Procedural vs Handcrafted

Thanks for that!

When you say "juice up a turd" you're suggesting that there is something already very wrong with the game, which may be the case. But that's not what I'm talking about.

All I'll say is, if you go and try the average indie game, IME, even by the time you hit the loadout menu, you can often feel the 'missed beats' in terms of visual/auditory feedback, and a sort of unnatural stillness or emptiness. On the other hand, if you open up the average professionally made game (even from 2002 or something) it feels complete, there's nothing missing as such. It's not graphics, and it's not quality of audio, it's some kind of coherency and completeness that you get when someone simply knew what they were doing. There's a constant flow of feedback to the player.

@frosted you said once, I think, that UI is a very important, often overlooked part of a game, and I agree. IMO a good UI is almost a tiny, mini-game in itself!

 

What indie games have you played recently? Usually I find them covered in tons of juice.

Hard West:

Darkest Dungeons:

Death in Bermuda:

Soma:

Endless Legend

In my experience, adding juice isn't too hard and is even kind of fun. But general polish is extremely time consuming and it's just really hard to know when it's worth it and when you have a turd.

I guess the trick is - how do you know the difference early on? How can you tell when something's a turd or when it's just got some rough edges? Or what parts are turd-like and what parts just need polish?

I think this part is really an art, it's also very difficult to learn.

 

Lately I watched a playthrough of Eterium which has the issues I'm talking about. I also thought that Interstellar Marines had these issues too. I haven't watched too much stuff lately, but it's definitely something I've picked up from my youtube surfing.

 

I never played Eterium, but it's a pretty nice example and definitely lacks some of the serious polish of a larger game.

Why do you think Etherium doesn't have that much juice? What do you think the obstacles were in the production there?

 

Who knows? Budget, skills, motivation, I wouldn't be sure. Maybe they thought the game should be free from feedback pill-popping.

It's an old-school game so there's a reason why the graphics are not very complex, but watching the playthrough it was just clear to me that it wasn't done too skillfully. Things like the battle music during gameplay sounding like electronic dance. Also, near the beginning, when you're about to jump into the cockpit, you suddenly have to watch a long boring cutscene that isn't too relevant, basically just character introduction. And then when you die you have to watch it all again. Also when the playtester hit an asteroid and died, there was just a simple doink sound and suddenly you explode.

That's the sort of missteps that can add up to make a game sink IMO, because they aren't paying attention to how the player feels at each point, whether they're getting the feel and feedback that they came for, they're just going through the usual paces of what makes a game.

 

Looking at some of the video of the game - the biggest problem is the audio. The audio throughout leaves tons to be desired.

It's pretty clear they didn't have a dedicated sound guy or a lot of expertise in audio. I do think that it's hard to get this part right and it's something many of us underestimate, while also being a huge part of the game. Audio is especially difficult because it's so specific and but hard to communicate. The only time I was ever unhappy with the results from a freelancer involved audio (not music, but fx and ambient). As a solo indie dev, I've really struggled with sound.

It's the kind of thing that get's placed low on the priority list, despite being so important. The skill set involved is also very unique and removed from most other aspects of production.

As for the game play itself, it seems like the actual reviewers usually maxed out at around 3 hours. At this point either they ran out of content or the content became boring and repetitive.

If you were in the dev's place - and you had a limited amount of time - what would you focus on?

• Would you try to build out more content to extend the game's length beyond 3 hours?
• Would you invest your time and attention in the audio?
• In adding more animation?
• Would you invest in hiring, managing and integrating voice actors?
• Would you work on the menus?
• What about the cockpits or enemy ships?
• What about the particle effects?

All of it needs work, but you only have a limited amount of time and attention - where do you focus your time and effort?

I kind of think that indie dev is like running triage. Everything always needs desperate attention and you need to try to invest the time and effort where it matters most. Some things are going to die and some sacrifices must be made.

In actual practice, you gotta make choices on how to spend your time and focus. Is juice really what that game needed or is the 3 hours of game play the real problem?

 

Frankly I would simply pay more attention to how the game feels.

For example, I don't think you need to be a space combat fan to see that the combat is simply boring. It's sort of hard to tell what's going on when you're firing, and it's a dull experience in terms of vfx and sfx. When you die, there's no huge explosion, it's a disappointing way to go out.

For the first mission, there's little context, you just seem to come upon a few enemies and you start shooting. There's no build up, there's no pacing, there's no idle chat with your wingman, and then you shoot an enemy and there's some silly comment on the HUD like "that was for my wife!" if I remember correctly.

I don't think that voice actors are the main problem, the story just needs to be more than a bunch of slides with no pacing, and it needs great, dramatic music to fill in that void. The cutscenes are just filled with random anime characters with boring backgrounds, they might have shown instead some dramatic footage of a battle, or a huge space station or something like that, to get the juices flowing. At that stage, I didn't care about the characters since there was little context for what they were doing or saying. I wanted to know that the game was going to be fun and the story was exciting, but the cutscenes were simply lacking drama.

The menus could have been better, and had much better music. How about some epic battle music? I know some on the asset store that would fit right in.

I don't think the game's length is a problem at all. I really think you should check out Gigantic Army, for me it's a beautiful piece of game design, and it's only about 25 minutes long. In the reviews I read, some commented on the length but it was very surprising how little of an issue it seemed to be. Sure it's a different genre from Eterium, but three hours for the latter is more than enough to sell a good game. Besides, I didn't even know that! I'm commenting on what I saw from a 25 minute playtest.

Also, for a similar example of a space combat game that was done right, check out Starlancer. If you replaced the voice acting with text, I guarantee it would still feel great.

 

I just did a random search for starlancer and this was the first video I saw:

The crazy thing is that literally nothing happens for the first five minutes here. You fly in a straight line. The entire thing rests on it's audio for that time, a mixture of it's voice acting and excellent music makes that time build tension instead of boredom.

There is a lot more going on there than juice. This is also clearly a much bigger production with a much larger team. Could you make this scenario as the first scenario? If you gave it to your play testers in a pre-alpha state - what would their response be. I have no clue.

I wonder, did Eterium's mistake come down to a lack of skill - or did they just scope too large for their resources and had to triage out too much. If the designer of Etherium worked on a much smaller game, do you think they would have succeeded or did they just fundamentally misunderstand good game design?

I think that a main point of our disagreement is that you tend to look at these projects like their developers just didn't "get it", and I tend to look at it like "there were obstacles they couldn't overcome and mistakes they made".

 

First of all, by Mission 21 you're either hooked or you're not. If you check out the first mission, there's context and there's build up in the right sort of way. Also, even in that mission you linked to, there's a buildup with the cutscene of the ships flying out, there's a nice planet in the background, there are ships around in some sort of convoy, and all that feels dramatic, like something cool is going to happen.

Instead of audio, why not put text dialogue in the corner, with a helmet face like Starlancer did? Eterium didn't do that, they just sort of left it with no build up or communication at all, your wingman is someone that just hangs around, who you have little interaction with, as you randomly come across enemies and shoot them.

The music in Eterium simply doesn't belong there. If they put a simple action loop there, instead of dance music, it would have come off a lot better and helped the atmosphere. There are good ones out there, I have one from freesound that fits right in.

Apart from voice acting, I'm curious to know what obstacles you think Eterium could not overcome (apart from lack of design skill), and what Starlancer did right that would cost a lot of money or would be very difficult for an indie to emulate?

 

You might think this is a cop out - but my response is going to be the same as it ever was. The thing that's very difficult for an indie to emulate is the total package. It's all those parts moving together so well. It's all the little details they got right and all the small bits and pieces that come together to make the game feel so good.

I don't think the Eterium dev thought his audio was awesome. I mean, it's friggin terrible, it's clearly terrible. If I had to guess, those were supposed to be 'placeholder' and he was never able to replace that stuff. The drums in the mission briefing are especially awful, the loop is like 3 seconds long and you hear it immediately. He didn't have the resources (time, money, skill) to overcome the problem. He didn't have a dedicated audio engineer, didn't have the money to hire one, or the time to learn enough to do the audio himself.

Starlancer clearly did. In fact, there were four people including two attached to the publisher who were credited on the music and sound.

Listen to the music in the first 60 seconds of mission 21 - they have something like 4 different transitions happening just in that one minute!!

That said, I think you could make a Starlancer, or at least come close. Your odds go way up if you've got a background in film, tv or theater.

Just watching the video for mission 21 though, man, they did an absolutely top notch job on the voice. The audio absolutely carries that mission, the tension, suspense, surprise the actors convey resonates with the music and evokes the right emotion. The voice/music really work together to make what should be an immensely boring video surprisingly engaging.

That team that put that together was damn good, 2000 or not.

 

Yep, no doubt Starlancer is a great game, and well put together, I have it and I even play it every now and then myself. But I think that Eterium could have been a lot better if they had done the bare minimum, for example if they'd put in reasonable quality, relevant music, i.e., music that belonged in a space combat game. Whether they would need to put in as much variety as Starlancer did, in order to be a really good, fun game, is debatable.

And yeah I know you say that it's all the little things, getting all the moving parts working together, that's difficult - and you're right, it is. But that is our job as developers, that's the core of the game development experience, and the one thing we can't blame on lack of resources - because in my opinion, 95% of those little things are not so much a question of quality but a question of knowing what to put in, where to put it, and when. That's something we have to simply keep working at until it all runs like clockwork. And after all, that's the one thing we don't want to generate or outsource, it's what differentiates the game from anything else out there.

 

With a little tweaking, you've got a good start on a procedural ski slope generator. If you don't ski, go online to some ski resort's web site and take a look at the slope map.

 

Heheh, makes sense.

The basic idea for the map generator was:
- Roads in general connect a location to another location
- If there isn't a road directly connecting a location to another neighbor, it's because there is some geography in the way blocking it.

So town A should have a road to town B, but there's a mountain in the way, so no road.

With that assumption, I just built it in reverse, I dumped a bunch of roads down connecting locations then if there wasn't a road connecting two dots, I stuck some kind of terrain there (mountain/forest). The bigger the distance to the closest road, the bigger the mountain was.

Simple:
height = distance( closest_road_vertex )

Because I was using a delaunay triangulation, dropping the shortest edge still kept the vast majority of the nodes connected, so islands were very rare.

A lot of random assumptions and definitely not a fully polished cutting edge design, but it worked well enough.

 

+1, insightful. Interestingly, when brainstorming some concepts for my current project, I invited my wife to act as the "player" controlling a physical prototype with coins, paper cups, and pencils. She would tell me her intended controller action, and I made the prototype respond according to the vision that was in my head. Not only did we find some weaknesses in my concept (mostly along the lines of, "Oh, crap...I never thought about the player doing that!"), but more importantly she encountered one feature that I considered a minor detail. Au contraire! She was delighted with the idea, and said it was in her opinion the best thing about the whole game. So a very simple feature has moved from the wish list of "do it if I have time" to being a core design target.

Sturgeon's Law: "90% of everything is crap." (The trick is to identify and keep the other 10%.)

 

Even though at a meta level, the entire point of this subforum is to overcome Sturgeon's Law, I feel that the Game Design Zen series from @Gigiwoo should address Sturgeon's Law and ways to overcome it (even though that's the entire point of those podcasts, too. Meta episode? I don't know.)

Once again, TV Tropes is applicable, and has some very good discussion of said law.

 

Hypothesis: Part of this is due to the fact that in a large studio, more people playtest and internally review the finished product, who are not the same as the people who created it.

As a part of my day job, I routinely create lengthy (100+ pages) technical analyses for customers, and I also am often asked to be an internal peer reviewer for others' work. When mentoring others in writing, I share advice I received from a teacher long ago: "Read your work aloud to yourself." When we read our own writing silently, most of us see what we think, not what is actually on the page. Reading your own words aloud will force you to slow down, and it exposes awkward sentence structure the way tapping a bell with a striker will reveal a crack. This is also a good way to review your tone for the intended audience: Is it too formal? Too informal? Too argumentative or confrontational? Did you use the same word or phrase repetitively? Is it downright boring?

I'm not intending this post as a writing lesson, but I think (hence the word "hypothesis") that the same basic principle applies to other creative endeavors, including game design. It is extremely hard, even for a seasoned professional, to effectively review and edit one's own work. In a small studio, the core team can peer-review one another, but it is of necessity a creatively incestuous process. Large studios have marketing droids, executives, and sometimes third-party advertising/promotion who will see the pre-production work but have not participated in creating it.

TL;DR: Good editing usually requires a fresh pair of eyes.

 

That's fripping brilliant. Turn the problem upside-down and see what falls out!

/me files this concept away for future creative inspiration...

 

Agreed. My "Crazy Idea" files have a new entry thanks to this.

 

There's tons of value in that. But (hopefully without reopening the entire argument) I want to reiterate the fact that often the developer knows that something is bad or doesn't quite fit. Unlike editing an article, fixing problems can really eat up massive amounts of resources. Especially when the fix still isn't right.

My last iteration had audio that was just awful, it had a lot of similarity to the Eterium game we were discussing - the audio was empty and awful. I was deeply aware of the fact that it sucked, but I just didn't have the bandwidth to deal with it adequately, there was always something more pressing that had to get fixed. Audio is something that I think people deeply underestimate both in terms of importance and in terms of difficulty. The Starlancer example has absolutely top notch audio, like really, really good, and it does some extremely heavy lifting in that game. It also was not remotely half assed, many, many hours were invested in it. Comparing the audio alone in those two games is really instructive.

I also found that 'placeholders' have this weird habit of becoming final results. Especially when you are under resourced for the task at hand.

 

All very true, but Eterium's audio was not simply ordinary, it was absolutely awful, especially the music. I think some average generic combat music would have changed things a lot.

I haven't found audio incredibly difficult. I've spent some time lately in Audacity editing stuff from freesound for my space game kit, such as the explosions and weapon sounds, and all of the HUD bleeps and bloops I generated from scratch using tones. You can be the judge of it, but I don't think it's too bad, and I certainly didn't spend more than a couple of days on it. And I'm neither much of a music nor a sound guy at all.

I think that the greatest difficulty with any artistic thing, such as designing a game, creating graphics or audio, is the ability to be able to see something from a fesh point of view, to empathize with someone who is seeing it for the first time, and not get used to it. It's something I try very hard to work on but it's hard to tell whether you're actually doing it or not.

 

It's very hard even for professional game developers to build really good game systems, but what they're good at is recognising objectively bad systems and fixing them. Both game systems and art can be objectively bad, it takes a very skilled artist to create good looking bad art, everyone else just creates bad art. Any game system that doesn't work is objectively bad. If one is creating a platform game but the character controller is so bad that the player cannot judge where the character will land once they jump, or know exactly how close a character can get to a platform edge before falling off, then the player gets poor feedback and can't really play the game. That's an objectively bad system, and they're not really that hard to recognise but for some reason quite a few indie games get to publication and still have poor controls and poor player feedback. Professional teams solve all that stuff before development really even starts. The systems that come from that a player may not like because it lacks features but that is all subjective opinion, usually professional games are playable, they don't have obviously broken systems. The player might fall through the terrain in assassins creed bud that's a bug not a broken system, and that often happens because most game engines are not structured similar to Unity and have a different process to get art in and what needs checking.

 

Human voice, music, and scale can really add in a lot of work. Although I think the genre you're working on is really an asset in terms of not needing a billion variations of sound. I looked through Pillars of Eternity's audio. They have something like 50 different sounds just for dropping items you move around in your inventory. You barely notice this as a player, and these are just tiny icons - but once again - it's just an issue of polish leading to scale.

Music is also very tricky if your aren't working with professionally cut elements. Generating a loop from audio that doesn't naturally loop can really take a while and will most likely never be correct. Building in smooth transitions without the source audio tracks is a major headache. Etc. There are a lot of tricks that can help make these things possible, but that requires time to learn the tricks. Audio is a solid discipline and there will be a huge gap between novice and expert.

Lastly, when dealing with some sounds (especially voices) you need to have a billion variations. Just doing pitch modulation and stuff doesn't cut it. People will very quickly hear the repetition. Exertion sounds and the like can really need wholesale mountains of variation.

Worth noting that you can fall through floors in unity. Wonky stuff happens on scene load sometimes.

 

Of course, usually it's a broken mesh, I was just using it to illustrate a bug rather than a broken system.

 

Voice work is a totally different ball game, I was only referring to Eterium's horrible music and sfx.

I don't have it, but it seems like Star Citizen's Arena Commander module, which is all that fans played for a long time, only had pretty much one action loop, and it was extremely simple. I could be wrong though. But it didn't seem like there was much variation.

I know you think that a lot of variation is necessary to have good audio, but I'm not sure I agree. It would be good to find a game that had good quality sfx and music but little variety, and see how it comes across. My guess is that you'd hardly notice. IMO 99% of what makes a good game is not so much how far above average the quality of the stuff is, but rather the simple fact that the developers put in all the bits and pieces at all. I agree with what @sicga123 said, I think that at professional level the biggest difference is the consistency of detail and the diligence to finish something properly, and quality itself is not always such a defining characteristic.

 

I'm not necessarily talking about variety in terms of having a bunch of tracks. I mean having all the parts to a single track that you need in order to properly evoke the right feeling in the player.

Professionally cut music has many different parts and forms, from loops to building crescendos to stingers or fills.

Small stingers and the like are particularly difficult, especially when the audio you may have access to has fixed length and doesn't synchronize properly against the events you need to present. What if you only have a 15 second stinger, but you want the scene closure to be 5 seconds long? Do you edit the animation sequence to fit the stinger? Do you try to hack the stinger to be shorter?

A normal sized project doesn't need to ask that question (very often) because they have the man power to assure that the stinger is the right length in the first place - or they have so many variations they can just pick the one that'll fit. The parts all fit together right because they were all tailor made to fit together from the start. These kinds of things are not above average - they're entirely average.

Again, the genre you're going after is a huge asset for you - machines make consistent noises, so you don't have to worry as much about variation - people expect the laser cannon to sound the same each time - maybe with small pitch variations. With exertion effects and stuff, it just feels weird and wrong when you notice the repetition.

"hey that guy made the same sound as this other guy and that's stupid!"

Final note - your sound might be great, and you may find better ways to handle audio than I do. I just mean to caution you that it may end up being a larger challenge than you initially expect it to be.

 

I wonder if you can manipulate something like exertion sounds at runtime from one audio loop, i.e., change pitch, speed and so on to give you something more natural and fitting for a given situation?

I kind of like what that animation guy did in some video we were looking at on the other thread, using a single keyframe I think he managed to make a pretty good looking walking animation (for a kung fu rabbit anyway!) And then using another keyframe he made crouching and landing animations. I wonder if it would be difficult to do the same sort of thing for certain types of audio?

 

This is standard practice, most audio will randomly select a range of pitch in order to help extend the amount of variation. For my game I have a 'manliness' range assigned to each character that further sets the acceptable range of any sound they make.

That said, even with this kind of thing in place - you still really need somewhere in the range of 100 or so exertion clips to really do an adequate job, much more if you really want to give different character types meaningfully different feels.

My game currently uses about 20 exertion clips, 15 impact clips, 10 footstep clips, 7 'clangs', 4 bow pulls, etc. It's still in very small prototype form, and it's already wholly inadequate. To do a proper job (on the prototype) I would probably need around 2-3x the exertion effects and 2-3x the impact effects.

In terms of manipulating loops at runtime, this is also standard practice. Stuff like really making coin sounds good for instance - you want the coin drop effects to synchronize against the money label increasing, so you need to loop the audio and/or re-layer the effects depending on how many coins you want dropping and the speed (build up/fall off). You can perhaps get away with something crappier, but coin sounds are important juice, and I think it's worth making that excellent.

EDIT:

I don't mean to ramble on forever about audio, but I really think it's really important and interesting.

Film Critique

I don't agree with this discussion of action films. But at around 7-8 minutes he talks about how with the way that modern action films frequently work with frantic cuts and insane visuals but how the sound tracks actually explain context and explain what's actually happening even though visually the action can border on incomprehensible. The sound design plays a tremendous role in explaining what's happening.

 

In my day job (related to IT process and control center design, and sysadmin), we call things like that "Quonset huts" because they're intended as temporary but are still around decades later.

 

On the large modding project I mentioned earlier, we had the amazing good fortune of two professional composers who graciously volunteered their time (collaboratively) to create tracks for the in-game and trailer music. I had never worked in this area at all before, but the composers were extremely supportive and patient, teaching me what I needed to know without trying to override the artistic vision of the project. The level of detail and depth of knowledge of a good composer is mind-blowing.

Further, I am an IT veteran with 30+ years experience and two engineering degrees, so it is not easy to intimidate me with a piece of software. The DAW (digital audio workstation) tools modern composers use make my Audacity look like a baby rattle, and are quite intimidating! So not only are composers artistically gifted, but also technologically skilled at an impressive level.

 

Just an aside, I wanted to thank all and sundry for perpetuating this thread, and thanks for letting this novice join in the discussion. I'm learning a lot from all of you. I hope the OP isn't too unhappy that topics have drifted all around the proverbial barn.

 

No worries about that, I've been around these forums awhile and the original point of most threads is largely irrelevant after page 1

I actually am quite pleased with the discussion in this game design forum, I had largely ignored it until recently and now it is my favorite subforum with interesting and intelligent discussions going on.

This audio stuff is particularly interesting to me, I sort of assumed most indie just buy some tracks from composers on soundcloud or the like and try to stitch it in there best they can. At least that is what I was planning on doing for lack of better options.

 

I think you could do a lot more with a lot less than that, but who knows? Maybe I'm wrong and you really need 100 breathing sounds.

Sometimes, sure, we're too optimistic. But sometimes, I look at No Mans Sky, generating plants, animals, oceans, planets, atmospheres, asteroids, stars, ships, space stations, the entire game ... and then I look at forums like this and I think, too optimistic? Hell, all I want is a pair of crutches to lean on when I don't have enough legs, I'm not asking for something that spits out a ready game, or designs a whole game for me or something.

No doubt a lot of that stuff is way over most of our heads in terms of putting in our own games, but I think that in our general discussion sometimes we're not doing justice to the reality of what has already been shown to be possible, let alone what is within the realms of near-future possibility.

 

Don't think of it like that...

Let's say I have 3 basic exertion sound categories: Attack, Charge, Death
Let's say that I have 5 main 'types' of AI character: peasant, brigand, guard, barbarian, priest
Let's say I want 10 different versions of an attack sound for each type, 2 charge sounds, and 3 death sounds.

Attack sounds: 10 * 5 = 50
Charge sounds: 2 * 5 = 10
Death sounds: 3 * 5 = 15

That's 75 sounds.

Maybe I could get away with less, maybe 50 or 60 and still present a variety so that each type of bad guy feels a little different.

Same kind of thing with locomotion sounds like footsteps, impact sounds. What if you're walking on stone instead of grass, or water. You need splashes or hard foot impact. What about wood creeks? An axe needs a different kind of impact sound than a sword, different clashes, a hand axe needs a different set of sounds from a giant two handed axe, and a wooden club needs an entirely different set of sounds.

Game assets spiral out in a crazy exponential kind of way, everything gets multiplied out. Once you want 'polish' and 'quality' it can start to get crazy once you really start to see all the crap that other games use to really present a polished product. Can I get away with less than that? Sure, but every time I do - it's another missed opportunity. Another chance for something to not fit quite right, and another thing that will annoy some gamer and break immersion. It's taking one step closer to Eterium.

Audio is really weird - it's one of those things that you don't often notice when it's done well. You only really notice it when it's done crappy. Well done audio usually just sort of fades into the background, but it's constantly giving you context, helping to describe and explain whats happening and telling you how to feel about it. It can do tons and tons to help with immersion.

 

Ah, I thought you meant per character, of course it depends on the number of characters. I was also only referring to exertion sounds. However, I've still got the sense that an indie game could be successful without as much polish and variety as you seem to describe as being necessary. But anyway, who knows? I haven't really got there yet.

 

Or maybe I could be wasting my time trying to build out more exertion sounds. You did sort of get me thinking, maybe I am working dumb, maybe there is a way to get more variation out of what I've got or cut some more corners without sacrificing too much quality. Maybe I'll try some experiments this weekend.

 

This is precisely why this comment...

...sounds to me pretty much like someone would say:

I don't want to offend, but I really think I'm not exaggerating much here. Sound design is friggin' hard and I have a hell of a lot of respect for people who specialize in this. It's on the list of things that I'll probably never gonna attempt in a serious fashion, just like animating. It just takes too long to get good enough at it and is a massive time sink. And Audacity... the DAW I'm using for music production has so many entries in the options panel alone, that they implemented a search function just for that (Love it! I wish all programs had something like that.).
You want a challenge? Find me the settings for either a free VST synth or any synth in Komplete 10 Ultimate, that 100% replicate the two synth sounds (the ostinato and the pad) right at the beginning of this track:

(System Shock 2 Ost - Engineering)

I've tried to do that and spent some 30+ hours doing nothing else but trying to replicate the exact sound of that melody (the pad probably would be easier but I'm not sure) and failed! Others have tried before me and the only person I could find, who came very close, used a hardware synth similar to the one used by the original composer.


Now about Eterium, this is from their FAQ:

I think for a team of 2-3 people being on their first big game some slack cutting would be in order. We've all seen much worse.


Yesterday I was reading through two long posts on mixing for sample based orchestral music:

http://music.tutsplus.com/tutorials...tips-for-producing-sampled-strings--cms-21216

http://audiojungle.net/forums/thread/trouble-with-mastering-mixing/105382?page=2

I thought that was some top notch information presented by the guy, so I wanted to share it in case we've got some composers silently following our discussion.

I'm currently contemplating making a dynamic music asset (Only .wav loops, transitions and stingers in the asset. And as an example, distributed for free and without guaranteed support, an FMOD studio project (set up to dynamically adapt the music based on 2 input parameters).). How would you guys think is general assetstore interest for loops intended for adaptive music use, compared to just static tracks? And how are you feeling about using FMOD in your games?

 

I'd focus on finishing gameplay first and then see how much time is left for audio polish. Some people care a lot about audio, some people don't care at all because they listen to podcasts while playing. Good gameplay benefits everyone. I like to call those things "procrastination tasks" because they are more fun than the important work that needs to be done and easily make you get distracted from what you should be doing. I think I spend somewhere between 90 and 100% of my dev time on such procrastination tasks and that doesn't get games done.

 

Dialog voice tracks also multiply rapidly. Even Bethesda, a AAA publisher with circa 100 people on a game team, struggles with this. In Oblivion, for example, they have 8 (?) character races. For non-principal NPCs they had one actor per race, per gender, but overlapped a couple of races that they decided could believably sound the same. So figure 6 voice actors for generic NPCs, times two to get male and female. Now, each character needs (among others) the following (uppercase words are the game's internal ID string for special generic dialog):

• HELLO -- These lines are uttered randomly when an NPC comes near another NPC or the player, whether or not a conversation will ensue. They are not uttered 100% of the time, but often enough that there needs to be reasonable variety. A lot of Unity developers refer to these as "barks".
  
• GREETING -- These are uttered randomly when an NPC begins a conversation with an NPC or the player. The generic randomized variants of these are about half a dozen per NPC race and gender.
• GOODBYE -- These are said when NPCs end a conversation with each other or the player.
• Attack -- Barked during battle, randomized list
  
• Power attack -- Barked during battle, randomized list
• Death -- Sounds or final words when the NPC is killed
• Crime -- Random list of lines barked when the NPC detects the player in thievery or assault
• Commerce greetings -- Special (again, randomized) variants of GREETING for shopkeepers touting their wares. Some are shop-specific, but many are generic to the type of shop (maybe 4 or 5 variants).
• Commerce success -- Random lines said after player buys or sells something.
• Commerce failure -- Random lines said when the merchant refuses the offered price to buy or sell
• Persuasion -- Multiple variants of lines when the player is attempting to improve relations with an NPC
• Detection -- Barks uttered when an NPC detects the player's presence but not their exact location
• Combat yield -- Barks uttered when an NPC flees from or ceases combat
• Yield accept -- Barks uttered when the player sheathes weapons during combat, which is the game's way of asking for a truce, if the NPCs AI chooses to accept the yield.
• Rumors -- All the random bits of gossip that NPCs say to one another, or to the player when asked. There are probably over 100 of these, varying from questions and answers about the other's well-being, to where one can find the best armor and weapons to buy, to discussions of how dangerous mudcrabs are.
• Factions -- Every NPC is in one or more factions, and all the factions have specific lines that any of their members might utter. Merchants talk about shopkeeping. Imperial soldiers talk about how sore their feet get after patrolling all day. Mages talk about spell research. This is a huge area, again probably averaging 20 or more lines per faction times a couple dozen factions.
• Races -- The various races and species have specific lines that reflect their own cultural heritage, and which allow them to greet kinsmen with extra warmth. Comparable to the lines for a faction, so figure about 20 here.

Now, you can see there is quite a list above. For the sake of illustration, let's pretend there are only 40 total generic rumor lines, although Oblivion actually has a lot more than that when you add up all the situational types. So what's the tally, figuring 5 each on the other categories plus 40 for rumors and Q&A dialog, plus (20+20) for a very low estimate of faction and race dialog lines?

I get (14 x 5) = 70 generic lines, plus 40 rumor lines, 20 faction lines, and 20 racial lines, for 150 total. Now multiply that by 12 permutations of race and gender. If my math is right, that's 1800 lines of dialog. I can tell you from having modded Oblivion extensively that this number is way, way, WAY low from what's really there.

Now, the funny thing is, all of this just covers the generic NPC guards, shopkeepers, peasants, priests, and guild hall members, and not a single quest nor unique NPC. I have an exported dialog tree in my modding environment, and it shows almost 39000 MP3 files in the Oblivion game.

This doesn't include non-verbal cues like footsteps, weapon hits, and so on. I'm just talking about vocal sounds.

All the people who download Unity and jump right into the forums about how they're going to make a better Skyrim in their spare time will be really sad when they discover just how massive even this one facet of such a task is. For our mod ("Rathunas"), we had 102 NPCs, averaging about 35 dialog lines each for a total of over 3500 lines. This was well under 10% of what's in the base game, yet it was a 2+ year effort worked on actively and nearly continuously just to get the dialog written, cast, recorded, edited, lip synced, imported, and tested. Just keeping track of all of this required custom tooling and a relational database.

Oh, and all of the above assumes you're releasing in just one language. If you are doing multiple languages, it scales pretty close to O(N) magnitude.

 

This kind of exponential growth is the problem when you try to increase quality.

Right now, the most important thing missing from my audio is really a "Who's there!?" the 'surprise/alert' bark from an NPC. This would make a huge difference in my game and really help make it way better.

But again, I'm going to need to multiply out a hand full of bark variations across a bunch of different types of people. Even something soooo simple ends up being a pretty big hassle. I can get this done, I can hit up fiverr and hire people and collect the audio and stuff. It's all totally possible, but it's just another thing on an already very long list.

It's one of the things where having help, even non skilled help, just literally another person who can worry about something so I don't have to, that can make a big difference in making sure those little details don't get lost.

 

+1. My wife, an extremely smart but by choice nontechnical lady, was kind enough to take care of a lot of the organizational details, especially when it came to audition evaluations which we tracked in The Giant Spreadsheet That Ate New York [tm]. She also wrote most of the non-quest scripts and edited my writing for the rest (I wrote the quest lines because they were so intricately linked to game mechanics because of dialog being conditioned upon, and/or triggering, events in quest logic).

I'm a decent techie, but she's a much better project manager than I will ever be.

 

Just wanted to add a note - I'm always saying "more stuff = problem" and I want to point out a differentiation between "time" and "attention".

The problem isn't time, the problem is attention. Like hiring a few people from Fivrr and getting these clips probably wouldn't take that much actual time. Maybe in grand total 10 hours, 20 hours. It'd be spread out and just end up being an hour here and there. But in total it wouldn't actually be that much time.

But it takes up your attention, it's an interruption once every other day for an hour or so that throws off your work or distracts you from other things. It's another item that you need to keep track of and follow up on and, perhaps most importantly make decisions on.

The problem with splitting your attention isn't so much time - it's like mental exhaustion, and once your brain just has too much to keep track of you start to really lose productivity. The big challenge as an indie dev is not managing time alone, it's really managing attention.

That's why all those little details we take for granted are so hard to match in larger productions. It's that trying to do all of those different things, keeping track of "is that icon right? does that animation fit? is that sound level? did that bug get fixed?" keeping track of all of those very different kinds of problems... it's just really really hard.

I think the biggest asset for a serious attempt at a mid+ scale project comes down to organizational skills more so than anything else.

 

I totally suck at this and would like to hear how other people manage it.

 

Me too. It's one of the reasons I'm interested in @syscrusher's experience. It sounds like a major reason why his modding experience worked so well is that it was well organized and managed.

Is the secret having a wife who can act as your project manager?

 

ROFL! We were not well organized at all! We only added decent project planning and asset management tooling after learning the hard way that we needed it! Our 5 year project would have been 3 years, had I the wisdom to plan it properly from the start. To quote songwriter Tom Smith, "If Murphy's Law's a religion, I must be a saint."

Wife, husband, or significant other, as the case may be -- or a close and trusted friend who can tell you your baby is ugly without offending you.

 

Some sounds lend themselves well to procedural randomization from stock cues, others don't. The physics and math of sound are reasonably well understood, and fortunately sound clips are low bandwidth by the standards of modern CPUs, at least compared to graphics. Slight variants of the same sound are mathematically easier than trying to create a different cue procedurally. For example:

• The player is walking on snow, and you want slight random variations in the ADSR envelope or the pitch of the cue. That's not too hard procedurally, and you can get away with some imperfection for most games.
• The player moves from snow to rock. Procedurally changing the snow footprint into a stone footprint is damned near impossible.
• In a steampunk game, the player enters a room with dozens of machines of varying size and uncertain purpose. Any given machine will probably make the same looping sound if it's running continuously, but you could randomly start and stop some of the smaller machines underneath the general din, and put interesting starting and stopping clips in the appropriate places for those events. Toss in an occasional random clank at a fairly long random interval, and you've got a pretty convincing audio environment. Here you aren't so much altering the sound as altering the sequence of sounds.

But as others have pointed out, procedurally altering human sounds to produce different voices (even for things like exertion or coughing) is "nontrivial, and left to the student as an exercise."

Sidenote from an engineer with a background in industry: Machinery starting and stopping does not always just change the pitch and volume of its steady-state looping sound. Sometimes it does, but not always. Stepper motors and servos used in robotics are fairly close to the ideal except when they are moving very slowly (e.g, have just begun to start or have almost stopped). Sliding friction and metallic contact sounds tend not to follow the theoretical ideal very closely. I've listened to machine sound FX assets in the Asset Store, and IMO some of them are better than others at sounding like a real factory.

 

Oh, another free audio tip for you: Magnetostrictive noise, the hum you hear around electrical transformers and other inductive devices, can be reasonably approximated with a sine wave at the frequency of power in that locale (60 Hz US, 50 Hz Europe, YMMV elsewhere). Toss in a few third and fifth harmonics, and maybe a bit of second harmonic, and you'll be pretty close. Add more harmonics to simulate a bit of loose metal vibrating from the magnetostriction, if you want a truly annoying sound!

For three-phase power, emphasize the third harmonic more.

 

It's funny, I really think these kinds of organizational issues are probably the biggest overall challenge. It's also some of the least discussed - probably because it's the most boring aspect of development.

Even super obvious stuff like putting together a design document (not even a formal GDD, just a detailed outline) and actually keeping it updated. This kind of thing can be a huge asset in terms of keeping track of your goals and not losing track of the big picture.

The smaller your team is, the less obvious it might be that you're being hurt by a lack of organization. It's very easy to let things start to slip and fall into disorder. It's a major problem I need to work on, for sure.

 

Gamasutra has another article surveying the pros and cons of procedural content generation in open world games, by Joel Couture who interviewed Adam Saltsman about the procedural-heavy game "Overland".

 

I just read that article. I like that it doesn't spend a lot of time debating whether to do procedural generation or not. It just gives lots of practical advice on implementing and tweaking procedural generation with a focus on generating the best player experience. Lots of good tips and insights in there.

 

Agreed. Adam Saltsman was speaking from practical experience, and it shows.