TIGSource Forums

So I've been working for a while on a number of small prototypes, the latest one aiming to be a top-down action game with random generated levels.

Doesn't sound like a particularly new concept but I've got some ideas which will hopefully introduce a couple of interesting twists (based around looting and maybe stealth).

Until two weeks ago I was working together with Michael Hussinger (who goes under the name "Miguelito" in these forums, and posts his work at http://hussinger.blogspot.com (http://hussinger.blogspot.com)).  Anyhow, Michael decided that the scope of the project was unmanageable so we're no longer working together on this, although he was nice enough to let me use his work as temporary placeholders.

Here's the prototype that Michael and I worked on:
(http://dl.dropbox.com/u/22814947/proto3/caps/piggy.png)

Youtube gameplay video
http://youtu.be/1NNeFV9oquo

Playable build
http://goo.gl/2c9W7
(requires unity3d web player)

What's next?
Right now I'm spending my time experimenting with different approaches for level generation.  This is crucial for the gameplay and even the graphics, so I want to focus on this aspect for now.  I'm quite pleased with the results so far so I´ll get round to writing about that here in the next couple of days.  Also, finding a useful name for the game ::)

Gameplay is really smooth and interested to see what you come up with in regards to looting and stealth.

Had one issue when trying the prototype that I found an enemy who was stuck in the wall. I was also able to almost overrun the camera which didn't allow me to see what was up ahead.

yeah there are a number of issues in there.  it's only a prototype so I won't get into any sort of detail until the basic progression is there.  thanks for taking the time to try it.

I love the look of this. The pig looks so cool and I love how he got a red face while swinging his bag. All these small details really make the character more charming. I can't wait to see how this will progress!

Man, the little character looks soooo cool! Love the cartoon style.
The concept sound really fun. The deflecting hability the pig has with his bag full of something, interesting.

Looking forward to see more updates from this project.

yeah the pig is super cute!!
like I said Michael is definitely a talented guy.

however, for the final game I think it´ll be better to go 8-bit, so as to reduce the workload.  oh well, still a lot of work left before getting into any graphics so just rambling.

by the way, controls are heavily based on pocky and rocky, which in my view is awesome.
http://www.youtube.com/watch?feature=player_detailpage&v=02qEz-qn8SE#t=142s

Ah! That explains why I feels the control is really familiar! I've just played Pocky and Rocky last month or so and was like 'Man, they don't make games like this anymore.' I even went on youtube and search for the new Pocky and Rocky game and was disappointed with the latest one. Glad to see someone work on this genre again!  :-*

Too bad that the pig has to go though... But I understand that 8-bit would reduce a lot of workload than going full hd for the whole game. Anyway, looking forward to this!

yeah man pocky and rocky is a real gem, still as enjoyable as ever.

by the way I was reading puipui's devlog the other day, thought the characters were absolutely lovely, so best of luck with that!

I love pock & rocky, so this is great! The similarities are a bit too much I think though, it's essentialy a clone. Throwing stuff, sliding (rolling in this case) and deflecting projectiles. That 100% pocky & rocky. But I'm aware that this is in a very early stage.

The pig is incredibly awesome, would be such a shame if you would remove it.

IMO you should switch Z and X, so Z is fire and X is melee.  Makes more sense to me that way.

yeah I agree.
however I'm pretty sure the game will feel totally different with new mechanics in place.
these will possibly include looting and maybe stealth, so I don't think it's a cause for concern.

still, I think making a pocky and rocky clone is something worth pursuing these days in and of itself, particularly since I don't think natsume will be making this type of game anytime soon.


yeah I know, however this is Michael's creation so at some point the pig is going to have to go  :concerned:

So, errr... Someone rang?

Hi Señor and I'm glad you're showing this around.
Yes, Señor Barborito is right in that it was me who kinda told the project off - the pig character is sort of in the limbo right now, but after creating all these animations the task of equipping a full, sprawling game with that kind of assets looked hugely daunting to me, expecially since I'm not exactly a wizard at doing backgrounds...

I'm glad people seem to like the guy though, many thanks for this. :)

So Señor, you've made a great build and I'm very happy to have this tiny bit of playable content.
But you mentioned going more lo-fi... That does sound interesting, if I may say so...... :)

In any case, I wish you all the best! I'll be watching the project no matter what happens!

Hi Miguelito,

glad to see you dropped by.
as you can see your character is already a great success :)

but yeah, I took good note of what you mentioned so chances are this will end up being pixel art of some sort.  but that's still far away because I'm not going to even think about that until gameplay is finished.

also, feel free to drop your criticism around here as the game evolves, it will be most welcome.

random generation

So, as I've said before the game will have random generation levels.  I've been working on that stuff for a couple weeks and I think it's looking good.

first I tried using herringbone tiles:
http://nothings.org/gamedev/herringbone/ (http://nothings.org/gamedev/herringbone/)

the results are kinda neat for caves but they lack in structure, ie. they are too random:

(http://dl.dropbox.com/u/22814947/proto3/caps/rivers_and_bridges_t.png) (http://dl.dropbox.com/u/22814947/proto3/caps/rivers_and_bridges.png)

this is no good because I'd like the game to take place inside some sort of real-world building, not a cave.  I was thinking maybe a museum or a haunted mansion for example.  but in any case, something with rooms.

so I started writing something which would generate rooms of different sizes but which would still have some structure to it.  here are some results:

(http://dl.dropbox.com/u/22814947/proto3/caps/GeneratedTexture0_t.png) (http://dl.dropbox.com/u/22814947/proto3/caps/GeneratedTexture0.png) (http://dl.dropbox.com/u/22814947/proto3/caps/GeneratedTexture6_t.png) (http://dl.dropbox.com/u/22814947/proto3/caps/GeneratedTexture6.png) (http://dl.dropbox.com/u/22814947/proto3/caps/GeneratedTexture15_t.png) (http://dl.dropbox.com/u/22814947/proto3/caps/GeneratedTexture15.png)

the thumbnails look kinda samey but if you open the images I think you'll see that the technique can generate a good variety of room sizes while still being beleivable, ie. it's not just a load of random interconnected rectangles.

the only problem is that at the moment it doesn't generate corridors.  I'm not sure if I can just do without corridors or if I´ll have to come up with a different approach.

next step, doors.

There's always SLIGE .. http://www.doomworld.com/slige/

looks interesting, thanks

improvements
-filled the rooms
-added room connectivity

(http://dl.dropbox.com/u/22814947/proto3/caps/2/GeneratedTexture0_t.png) (http://dl.dropbox.com/u/22814947/proto3/caps/2/GeneratedTexture0.png) (http://dl.dropbox.com/u/22814947/proto3/caps/2/GeneratedTexture4_t.png) (http://dl.dropbox.com/u/22814947/proto3/caps/2/GeneratedTexture4.png)

these images need to be zoomed in something like 2000% to get a feel for what it´ll look like in-game when the map gets transformed into tiles.  otherwise it just looks unplayable.

the most obvious addition is that once all the rooms are placed, they're filled with premade blocks.  a block is simply a square of 20x20 tiles with some edge connectivity info so the system can choose which block to place next out of a set of blocks which "connect" with the current one.

the idea here is to eventually have different types of room, where each type places emphasis on an a different mechanic of the game.  for example one room type might emphasize shooting, the other exploration and the last one stealth, for example.  of course room types must be assigned in some meaningful way, otherwise it´ll just end up feeling like a mish-mash.


next
-put a character in there with some basic controls
-put some enemies in there with some basic AI
-make blocks for different types of rooms
-more complex connectivity between rooms (currently all rooms are connected to each other)

in case anyone reading this is familiar with the game design aspect of having procedurally-generated levels
tips, feedback, ideas will be much appreciated since I´m a bit lost at the moment!

been working a bit more on the procedural generation of the levels, so I made a video to show how the levels feel in the game.

there's no actual game going on yet, but I'm quite happy to see that the levels are already better than what I'd be able to produce by hand  :beg:

so here's the video!

(http://dl.dropbox.com/u/22814947/proto3/caps/2/videoThumb0.png) (http://www.youtube.com/watch?v=8x3yS5WR54c)

again, any experienced PG game designers out there, advice much appreciated  :gentleman:

Just my opinion, and not as a designer either, but...

Looking at that video, it seems immediately frustrating when the player spots a dead end and has to turn back. Same for the smaller openings with nothing in them.
This is probably future talk, but you might perhaps want to fill every last dead end with something - either make a key enemy spawn there, put a collectible there, or even a secret invisible bonus. It's what's done in every game ever (long, boring walk = reward).

I'd also watch the width of corridors/openings etc - at times, there are very narrow passages to squeeze through, some of them looking just like corners with a tile missing. You might want to check and try to avoid situations where openings are obviously made too narrow by the random generation - it spoils the illusion a bit...

I like this.
however I can't really "detect" corridors with the way that the generation works currently. but I will definitely look into this when the game goes into production.  after the prototyping stage I´ll be starting from scratch anyway, so I'll be able to support long corridors with a reward at the end.  or maybe an unexpected ambush now and again with a rare drop :)



well spotted!
will fix that.

thanks!

I just want to point out that with practice, you can easily make maps that are more interesting than anything a procedural generator could pump out.  There's a reason certian games are revered for their amazing level design, and it isn't because they had the best procedural level generator.

sure, I totally agree.

but at least for now, I think that as a programmer, this is the approach that suits me best.  I should probably move on at some point, but right now I quite like changing the generation blocks/code to see what works and doesn't work in the game.  this allows me to test different strategies and it's a lot faster for me than sitting and placing 500 tiles.

but yeah, no matter what I do the levels feel a bit samey.

Well, do what a painter would do: Break up your strokes!
If you keep working with the same type of wall/corridor/corner etc., the level will certainly end up looking a bit uniform. How about inserting some completely different tiles? One completely empty, one with a Bomberman-style block grid, a spiral, meanders...
Try thinking in terms of what will actually happen there: Place some Space Invaders-style cover around a large open area, or put a single enclosed arena with one entrance only.

Then, you could clump certain types of blocks together instead of distributing them equally, imitating the feel of buildings (long walls) standing next to a forest (dotted obstacles) and to a desert (practically nothing).

Would make the whole scene more organic, I'd think...

I think it's clear by now that my approach doesn't allow enough control over the map and hence will not be able to provide enough "interesting" situations for the player.

For the final thing I'm pretty sure I'll need to be able to design each room by hand, so I can create complex situations like the ones you suggest.  still, there´ll be plenty of random elements in each room and they will be assembled randomly as well, so it should feel different every time you play.

This is probably the way to go, but for now I´ll continue with the current approach by incorporating different room types as well as enemies, so I can get a more global idea of the problem rather than just working on the room generation side.

In fact I've been working on different enemy types for the last couple of days.  Pretty basic stuff but I'm still looking forward to putting them in the game :)