Game Size

[kokoro nut]

Well, In my own opinion, if the game you are developing weighs in at around 20 Megs, then my advice would be to consider creating a Web Driver version of it. That way, people won't have to wait two hours in order to play it. Instead, all you would have to do is just install the WildTangent Web Driver,jump over to the lanch page of that game, and start playing it in just a few minutes as opposed to hours.

[PyTom]

The whole b.us/r.us/l.r.us complex runs on a server I rent from ev1servers. I pay $100 a month for that, but a friend of mine reimburses me $50 in exchange for half of the server. We get 1000gb a month, so that's $.10 a gig... if we use it all.

But, of course, we don't use it all. (And since it's $.75/gb overage, we try not to come too close.)

Also, it's a Linux server I administer, so keep in mind that you'd need some sysadmin experience to make it work.

[PyTom]

Well, In my own opinion, if the game you are developing weighs in at around 20 Megs, then my advice would be to consider creating a Web Driver version of it. That way, people won't have to wait two hours in order to play it. Instead, all you would have to do is just install the WildTangent Web Driver,jump over to the lanch page of that game, and start playing it in just a few minutes as opposed to hours.

I'm not sure how that would help at all. In a 20 mb game, the size of the game engine is small (~2mb) compared to to the size of the game data (the remaining 18mb). It's not like we're making the user download game data that isn't being used, so all of that would have to be on their computer before the game could be played.

That's not to say that for any particular game, there isn't data that is unnecessary. But usually, we're pretty good about that sort of thing.

[papillon]

By shareware eyes, that 10-15MB is already pretty big (bigger than I'd go for a demo at least). As far as what *users* are willing to download, I think the general business idea is that anyone who will do 10 will be almost as likely to do 100. The dropoff points are supposed to be 5MB, 10, maybe 20, 100, and after that point there's no limit - if they're willing to dl 150 MB they'll do a gig too.

Some adventure games have done voices in a separate download and rigged the code to be able to work with-or-without, so you can get the game and START playing quickly and take extra time to install the voices if you want.

[ShiraiJunichi]

I think having separate downloads for additional content is a really good idea. Seeing as RenPy supports voice, I am very much considering using it for my game. Having it as a separate download would allow people using dial-up to still download, and enjoy the game- while giving something extra to those with high speed connections. The same could be done with videos.
We could also encourage that creators use this method, even if they don't have a really large amount of media files. Putting the few mp3 music files of a game in a separate download could probably decrease the download size of most games by a few megs at least.
Perhaps someone could create a tutorial about the best ways to slim down file sizes of games. It could discuss things such as sound formats, image formats, different compression methods, paper doll method, etc... Now, I think most creators are fairly knowledgable in these areas, but it could possibly enlighten a few about things they hadn't considered before.

As far as Bit Torrent goes, I think it should avoided as long as possible. And when it does become necessary, it should only be used for the larger games.

[PyTom]

Okay, found out several things about Ren'Py games and size:

Apparently, some windows tool does not compress zip files when they are added to other zip files. This is a problem, as lib/renpy.zip is an uncompressed zip file storing a number of python bits. Compressing this file significantly reduces the zip size. (Uncompressed, it was 1.6 Mb, compressed it's 460 Kb.) To fix this, I've renamed renpy.zip to renpy.code, in new distributions.

I also took a look into the zip file, and saw it was including some unnecessary libraries. By manually excluding them, I reduced the size further to 1.1 Mb uncompressed, 302 Kb compressed.

I also noticed that some games ship both rpy and rpyc files in the common directory, when either is sufficent. So I yanked those out.

I also noticed some games do not zip at the highest level. So we need to be careful about that.

Since Metropolitan Blues was handy, I did some experiments with it... By making these changes, I was able to reduce it in size from 13.3 Mb to 12.0 Mb. This is without touching the game data.

A quick experiment shows that another meg can be saved by applying pngcrush to the pngs... but I haven't tried zipping them up to see what the actual savings is. I think it might be close to that.

So 2 megs saved from 13 without losing any quality is probably a good start. The fixes to Ren'Py stuff will be in 5.0.1 when it comes out.

(I'm certain other games, including MW, could also save, but I just had MB handy.)

[Megaman Z]

my input on the general topic: when programming, don't worry too much about the game's size at that time. it probably will drop when compiling (this probably applies to ren'py, but I haven't had the pleasure of actually trying to compile a game, and when [wait... where's the strikethrough button?] - if I get to my computer and it works, I still don't have enough of TOZSP to do a release [that, and I'm lacking the mpeg to episode 0... wait... that was supposed to be a secret. DARNIT! LOOK WHAT YOU MADE ME DO!])

I think the idea with this would be that the Ren'Ai archives would seed the torrent at a constant upload rate, and people downloading it would also add more bandwidth, improving speeds.

now, about torrenting the larger games: first, I think Azureus would be better. second, if someone can get the game (or game set) to me in the first place, I think I can make a .torrent file for it and, using Azureus, super-seed it for a couple days. hopefully my internet won't be screwy (blame Dennis if it is), and a few people will have full copies so they can seed. the only problem is with tracking the torrent: I'm probably suffering from a NAT problem right now on my home computer, and that prevents decentralized tracking from working. if one of you can get a tracker working, then that problem is solved. (granted, my room's temp. will raise five degrees, and other people will be ticked off about it, but hey, it helps you guys.)

(if you want an explanation of super-seeding, I'll post an explanation. otherwise, I won't)

[PyTom]

A dedicated server is probably better than azureus, as I can guarantee its uptime, and it will almost certainly use less resources. Also, super-seeding probably isn't a good idea, as we may not have large numbers of people downloading the game at once.

This is all hypothetical, and probably won't happen any time soon. I mean, we still have plenty of bandwidth, and people seem opposed to BT, so I think we should avoid it for as long as possible.

[mikey]

Well, the reason I asked is that I really have a problem with music extra. I have the feeling that especially in games which are free or aren't hyped up, people won't be inclined to download a music extra. Once the game is playable, they will play it.

And the second reason is simply that you actually ADMIT that the game CAN be played without music. Which perhaps means that you think the music isn't entirely necessary (from a player's point of view). It also means you give your players TWO different experiences. The moment you separate the music from the game is the moment you break its consistency.

I don't know the mentality of the downloading person, but normally if I had a 9MB game and an extra of 15MB music, I wouldn't want the hassle of unpacking and copying (be it even automatically installing) the extras when the game is a separate file and thus implies the above paragraph.

The best way to slim down games? ^_^ The best way is planning it from the start. The number of scenes and unique BGs was a major factor for Kaori. So that's phase one - here you can really drop the size (2.3 Kaori vs. 11.5 Milkswim). The second phase is as suggested, is in the tweaking and balancing phase, where you can shave off a megabyte or two. Anyway, I just wanted to mention this, because if you really want the final product to be a small download it's best to think ahead - and not make everything as monstrous and carefree as I did with Rivertrap and then when it's too late decreasing color depths and so on. Yeah, you have to learn from something I guess...

I guess my feminine side makes me very conscious of my game's "weight" and actually the renpy.zip problem PyTom described was really getting me down (I saw a zipped file and had to find out that it's needed), so thanks for every little byte you can save us!

[PyTom]

I don't know the mentality of the downloading person, but normally if I had a 9MB game and an extra of 15MB music, I wouldn't want the hassle of unpacking and copying (be it even automatically installing) the extras when the game is a separate file and thus implies the above paragraph.

My gut feeling its that it's less onerous when you have two complete installers, a broadband one that includes the music and a modem one that includes no music, or lower-quality music, or whatever. This way, instead of a second install, the user can simply pick the package that's right for them. Disk space on a server is cheap, at least when it's measured in megabytes.

[ShiraiJunichi]

And the second reason is simply that you actually ADMIT that the game CAN be played without music. Which perhaps means that you think the music isn't entirely necessary (from a player's point of view). It also means you give your players TWO different experiences.

Well, from all the games I've seen, it's true- it CAN be played without music. I've yet to see a game that required being able to hear the music in order to play it. Aesthetically speaking, you have a point. But the fact is, realistically speaking, that you can play without music, but you can't play if you can't get the files onto your hard drive. IMO, I think it's better to give a wider audience two similar experiences, than to provide a smaller group with a single experiece. Especially if you can get all the people who would download the full version if that's all there was, to download the larger version, even if a smaller version is available. I think PyTom suggested a great way to accomplish this- don't have "extra" downloads for the additional content, but a completely separate download that contains the game plus the additional content. That way, no extra work needs to be done by the player to make it work. They can look at the downloads and see, "Oh, this download has the music with it. I'm on a fast connection, so I won't mind waiting a few more seconds to download that"

[Megaman Z]

I've yet to see a game that required being able to hear the music in order to play it.

c'mon... I know there's one... can't... think... of... it...

closest thing that comes to mind (not counting bemani and bemani-genre games like DDR and ITG) is Biohazard: Outbreak (file 2) [AKA Resident Evil: Outbreak (File #2) in the US]. I've been killed so many times from Titan surpise attacking me when I had the game muted. (you think a giant zombiefied elephant would be easier to spot...)

[ShiraiJunichi]

Well, I was referring to the free ren'ai games

[papillon]

Hrm... In some sound effects are a helpful hint/clue of character mood, but most of us don't have a very full soundtrack to really exploit that. (Schoolgirls has a particular theme to play when you've managed to reach Deep Emotional Chat time with one of the girls, but it doesn't play any little frills along the way to indicate when you've just scored points with a girl, which would probably make the game easier. )

[mikey]

I have to say clearly I really RARELY play games with music. Usually I put on my favorite music and play like that. As for fan-made free I am also a barbarian ^_^. Most of the time I play through the first time(s) with the intended music, but then WinAmp wins over and I listen to my other tracks. Of course that doesn't mean I play silently. I need to hear the engine in all those racing games ^_^

So my release scheme is more dogmatic and aesthetical than rational, that I admit. You can make several good things with music which will better up your game (experience), just as papillon mentioned. But it's also true that very few games critically need music. I had to decide with almost every game do I include music or not... But once the decision is made, I like it to stay like that.

This leads me to an interesting thought about additionally releasing music for games. Surely it's easy with ren'py, "just add music" and release the files along with an updated script. Of course adding music will improve your game... But I doubt that I would re-play a game with added music.

It's the mentality of a game-maker (I suppose), that when you finally label something as "finished", you like to consider it finished. It's a liberating feeling and you can move on to the next project.

And also, normally you spend some time getting or composing music and it's not always possible to re-use that music after you can't include it in the project. Plus, you want your project completed and it's a strange (if not demotivating) feeling when you KNOW you're going to release two versions of a game to have one finished and adding and composing music for the second, because your general idea is already DONE. So that extra work for that extra effect sort of feels in vain.

It may seem like an excuse of course, but that's the way I see it.