector math. It uses a grid of seperate "invisible" bitmaps to check for collision, ie you can walk through white pixels, but not through black pixels, and gray pixels you can walk/jump through, but you can also stand on them. With this system, almost any shape is possible. (In the demo, the actual tiles are the same, but they don't have to be).

Comments on the overall feel are welcome, and let me know if you find any bugs.

Next up will be moving platforms and a finite statemachine system with pathfinding for enemy AI. That will be a challenge I imagine, I know how to do it in a topdown game, but for a platform game it doesn't seem so easy.

Also, anyone know where I can find a free platform tileset, with a variety of tile shapes (slopes, etc)? A rip would be ok, it's just for use during development.

so a cheating engine aka art based engine? not sure if I understand the collision checking but wouldn´t vector math be way faster with bigger maps and more complex scenes?

No it's completely tile based. You have the display-tiles, and a grid of invisible shapes (which, in this demo, would look the same as the display if you would display it). It's just a 2d array of bitmaps, and if I need to know if a point hits something, I only have to check one bitmap - divide the coordinates by 32 (or whatever) to find the right tile, and then check the bitmapdata with getPixel. So I think that's pretty performant even for big maps. I've done something similar with AS2 in a few games, and it ran smoothly, so in AS3 it won't be a problem at all.

Off course, the "physics" are more complex than that. Basically, I move my character down the amount of DY, and while one of his bottom points (I use 3 here, but two is enough if you don't have spike-shaped land) hits something, I move it back up until it doesn't hit anymore. Well, that's the jist of it, it uses a lot more logic than that (checking for walls or when bumping with the top, etc), but I can't explain the entire engine.

The principles behind ai should be the same as top-down. Just consider your level in the same way as a top-down map, but with restrictions on movement - ie gravity. If its path based, your paths move along platforms and through jump and fall points at the end of platforms.

Nice engine, btw. Looks very art-based, but thats only because my best attempts have been.

Btw, art-based is incredibly re-usable - designers can go in and change the level assets easily. If you're targeting it at other developers then disregard this point.

It's smooth, I have to say that. Very nice scrolling. 2 bits of error:

Your angle collision on slopes wrong based on TonyPA's projections of realistic slope collision. (The character's bottom center point should touch the ground, not the corner.)

Also, when jumping sideways into a platform (a black one that isn't a "cloud"), the character was able to go right through it and "glided" to the ground.

It's a start, at least.

Thanks for the comments! I've updated it a bit, upward slopes now slow you down a bit, whereas downward slopes go faster, and you glide down them just a bit.

Quote:

---

The principles behind ai should be the same as top-down. Just consider your level in the same way as a top-down map, but with restrictions on movement - ie gravity. If its path based, your paths move along platforms and through jump and fall points at the end of platforms.

---

Thanks. I was thinking along similar lines. The only problem I'm having is that if a certain enemy can't jump as high as others, hed have to (try to) find a different path sometimes. But I'm sure I'll come up with something.

Quote:

---

Btw, art-based is incredibly re-usable - designers can go in and change the level assets easily. If you're targeting it at other developers then disregard this point.

---

It uses a tileset image (or more than one per level), and levels are made with Tiled, so it'll be pretty easy to create a new game with it.

Quote:

---

It's smooth, I have to say that. Very nice scrolling.

---

It uses my new and improved scrolling engine. It works fairly well, but could still be improved.

Quote:

---

Your angle collision on slopes wrong based on TonyPA's projections of realistic slope collision. (The character's bottom center point should touch the ground, not the corner.)

---

Yeah... But I can fix that without changing the code. I use a collision tileset, and I just have to move the slopes down a bit, without changing the displayed tiles.

Quote:

---

Also, when jumping sideways into a platform (a black one that isn't a "cloud"), the character was able to go right through it and "glided" to the ground.

---

Damn, I've tested it a lot and haven't seen this happen. It's always hard to fix errors you can't reproduce... Thanks for bringing it to my attention though.

Actually I misspoke myself, it was suppose to say "a gray one that is a cloud". :p The glide error still happens, but that's when you hold UP after jumping, which I presume to be a matter of not checking for button held down after initial jump.

I get the same thing as IP. I'm not sure if it's becaus you tried to incorperate some kind o double jump feature pr something, but you may want to fix it.

Otherwise good work. Is it going to be open-source?

Viza.

EDIT: Ah, I just read your first post properly and apparently this glide 'feature' is caused from your double jump code... You should limit a double jump to occur only once, and also only when the velocity is a certain value (to stop it from occuring too early so it's noticable).

Quote:

---

You should limit a double jump to occur only once, and also only when the velocity is a certain value

---

I already do that. The gliding has nothing to do with the double jumping. It's a seperate feature, kind of like in Ratchet and Clank when you have the helicopter extension thingie. But it can be turned off easily.

Quote:

---

Is it going to be open-source?

---

I think I'll do at least one complete game with it first, but after that, chances are I will. But I wouldn't want to release it as open source before I used it myself, and then have to compete with other platformers that use my code :)

Quote:

---

Originally Posted by Fall_X

I already do that. The gliding has nothing to do with the double jumping. It's a seperate feature, kind of like in Ratchet and Clank when you have the helicopter extension thingie. But it can be turned off easily.

---

Oh, I see. That's a pretty neat idea, but you might want to make this obvious to players by representing it graphicaly.

I still think the double-jumping needs work though, because it's really hard to time it right (I can only double jump when the player is just about to start falling back down ??).

Viza.

Quote:

---

Oh, I see. That's a pretty neat idea, but you might want to make this obvious to players by representing it graphicaly.

---

Yeah, in one game I used this same feature in, flames came out of the robot's booster thingies, and in another I had a character with a cape, so it was also clear what was going on. Now off course, here we have just a placeholder block, so it's not very obvious.

Quote:

---

I still think the double-jumping needs work though, because it's really hard to time it right (I can only double jump when the player is just about to start falling back down ??).

---

Double jumping is currently limited when dy (the vertical movement) is between 5 and -5, but this is easy to change.

Thanks for the comments.

Small update : now you can climb ladders. There's still a small bug/glitch in that, but I doubt anyone will notice :) But do let me know if you come across anything odd.

That's awesome. If that's tile based you should definitely write a tutorial dude, or at least illustrate somewhere how you did the slope collision testing part. Probably one of the best tile-engines I've seen on FlashKit so far. I'm wondering if it can deal with collision sweeping yet?if an object hits the wall at high velocity can it handle that?

It works very well so far :)

That's funny, over the past few days I was working on exactely the same kind of engine (although I needed several days of work compared to your "several hours").

[EDIT]
if an object hits the wall at high velocity can it handle that?
HA! That's a common "bug" in the tile based engines - or just an intended omission to gain some speed (by removing checks).
Personnaly my engine was handling that (heck, you could even have a speed of 1'000, it was working :P ), but since the tiles are 32x32, I removed it and the max speeds are set to 32; it's enough for me, 32 is already a very fast speed.

Updated, I think the climbing is now perfect. I also added a larger climbing area in there.

Quote:

---

That's awesome. If that's tile based you should definitely write a tutorial dude, or at least illustrate somewhere how you did the slope collision testing part. Probably one of the best tile - engines I've seen on FlashKit so far.

---

Thanks for the compliment. As for the tutorial, well, it's all a bit hard to explain for me, especially since English isn't my native language, but I might give it a go. Either that, or just share the code.

Quote:

---

I'm wondering if it can deal with collision sweeping yet?if an object hits the wall at high velocity can it handle that?

---

See Marmotte's reply. I could implement it, wouldn't be too hard, but I'm not sure if it's worth it.

Quote:

---

That's funny, over the past few days I was working on exactely the same kind of engine (although I needed several days of work compared to your "several hours").

---

Damn, I guess that means you won't be interested in making a platformer together? :) The several hours is relative, I'm never sure how long I work on things like this because I do it on train rides and whenever I find some spare time. Could be 5 hours, could be 20, could be more. Plus, I already had the scrolling engine.

I like platform games but there are some things I'd like to see more often:

acceleration based running (like mario 3 / sonic)
less jumpy jumping...slow it down a bit or something

could you get the actor to face at the same angle as the slope he is on?
maybe add some fun extra stuff like wall jumping etc..

I know your engine is early, I'm eager to see how it comes along! Keep up the good work.

Quote:

---

acceleration based running (like mario 3 / sonic)

---

It is acceleration based, but it accelerates to the maximum speed very quickly - but that's all parameterized (if that's a word).

Quote:

---

less jumpy jumping...slow it down a bit or something

---

Also easy to change, just a few numbers... It depends on the game, how far you can jump, etc.

Quote:

---

could you get the actor to face at the same angle as the slope he is on?

---

Not with this engine I think... I didn't use vectors because a) I've made a couple of platformers using this technique in the past, and it always works pretty well, b) I didn't want to learn vector math and c) I want to be able to design levels without having to define them as vectors, but without vectors I think the angle would be difficult to calculate. It's something I can do without though.

Quote:

---

maybe add some fun extra stuff like wall jumping etc..

---

Yeah that was one of the thing I had planned, as well as hanging on edges (and swining around/climbing up), platforms that cave in, etc. But I think with what I have so far, I can start making a game out of it already.

Why would the angle be hard to calculate? Wouldn't it just be a collision check at the front and the back of the actor/avatar? Not sure how this tile engine is doing collision though. It would also let you do jumping at an angle when on slopes as opposed to always jumping straight up even when on a slop.

That's not possible because when you're on a slope, usually only one point would hit the floor. I could try dropping the other one(s) a bit until it/they hit the floor as well, but I'm not sure if that would work, and how performant it would be, and since it's not a feature I really want (at this point), I probably won't bother :)

Quote:

---

Originally Posted by Fall_X

Damn, I guess that means you won't be interested in making a platformer together? :)

---

Haha, are you serious?
Well, my plan was to develop another personal project (like I did with Bubble Islands), and it's either a fighting game (Street Fighter-like) or an adventure-platform game (like 2D Metroid, an open world that you can explore fully only if you get the proper items). Or another game, I have soo many more original game ideas that I still never tried.
Anyway, since I'm really busy with works, I don't have that much time to work on those, but maybe in 4 years something will be released :P

What's your platform game is about?

Also, I don't think that calculating the angle of the tile is that hard...? Since you probably stored the slope values of each tile somewhere...
In my engine, the slope only affects the player to it's center (like IP said: "The character's bottom center point should touch the ground, not the corner"). So it's quite easy to detect the angle directly (even though I'm not interested in that). In that case, there are several options I guess:

1) Assign to your tile an angle value, so you don't need to calculate it in real-time.
2) Calculate it in real-time, for example check the values at your left/right extremity, and make a medium, then calculate the angle thanks to that

(oh btw, I'm crap with maths :) )

Being quoted by Marmotte - very awesome. :p

Would it not be easier to simply store appropriate rotation values in the tiles themselves? He said himself he's not using vectors for collision, so instead of comparing points, you could just rotate the clip to the rotation angle specified by the tile data?

Yeah, except that the collisiontiles are just a big bitmap (tileset), and I don't store anything else. But I could try precalculating this or something... if I ever decide my engine needs characters that rotate with the slopes... But that would also exclude more complex shaped tiles (right now about any shape is possible, as long as it's big enough).

Marmotte: wasn't being serious, though I wouldn't mind doing a game together some day... I'll send you a pm/talk to you on msn with more info on my platformer idea... I think it's wicked cool and very original, so I don't want to talk about it too publicly yet.

Quote:

---

Originally Posted by ImprisonedPride

Being quoted by Marmotte - very awesome. :p

---

Here we go again, then :)

Quote:

---

Originally Posted by ImprisonedPride

Would it not be easier to simply store appropriate rotation values in the tiles themselves?

---

That's what I said as option #1 (or, what I WANTED to say ;) )
(And that's another quote with you)

Fall_X... bitmap collision? Oh yes, I understand now... I've re-read one of your old message, you are using bitmap detection for sloppy tiles... Is that also called "tile-based", then?
In my case, I have an array that stores the data of the irregular tiles. I'm also able to have any possible shape, since the array stores for each "row": how many blocks (or free tiles) you have in the 32 rows of the tiles (if it's 32x32).

I don't know if it's the best way, but it works... That way I was also able to assign many other things to those irregular tiles (ex: it affects your walking speed, when the sprite doesn't move he's sliding ...etc). I took days and days to achieve that, because it's still a bit abstract for me (I was used to work with art-based games years ago)... But I think it works well so far :)

And yeah, you can try to contact me. But beware, I'm always busy for at least 1 thing ;) (that doesn't mean I'll not listen to you)

Quote:

---

Fall_X... bitmap collision? Oh yes, I understand now... I've re-read one of your old message, you are using bitmap detection for sloppy tiles...

---

For all tiles actually. My platform code doesn't know what kind of tile it's on. It just asks my world for a few points if there is something there... and if there is, it moves back the way it came until there is no hit.

Quote:

---

Is that also called "tile-based", then?

---

Well, my scrolling engine is tile-based, and the collision detection is also done using a seperate tileset with 32x32 tiles in it, so I wouldn't know what else to call it than tile-based...

Quote:

---

In my case, I have an array that stores the data of the irregular tiles. I'm also able to have any possible shape, since the array stores for each "row": how many blocks (or free tiles) you have in the 32 rows of the tiles (if it's 32x32).

---

Not sure if I'm following... but if it works as I think it does, it's a similar system, except that you use an array of 32 numbers per tile, and I use a 32x32 image. You have the benefit that you can add other stuff (angle, type of surface, etc) in that array, but I could always extend my system so that I store these extra properties per tile somewhere...

Have you implemented climbing in your engine? For me it was pretty tricky to do, more of a challenge than I thought it would be.

And can I see it somewhere? :)

Quote:

---

(...) except that you use an array of 32 numbers per tile, and I use a 32x32 image.

---

Yes! For the sloppy / irregular tiles, I have an array with 32 slots. And each slot informs me how many free vertical pixels you have from the top of that tile.
And yes, you can extend your system too :) You would probably need another array/grid that is holding the extra informations of the map.

Quote:

---

Have you implemented climbing in your engine? For me it was pretty tricky to do, more of a challenge than I thought it would be.

---

Not in my current engine, but I already wrote ladders with my several of my "normal" tile based games ("normal" tile based game = without any slope). Hmmm, I don't remember if it was that tricky, but I'll see when I'll have some time :)

Quote:

---

And can I see it somewhere? :)

---

Haha, not yet, last time I worked on it, I decided to rewrite/optimize several parts, so it's not working as great as before right now :)

Quote:

---

You would probably need another array/grid that is holding the extra informations of the map.

---

Well, it can be done with the same grid because I use objects that contain the bitmaps, so I can have extra properties that can be read from xml or something...

Quote:

---

Hmmm, I don't remember if it was that tricky

---

Well, the climbing isn't that tricky itself... but I now have two types of climb-tiles, one for ladders that you can also walk on without falling through them (as in the demo), and also a type for larger climbing areas, where you can't walk on (not in the demo yet). And I also support different keys for climbing and jumping, but they can also be the same (ie up could be jump and climb, but space could be jump and up could be climb), and the down key is used to jump through platforms, but also for climbing down, but again it could be double-press for falling through platforms, etc, so it took me a long time (=4 hours or something) to get all this working correctly, taking into account all possibilities. I'm happy with it now though, the code is quite complex but still readable and performant, and works a charm.

Updated with a nice 3D parallax/mode7-ish effect in the background (which glitches when the camera's x position is negative, but this won't happen in the actual game)... Graphics are still temporary, off course.

Cool effect. How does mode7 work anyways? Is it pretty complicated?

Well, this doesn't work the same as normal mode7 engines, because there's no rotation involved. But what is the same is that you basically have a number of independant horizontal lines (in this case thay're 4 pixels high to save some speed), and you draw into them.

(modern) Mode7 engines in flash usually use beginbitmapfill with more than 100 of sprites. You can find some examples in the forums here.

My effect just uses copypixels into one bitmap's bitmapdata to increase speed, which I can get away with because I don't have to rotate parts independantly. It's basically the same as parallax scrolling, you move the different parts at different speeds. I also did not want to scale in realtime, so my image is skewed in photoshop, ie the top part of the image is smaller than the bottom, because the top parts will scroll slower.

I can give you the code if you'd like, but it needs the rest of my scrolling engine (which you already have), and it's also a bit messy due to heavy experimentation to get it working.