Plague - 3.28

I would love to try this out, but your archives aren't unix-friendly. Debian file-roller doesn't interpret the "\" as "/", so I end up with a bunch of files in the main directory with names like "res\sound\rip.ogg"(as literal strings, unix doesn't interpret the \ as a directory) resulting in the game not finding the files.

im not planning on having buildings that train units as a in a traditional RTS - the plan is to have buildings like a tower defence game, and units that you find (so until you find more units you have a fixed number which makes them more valuable).

It looks alot like warcraft 2. That would be a good feeling to have in the game. Able to build buildings and select them to train troops and stuff. I think it looks good. I like the 8 bit feeling. Makes me think of the original Nintendo. Keep up the good work.

For A* - you could try caching paths.
Naturally, it probably wouldn't be necessary to cache every grid against every other, but you could, say, cache every 10 grids, then you just have to do A* check for -10 to +10 grid squares til you find the closest path, and then a check for the destination.

¿É°®µÄpython

thanks for the __debug__ hint

Another hint: put output in a 'if __debug__:' block so if you run it with 'python -O main.py' you wont see (and it is faster) the output

The mem leak still exists. Maybe you a reference is kept around? Or loaded twice? It consumes about 300 mb per game you start. If I die and press 'esc' and start a new game, another 300mb are allocated. Besides of that it works great.

Hi Rene,

Thanks for pointing that out, if thats the mem leak it would've taken me a long time to find.

I'd be very grateful if you could make a patch for it - I'm still busy looking for a new car (as the old one is almost dead :) ) and coding is taking second priority, so its unlikely that I'll get enough work done to release something by end of the weekend.

hi, I think the mem-leak is in the rotate/scale cache code I sent you... so maybe disable for now.

It needs to have an upper limit on the cache size. Also because I think you reload images between levels each time you reload it adds more cached versions for each one.

I'll send you a patch at some point... probably on the weekend... if you don't do it first.

cu.

I use ''' some text ''' as doc strings (I think this is the standard?). If I've done that anywhere else its because I wrote that code EARLY on (and havent switched to # which is what I do when I see it).

Are there any profilers/tools to get GC stats/profiles (eg: like cProfile)?

Just another sidenote: why do you use """ some text""" in your code instead of a comment? I think the interpreter makes a string out of this and so it slows things a bit down (just guessed, not profiled).
Yeah memory leaks are hard to find, maybe it is not at all your fault. It could be pygame, numpy or any other lib.

Awww man, not another memory leak. Those are usually so hard to track down and kill. Thanks for the warning, I didn't realise I had one. Glad you were enjoying Plague up to that point.

Yesterday trying it out it crashed due to a MemoryException (I have 4GB ram!). I guess there is a memory leak somewhere. I died approximately 20 times and restarted the game. After around this 20 times it crashed. I guess some memory is not deallocated properly. Otherwise it is a cool game.

Sounds awesome, np

That IS a simple solution. I like it. I think I have been programming my game for too long, and not playing enough other games... *click*.

Now that my brain is switched on, I remember I wanted to use the Chronicles of Riddick health system (since I think it's brilliant). I'm thinking 3 (or n depending on how you level up your character when I get that done) health bars, health recharges in your current health bar, medikits replenish fully depleted health bars, and when a health bar is depleted you become invincible for a short amount of time (and drop into the next health bar).

Thanks for bringing this up PyMike!

Simple solution:

You get 3 (or whatever) lives. When you die, you simply respawn in the same location without changing the map at all. (Remember to give the player a short invincibility period). When you lose all your lives, reveal the map so you can watch the humans battle it out.

Ah, I see that what I have just written may also be unclear. When you die you can respawn, by adding another "player" to the map. You do this right now via "test mode". This wont work while you are alive because there can only be one player.

PyMike: I should add that I'm personally thrilled to read that you want to keep playing when you die. Plague must be heading along the right lines. :)

Rene: I should add to that that I'm even more thrilled when someone looks through the code and improves on something. Thanks!

Plague doesnt support "repawning" at the moment (except via test mode as I detail below, which is what I do when I want to keep playing). I think Sean is planning on adding a respawn key.

The reason why I haven't implemented it is because I reveal the map to the player when they die, and they can then see all the weapons, hives, etc - I think a good amount of the fun of Plague is exploration, and this ruins it.

I have previously toyed with the idea of dropping you into a remaining human at random (when you die), but I guess I quite enjoy watching the remaining humans try to battle it out. I'd also have to do some work (which would be quite involved with bug testing) to draw the sims memory when you possess them (otherwise you wouldnt see what they had memorised).

Thoughts?

I mean when you die, can you respawn? It's annoying having to wait for the map to regenerate every time you die ^_^

Rene: Cool, thanks, I'll review and incorporate those changes today. Those are some good suggestions as well, and they all look fairly straight forward. Thanks again.

PyMike: Press "h" to enter "test mode". You can then use the arrow keys to change item (left, right), create item where the cursor is (up), delete item under cursor (down). There can only be one player at any time, but you can make an unlimited number of anything else. Press h again to leave "test mode". The item you will create is displayed in text in the lower right.

Yes this runs much faster, I'm getting a steady 30fps. Is there a way to respawn in-game?

The new version is very speedy :)

I've started on some optimisations... and I've made it quite a lot quicker!!!


http://rene.f0o.com/~rene/stuff/resources.py
- loading is faster, and also can now specify types of images.
- depending on the type of image, it can be drawn faster.
- a big one is the background image using .convert()

http://rene.f0o.com/~rene/stuff/simobject.py
- cache scales, and rotates.


Pre-computation is what you want to do. So once you scale, or rotate things save the value. I've added some scale/rotate caching stuff.

For the zoomed out view... why not save that as a surface so you only need to update the changed areas?

For the font drawing of the status updates... why update every frame? How about once every second? Save the rendered font if you do not change it.

Alternate the updates between frames... don't update everything every frame.
eg.
frame 0 - update fps.
frame 5 - update score
frame 10 - update speed.
... etc.
frame 0 - update fps.
... etc.


Try changing some of your images with no transparency to use .convert() rather than .convert_alpha() Especially those that do not have any alpha parts in them.

Most of your sprites look fairly low color. So RLE encoding should speed them up. Rather than convert_alpha(), use color key transparency: sprite_surface.set_colorkey([0xFF, 0xFF, 0xFF], SRCCOLORKEY|RLEACCEL)


re: opengl. For non-opengl computers, porting to opengl will be *way* slower. That's about 30% or so of computers.

Best to ask on the mailing list for help optimizing it -- you'll get a lot of help.

Hi PyMike,

The random fps drop spikes are gone (as near as I could tell). They were caused by large number sof physics calcs due to bullets reflecting through solid materials (ie: between two parallel planes of rock with no gap in between).

How hard would it be to port to something like Pyglet, and what sort of performance increase do you think Plague would "generically" get (given that drawing is the #1 CPU time hog)?

Hi PyMike,

I agree the random fps spikes are annoying (on my fast PCs). I put too much into the AI over the last few weeks and now its performing horribly. I guess keep an eye out for a release that says I've optimised the AI and have another go then (thankfully I know what I've done and how I want to optimise it). :)

Unfortunately, I found the game to run too slow for me to enjoy it, especially with random framerate spikes (5-30 fps randomly), which really mess up my timing.

And yes, my computer's plenty fast =)

Wow this game is one of the best pygames I've ever played... could you host the file here too?: http://marislabs.org/ext/apps/

hey Skizzaltix and Kasperi, I've sped up the map loading, shaved 2.5 seconds off it.

Hi Sonolin,

Thanks for the feedback. Plague does need a lot of "balancing" - for the current build I've concentrated on fun/easyness rather than realism or challenge (Plague went through a "challenging" phase and all the feedback was it was too hard - I learnt from that that balancing is hard!). I also think the SMG is too weak, the Shotgun isnt working right, the Minigun is also too powerful - pretty much the only gun I'm happy with is the Assault Rifle.

If you would like to get involved with balancing have a look at "simdata.py" and let me know (preferably via the blog). Pretty much all balancing stuff is in there and it should be straight forward. Otherwise I'll get around to it eventually, but right now I want to concentrate on getting a "city builder" map algorithm - I'm bored of wandering around forests and rocks.

I'd love it if you had a look at fire algorithm ("cell.py"). Michael Lange has hooked NumPy into it, and a had a brief peruse of the changes, but I'm not a NumPy expert atm, so I havent made any modifications.

Oh and also - I think limiting the amount of health/ammo upgrades would benefit the game. It is pretty silly when I can just run around, let my companions pick up all weapons, and shoot bushes engulfing my enemies in flame, then run through the flame, and stand on the guaranteed health pad on the other side.

I did not hit much ammo problem with my laser of course ;)

Wow I really must say I love where you are going and what you are doing with this game - it really is a great game. I gave it a 3/5, however, due to some things that need to be worked on (but can definitely be fixed!). Some of the things that stood out to me over everything:

- Initially, I stopped playing the game after about 10 minutes due to the quite bad framerates on my (decent) system (nothing to brag about, really, but I thought the framerates were quite bad). Simply because I loved the idea (and remembered how fun Gauntlet for N64 was :)) I tried again today with psyco installed. You REALLY should say "Psyco is required for any decent speed" not "Psyco is recommended for a decent framerate increase) ;) - the result was awesome!

- Adding upon above, after psyco, I realized the only significantly slow part was the rendering of fire. I will take a look into that part of your code and see if I can be of any help to speed that up but no guarantees ;).

- Lastly, certain weapons are DEFINITELY overpowered. I have not played much, but I really recommend you to rethink/take out at least the Sniper Rifle and Laser. Both of these could EASILY be fixed (with realistic fixes).

1) Sniper rifle: add some sort of larger delay after each shot, to represent you inserting a new round into the chamber. Possibly take out reloading. This will make it so you cannot go around unloading with a SNIPER RIFLE on the enemies ;).

2) Laser: limit the ammo a bit - possibly at 100 - so it does not keep increasing. My last game, I simply ran around with a laser 100% of the time until I got bored and then quit.. not really much challenge.

Now, I understand it gets harder (I only was on the 4th level), but it was pretty simple for me to just waste things with the laser and odd grenade/sniper shot to mix things up. Also, IMO I think you should increase the difficulty towards the start of a game a notch or three... some may be turned off by this initial easy-ness (me included).

GREAT game, and I believe that much success can come out of it :). Oh, and to a minor note, the interface needs a bit of improving - it took me a while just to figure out that my list of guns were in the top left LOL! Add some gun icons or something, and put them in easier view (I have a large monitor, so this might contribute).

thanks for the feedback guys,

making the map is kind of slow - the slow part is an algorithm that partitions the world to find "connected" spaces, so I dont add zombies, guns or survivors to "blocked off" regions.

I can rewrite that algorithm to make it a little faster, but since map generation is procedural and various checks have to be made to make sure the map is "playable", its always going to take some time.

if you think Plague is too easy, keep playing for a few levels. I havent gotten past level 5. :) (Although it might be a bit easier now that I've made squads, weapon sharing voluntary, etc).

Very good game! Only bad thing was it loads slowly. But very well done. I liked it!

Cool game! I like the variety of weapons/etc.

Though, it loads pretty slowly--Whatever happens right after it loads the rocks takes forever on my computer (MacBook running OS 10.5), which is pretty fast.
Also, I found it was pretty easy... Maybe a few more zombies and a few less ammo/health pickups?
Anyway, great job!

To change the screen res, look in constants.py. It's (1024, 768) atm. (800,600) should work well for you.

Things like the help screen wont fit onto a screen though, same with title page. I dont resize fonts to fit onto the screen (but I should...)

no probs with a triple post. :) My brain works like that too.

Sorry for the triple post - is there a way to change the resolution? The screen is simply too big for me (technically it is as big as my screen, but the window decorations and the window bar on the bottom make me not see stuff :S)

Thanks

Ahh my bad on the gauss rifle - I had it confused with the mini-gun :S
Looks like it was LOL

Ahh, I thought those were intentional (why wouldn't you find loaded guns lying around an zombie infested area? why would you find guns at all if it is illogical?)
But yeah, I was thinking it would make sense if you simply refilled ammo for the currently selected weapon only, and you have to cycle to load them all.

I will look at that file you suggested here in a bit (I would devote more to it, I really love the project, but I have too many of my own :( )

I haven't done anything to the fire, so thats odd (and good) that its sped up. Maybe I've just lightened the load elsewhere.

Balancing wise, it is too easy. I'm going to add "difficulty", hopefully in interesting ways (not just smarter healthier zombies - I find that boring), but I need to build the game up first to support that first.

Leaders and packs are definitely doable - I designed the code last week, but havent released it as I was concentrating on inventory. I'll try to whip that up one night this week and release straight away.

I've started revamping the phosphorus grenade to make it different to molotov (molotovs burn longer but cooler).


I'll spend a little work on the rebalancing the guns to differentiate them, or if you want to edit simdata.py and pastebin it on the plague blog (with detailed notes on your changes and why) I'd be quite likely to incorporate your changes. I can only do one of the following at anytime: designing, programming, art, sound, testing, balancing, etc. As such, any help is massively appreciated.

I agree with you, Concussion grenades have been bothering me for a few weeks now, and I always use molotovs since they are the best, but I havent been bothered enough to fix them (as this is a balancing task, and I have lots of tasks). I'll rebalance these a little this week.

Just thinking about difficulty, I've had some thoughts while writing this which wont require much work. I'll use the score (and keep it building between games) and drive difficulty from that so successive games you live through build the difficulty. Quick and dirty requiring no work, so should have something in short order.

No bugs? I've noticed that guns "reload" as soon as you pick them up, and then instantly gets some clip_ammo when you pick up ammo. I'd say thats a bug. Any comments on things along those lines where stuff doesnt work "right" are welcome.

Well, I liked the new version (especially the shotgun) and the fire's definitely were not so slow this time (I had three molotov cocktails going once and the game was still playable :) )

As far as balancing - this version if probably a little too easy - perhaps a difficulty setting or something? Because all the other humans were dead right away, and I just ran around and killed everything pretty quick. Perhaps have fewer health places - and try and put zombies in bigger packs that act as a pack and attack together - and perhaps have some of the other humans follow your lead - if possible.
But definitely have a difficulty...

As far as weapon balancing - I don't note much difference between the gauss and the sniper rifle - most of the time I just ended up using the gauss as an extra ammo sniper rifle anyway :S
I didn't find the laser anywhere - or my favorite weapon the chain gun thing (sorry, kinda early, but I got the impression that that was the gauss, and you changed it...) and the rest seem ok - though I don't have a clue what the difference is between incendiary and molotov cocktails (except the cocktails run longer) and I find the concussion close to useless (why blow them back for no damage when you can just kill them?)

Perhaps add more kinds of zombies that are a little smarter, maybe some that can make new hive things when they get destroyed, etc.
Perhaps even add new kinds of badguys besides the traditional zombies - I was thinking vampires or werewolves would fit the theme, and actually make a use for having several kinds of weapons, etc.

Cheers, and good game :)

Hi Roebros,

Thanks for the feedback. Yeah, It's not obvious to me how to speed up the fire simulations - it doesnt seem to me that NUMPY fits well (although I would've thought it would). I'll keep looking into that.

I'm glad to hear you are enjoying Plague. Winning will be a bit easier in this release as I've made the zombies a bit slower and have less health - I now find it more fun. Still trying to balance the weapons/zombies. If you have any balancing suggestions, let me know.

The FPS do get pretty slow when you throw one (or more) grenades - but otherwise quite nice - I love these kinds of games...

Spent 2 1/2 hours playing last night and won like 4 times - honestly though - unless you start in a bad position it is pretty easy to win every time - just got to make good use of the molotov cocktails and get to the minigun or the sniper rifle...
Perhaps make the zombies charge or run out of ammo depos or something...

Hi Saluk,

Line of sight is expensive - thats why you (and everything else) has a small LOS radius. Try pressing T (once). That sets an optimisation option where the game tries to pick the smallest set of "blocking" tiles possible. This leads to artifacts, but should speed up the game by a fair amount.

The artifacts are being able to see things that should be blocked, but arent due to a blocking tile being discarded (because it itself was blocked). This means you can see stuff you shouldnt, and so can AI sims. Its a speed accuracy tradeoff you can make.

Glad to hear it was fun. I'm still planning on optimisations in the short term to bring up the game speed, but after that I'm going to concentrate on making it even more fun :)

The framerate got a little better with this release. Besides the fires, I think line of sight is the worst culprit. I get 15 to 20 fps if there are no bushes around, even with many zombies, but with few zombies and many bushes the speed goes very low.

I managed to play a complete game though, it was a lot of fun!

I wouldn't say it's worth incorporating, it's just an extra gun with invisible burning bullets.

Wooo, what's that I see? A flamethrower? Would you like it incorporated into the next release of Plague? If so, please paste the code into something like pastebin and I'll review the changes.

http://i55.photobucket.com/albums/g155/NathanWoodward/flamer.jpg

Haha. Kill it with fire.

They seem to be fine.

I think your framerate may have been low because there was a cluster of zombies trying to get past each other. Your fire bombs cleaned them out stopping all the collision checks and calculations (just a guess though).

Sean Lange made the new icons, and I'm very thankful to him. They are much nicer sprite art than the stuff I produced. I think he's planning on producing more sprite art, included some animations, so Plague might get a whole lot better looking in the near future.

Any issues with bullets now?

That was odd, my framerate was averaging about 15FPS, but then I lobbed a few firebombs and it went all the way up to 30. I killed a few zombies, but they soon maxed themselves out again, and the framerate stayed ~30.

I like the new icons, and the way you can no longer heal up while running full tilt.

Bouncing bullets were causing performance problems, but that was before the incident with the grenades.

Hi Saluk,

Thanks for the feedback and constructive criticism. I'm working on the speed - lots of code rewriting going on atm to make it more extensible and faster.

You can outrun Zombies really easily - movement speed is based on direction (so look forward) and hold SHIFT to sprint- you cant sprint indefinitely, but at the moment the game is too easy once you know all the tricks of playing.

I guess my challenge from your comments is to let you know what you can do in the game :). Should I add another "instructions page" or something like that? The alternative is to have a tutorial(?) - might be necessary since there are a lot of non-standard features in Plague. I'd like to add comments while the map is building but that involves a rewrite of the map building algorithm(I guess I have to do it though).

Pretty impressive game, but still far too slow to be fun. Ignoring technical issues: The player should start with a weapon (it was really bad when my teammates keep picking up the weapons before I can get one!) and the starting weapon should preferably not have a muzzle size. Many of the games I play I end up with the sniper or automatic rifle, and am unable to fight off the zombies that are too close. Since I can't outrun them, this feels unfair.

I have a Pentium Dual Core (2GB), so I don't think it's my processor.

That was quick, I'll have a look in a bit.

Hi Nathan,

Thanks for the feedback on v3.03. Can you have a go at v3.04 and let me know your thoughts?

Much improved - it's still a bit choppy but for the most part the FPS stays in the high twenties.

Yes, it runs slowly because there is a lot going on under the hood. However, I've made some algorithmic optimisations (v3.03) so it should be faster. I've also made some gameplay changes to make it more fun.

What are the specs (CPU) of your computer? I'm running a Intel C2D6600 and I get 30FPS (new max FPS) most of the time now. Let me know how you get on.

Cheers, Ben.

It works now, thanks. Runs kind of slow though, less than 10 FPS.

Hi Vitor,

There is a new version. Try to unpack the source file again. (Is it failing when trying to unzip the zip file? If so I probably cannot help you).

I'm getting an error extracting the source file. Can't play the exe as I'm on Ubuntu

I spent a good while just ricocheting chaingun bullets into zombies and ended up killing my self with my own shooting.

very nice! I killed heaps of zombies. Runs 15-20fps with psyco for me too. Which is just fast enough to play in my opinion :) Looking forward to updates.

Hi Nathan. If you want to reduce the number of guns, edit line 315 in main.py (say 40->20 or 5 if you want a hard game).

Thanks for the feedback.

I'm taking a break over xmas, but #1 on my Plague TODO list when I get back is to use (numpy OR multithreading) or something to speed up the finite element analysis. Various other optimizations are planned after that because my goal is to have it run at 50FPS min (and max).

Wow, this is really quite good. However, it runs slowly even with Psyco (~15-22 FPS). I also thing there's too much armament available, there's more guns than trees :P.

Love the fire, even if it does slow the whole thing to a crawl.

Looks nice, I'm downloading it.