Visual novels as a medium

[Mihara]

Even though I'm new to this forum, there's going to be no introductions, since people don't really care about those, in my experience. Instead, I'll just try to get on with the topic and make it an interesting discussion.

I doubt there have been serious published studies on the subject of interactive visual novels as a medium, at least, I couldn't find anything offhand. Nevertheless, they are just as much a viable subject for such studies as music (which has trends, genres, styles, all heavily interconnected and independent) or any other kind of art.

I don't think I could get away with doing that sort of thing for various reasons, but it is kind of important for my own project, so I guess I'll try to bring it up, seeing as how this forum has enlightened people who played far more of these games than I ever did, and hacked them.

The history of FPS genre can be easily defined in terms of engine features that push the envelope. First we had Wolfenstein 3D which had, essentially, a 2D world to it. Then, after a few examples exploring this concept and making the most of it came Doom, which had a world which actually had more of a vertical dimension but still was designed in terms of flatness (for example, no two objects could occupy the same X-Y coordinates, regardless of their Z coordinate). Then came Quake, which was fully three-dimensional, etc, etc, with the demands of realism pushing the hardware to where it is now.

As far as I can see, having played a few games released from late 80s to present day, visual novels have evolved little over the years as a medium. It is to be expected, since the medium relies heavily on written word, and books have the advantage of language as their main medium and thus have changed very little over the years -- but still, new ideas form new genres of literature, and new concepts make new books worth reading, or we'd stop with a few thousand good books and stick to them. There must have been similar developments in visual novels as well, and not only in terms of plot, but also in terms of engine features and visual language. (Those who have read 'Understanding Comics' will get what I mean by 'visual language' - those symbols used to convey meaning graphically.)

What were those developments? Does clever use of transition effects enhance the game for you or distract your attention from what really matters about it? What purposes exactly merit the use of animated cutscenes and lip synch? Do voices make the game better or worse and what exactly do they add to it?

What exactly the reason behind the slowed down text display speed - is it the japanese language, or a necessity for the medium itself, regardless of the language used?

Specifically about the features, why modern anime-style art frequently makes use of vectors, and yet, the only visual novel games that actually do are made exclusively with Macromedia Flash? Why exactly most visual novels stop with 800x600 display resolution, when most other kinds of games have long since moved on to 1024x768 as the minimum possible resolution?

Why do the visual novels look the way they do, which of their aspects are defined by necessities of storytelling, and which are technological limitations? Which are defined by the demands of their traditional audience and capabilities of their computers? When and how these technological limitations were first overcome and for what kind of storytelling purposes? What sort of engine features could define a game's entire development cycle?

I do not have enough information to discern the trends and reasons the visual novels developed as they did, but I hope some you can think of something.

[AstCd2]

Originally Posted by Mihara

What were those developments?

Insofar as engine developments go, some noteworthy ones have been:

• Opening/Ending songs and movies
  Flag-based story paths
  Voice
  Animation (both movie and engine-based)

Originally Posted by Mihara

Does clever use of transition effects enhance the game for you or distract your attention from what really matters about it?

I've got no issues with clever effects. Looking nice isn't what makes a game great, but it certainly doesn't hurt.

Originally Posted by Mihara

What purposes exactly merit the use of animated cutscenes and lip synch?

People like seeing animation. It's pretty and it helps the audience envisage the world in which the story takes place.
I don't think anything justifies lip synch. It's weird and creepy when characters can move their mouths but not the rest of their bodies.

Originally Posted by Mihara

Do voices make the game better or worse and what exactly do they add to it?

I don't think there's a universal opinion on this. Some people think voices add to the suspension of disbelief and the persona of the characters, while others think it robs them of the chance to 'fill in the gaps' with their own imagination.

Originally Posted by Mihara

What exactly the reason behind the slowed down text display speed - is it the japanese language, or a necessity for the medium itself, regardless of the language used?

I guess some people feel it makes for a smoother transition compared to the alternative of having big slabs of text hurled at you one after the other.

Originally Posted by Mihara

Specifically about the features, why modern anime-style art frequently makes use of vectors, and yet, the only visual novel games that actually do are made exclusively with Macromedia Flash?

I'm not sure what you mean by vectors, here - but in any case, there aren't many visual novels written in flash at all (I can think of maybe two, both of which are doujin games).

Originally Posted by Mihara

Why exactly most visual novels stop with 800x600 display resolution, when most other kinds of games have long since moved on to 1024x768 as the minimum possible resolution?

Given that the primary focus on these games isn't the graphics, I'd say that there probably just hasn't been the demand for it.

Originally Posted by Mihara

Why do the visual novels look the way they do, which of their aspects are defined by necessities of storytelling, and which are technological limitations?

The only limitation I can really think of is how many pictures the background and character artists are willing to draw. What determines how modern visual novels look is a compromise between what's necessary for storytelling and how lazy the artists are feeling.

Originally Posted by Mihara

Which are defined by the demands of their traditional audience and capabilities of their computers?

The anime art style is probably a legacy of the traditional audience of the visual novel (ie. anime and manga fans).

PC limitations don't really come into it, as visual novels don't tend to be demanding on system resources - though perhaps that's changing, given the increasing number of developers taking advantage of improvements in PCs to make full animation/3d model-based games.

Originally Posted by Mihara

When and how these technological limitations were first overcome and for what kind of storytelling purposes?

In the last 3-5 years. Storywise, they allow for greater variety in the direction and more interactiveness in some cases.

Originally Posted by Mihara

What sort of engine features could define a game's entire development cycle?

Voices need recording, animation needs drawing, and flags need debugging. Each has an effect on the development cycle, but I wouldn't say that any of them define it, except perhaps in the case of full animation, in which you'd need to think the production through in a different way to a text novel.

[Mihara]

Originally Posted by AstCd2

Flag-based story paths

What exactly are those? Choices midgame which do not create diverging paths until much later in the game, or something else?

Originally Posted by AstCd2

I don't think there's a universal opinion on this. Some people think voices add to the suspension of disbelief and the persona of the characters, while others think it robs them of the chance to 'fill in the gaps' with their own imagination.

Actually, there might be more to it.
A (surprisingly, legal) Russian translation of X-Change 2 came out here, and I picked it up to see what they were up to.
Deinstalled ten minutes later. Never mind the bad translation, the dubbed voices made the game impossible to enjoy in any way other than blindly skipping through to get to the HCG. It's as if the voices were the one thing that kept the game Japanese.
Dubbing it somehow made it worse than renaming all the characters and pretending the story happens somewhere else would.

Originally Posted by AstCd2

I'm not sure what you mean by vectors, here - but in any case, there aren't many visual novels written in flash at all (I can think of maybe two, both of which are doujin games).

By vectors I mean 'images which are defined not in terms of pixels but in terms of curves and fills, specified by abstract spatial coordinates'. Flash is a typical example of those, most typography you see in this day and age is impossible without vectors. Vector-based images allow for very cheap scaling and rotation, among other things (it comes out to just fiddling with a few numbers instead of computationally expensive and bad-looking filling in of pixels which weren't there in the first place).
I would expect that using vectors would drastically reduce the workload (quite a few of the artists I know, especially those who are good at using tablets, swear by them and Adobe Illustrator. The rest swear by OpenCanvas...) involved in producing such a game, not to mention making it much less resource-intensive if used right, but I just don't see any examples beyond the doujin games made in Flash. Which don't even really make the wisest uses of the technology.

Originally Posted by AstCd2

PC limitations don't really come into it, as visual novels don't tend to be demanding on system resource...

Most PCs you'll find in the West (and here) will be able to do 1024x768, have over 600mhz worth of computrons and over 128 megs of ram. I can't help but wonder how much legacy hardware is still in use in Japan though.

Originally Posted by AstCd2

Voices need recording, animation needs drawing, and flags need debugging. Each has an effect on the development cycle, but I wouldn't say that any of them define it, except perhaps in the case of full animation, in which you'd need to think the production through in a different way to a text novel.

Actually, I meant something else. Suppose you've invented a new cool feature (like, possibly, the 'flag-based branching' mentioned earlier) which changes the engine design little but completely changes the structure of the game. That changes the development cycle rather drastically, reducing the workload in some areas, increasing it in others, and in general turns the story around to produce something new and fresh.
Can you remember any such innovations?

[Haeleth]

Originally Posted by Mihara

What exactly are those? Choices midgame which do not create diverging paths until much later in the game, or something else?

Yes; also choices midgame which adjust a value rather than directly affecting the game's outcome (a common system is to assign various numbers of "points" to certain choices, and open up story paths depending on the number of points received).

Originally Posted by Mihara

Vector-based images allow for very cheap scaling and rotation, among other things (it comes out to just fiddling with a few numbers instead of computationally expensive and bad-looking filling in of pixels which weren't there in the first place).
I would expect that using vectors would drastically reduce the workload (quite a few of the artists I know, especially those who are good at using tablets, swear by them and Adobe Illustrator. The rest swear by OpenCanvas...) involved in producing such a game, not to mention making it much less resource-intensive if used right, but I just don't see any examples beyond the doujin games made in Flash. Which don't even really make the wisest uses of the technology.

It's a question of skills. Most artists (with manga/anime training) are more comfortable working with pen and ink than with bezier curves, and most artists find raster shading far, far simpler than the laborious task of tweaking gradient fills to produce the effect they want. It's easy to find people who can produce decent raster graphics in any style you care to mention; it's harder to find people who can produce decent anime-style vector graphics.

Basically, it's hard to sell vectors when the existing staff and the existing market are more comfortable with raster graphics. :)

BTW, there have been one or two attempts at vector-based romance games, e.g. Tokimeki Memorial 3 (PS2). That game was panned, though the unmemorable character designs and certain unpopular changes to gameplay mechanics probably had more to do with that than the graphics.

Originally Posted by Mihara

Suppose you've invented a new cool feature (like, possibly, the 'flag-based branching' mentioned earlier) which changes the engine design little but completely changes the structure of the game. That changes the development cycle rather drastically, reducing the workload in some areas, increasing it in others, and in general turns the story around to produce something new and fresh.
Can you remember any such innovations?

Most of the innovations I can think of have been very much in the area of engine design, with gameplay changes being relatively minor. Things like scrollback and persistent scrollback (i.e. you can review previously viewed text even after quitting and reloading) improve the game experience, but don't really affect the development process: they're added to the engine, and games benefit automatically.

The only thing I can really think of that matches what you're looking for is the recent trend towards engine customisability. Where formerly game engines would be pretty rigid in what they permitted, offering creators maybe one or two text display styles, built-in menus, and a fixed set of transition effects, recent engines have started to become more general-purpose and to allow games using a third-party engine to look and feel as though they use dedicated custom programming. For example, RealLive's extensibility through DLL plugins permitted Tomoyo After to embed a simple RPG, while Majiro's structure (a general-purpose VM with most game logic implemented within the engine) permitted it to be used to write a Mahjong game, and KiriKiri's similar design allowed a wide variety of minigames to be written for Fate/hollow ataraxia.

To continue your comparison with the FPS genre, this is comparable to the increased reliance on game scripting that first appeared with Quake and permitted a much wider variety of gameplay within the FPS framework. Also similarly to FPS games, the full potential of extensible engines is rarely realised, with most visual novel titles, like most FPS titles, still delivering the tried-and-tested game structure that hardcore fans of the genre demand...

[Carl]

Originally Posted by Mihara

Actually, there might be more to it.
A (surprisingly, legal) Russian translation of X-Change 2 came out here, and I picked it up to see what they were up to.
Deinstalled ten minutes later. Never mind the bad translation, the dubbed voices made the game impossible to enjoy in any way other than blindly skipping through to get to the HCG. It's as if the voices were the one thing that kept the game Japanese.
Dubbing it somehow made it worse than renaming all the characters and pretending the story happens somewhere else would.

Sounds like the early days of anime in the West. I guess in an environment where you have little to no knowledge of what works and what doesn't you just have to guess. In Russia, and just about the entire rest of the world but Japan, visual novels are very much in their infancy as a marketable product, and even the compartively more polished examples like Ever17 are still quite a bit rough around the edges. Like anime, that'll change with time (I hope!), but we're not there yet.

Of course, even the Japanese are capable of screwing up their own games without foreign assistance. I love Canvas2, but the PC version is not far off unplayable because of utterly terrible voice acting, such that they completely recast it for the PS2 version and anime adaption.

By vectors I mean 'images which are defined not in terms of pixels but in terms of curves and fills, specified by abstract spatial coordinates'. Flash is a typical example of those, most typography you see in this day and age is impossible without vectors. Vector-based images allow for very cheap scaling and rotation, among other things (it comes out to just fiddling with a few numbers instead of computationally expensive and bad-looking filling in of pixels which weren't there in the first place).
I would expect that using vectors would drastically reduce the workload (quite a few of the artists I know, especially those who are good at using tablets, swear by them and Adobe Illustrator. The rest swear by OpenCanvas...) involved in producing such a game, not to mention making it much less resource-intensive if used right, but I just don't see any examples beyond the doujin games made in Flash. Which don't even really make the wisest uses of the technology.

This is simply a matter of the favoured graphical style of Japanese illustrators. There isn't a lot to it, they just prefer doing it the way they do.

Most PCs you'll find in the West (and here) will be able to do 1024x768, have over 600mhz worth of computrons and over 128 megs of ram. I can't help but wonder how much legacy hardware is still in use in Japan though.

Japan certainly has a lot of fast computers, but bishoujo games have never tended to push computer limits, unlike the American shooting game industry.

Bishoujo games grew up on a very cheap, mainstream computer system made by NEC called the PC88xx series. They were cheap, they were slow, they could display few colours, they lacked features, they were one of the last off the block with a CD drive, but they still became the most popular computer in Japan. The reason is that they kept themselves affordable and got software that people wanted to use. And because more people used them, more software was developed for it, thus leading to an upward cycle in their useage. Bishoujo games grew up on that feature lacking, not particularly powerful computer system, they didn't grow up on the latest speed machine with brand new top-model graphics cards every year, so the developers focused on making things within the technical limits, not pushing the boundaries. As time went by, and they moved over to Windows based machines and new technology, the culture never really changed.

But as most PC gaming in Japan is religated to the niches (bishoujo games, MMORPGs, and FPSs), due to the mainstream titles all being sold on game consoles, and the fairly high cost of PCs over there relative to the income of young people who make up much of the market, not demanding the state of the art is pretty understandable.

Actually, I meant something else. Suppose you've invented a new cool feature (like, possibly, the 'flag-based branching' mentioned earlier) which changes the engine design little but completely changes the structure of the game. That changes the development cycle rather drastically, reducing the workload in some areas, increasing it in others, and in general turns the story around to produce something new and fresh.
Can you remember any such innovations?

There haven't been too many innovations since Leaf began their visual novel series in the late 90s. Voice and RPG elements take quite a bit of programming, but even they are only a moderate change from what went before.

Bishoujo games are really just too niche a market to afford big programming teams and bold ideas.

Originally Posted by Haeleth

BTW, there have been one or two attempts at vector-based romance games, e.g. Tokimeki Memorial 3 (PS2). That game was panned, though the unmemorable character designs and certain unpopular changes to gameplay mechanics probably had more to do with that than the graphics.

I actually really like TM3. It's not without its flaws, but much of the criticism seems a bit exaggerated to me.

It tried to do something original, but was a few years and a game system too early for what they were trying to achieve. In spite of that, they did actually get pretty close to what they were trying to do, so they should have been commended for trying something new, rather than pilloried for rocking the boat if you ask me.

To continue your comparison with the FPS genre, this is comparable to the increased reliance on game scripting that first appeared with Quake and permitted a much wider variety of gameplay within the FPS framework. Also similarly to FPS games, the full potential of extensible engines is rarely realised, with most visual novel titles, like most FPS titles, still delivering the tried-and-tested game structure that hardcore fans of the genre demand...

Well said.

[Agilis]

This is simply a matter of the favoured graphical style of Japanese illustrators. There isn't a lot to it, they just prefer doing it the way they do.

It's not really all flash. At the very least, Hybrid engines seem to be in the works.

Age's rUGP seems to be heading towards vector-y artwork. Looking at Muvluv Alternative, the most beautiful CGs, backgrounds, etc, still tend to be rasters, but a great deal of the character art during the normal parts of the game involves textures and overlays, and the characters zoom in and out from motion and panning too well (near zero 'rough edge' problems) for them to be generic pixel-based standup character art.

And Summer Days, which makes extensive (albeit initially buggy) use of the same version of rUGP as Alternative, makes use of what seems to be a combination of rendered video overlayed on stills to get the "totally animated (but not really)" sort of feel to things.

I liked the way School Days was done better I think. just stacks and stacks of pre-rendered video stored in wmv, then somehow glued together using xml. It made for a rather rigid story and alot of redundant clips, but it certainly played quite well. Summer days probably wanted more flexibility and branching in a smaller payload and went this newer way...

[Mihara]

Originally Posted by Haeleth

Yes; also choices midgame which adjust a value rather than directly affecting the game's outcome (a common system is to assign various numbers of "points" to certain choices, and open up story paths depending on the number of points received).

I wonder, if there's anything else possible in the choice-branching methodology that hasn't been well-explored. Random pathways that only open up half the time have been done and I don't think they do much beyond frustrating the player. Using choice-branching interface to, essentially, disguise a minigame on the outcome of which the real branching decision rests (like in the aforementioned X-Change 2) is just an attempt to fit a square peg in a round hole (just making a round peg instead would work better but wouldn't ultimately change much) and has also been done. Completely splicing a visual novel with another kind of game has also been done. (tactical strategy of Sakura Taisen) Dating simulations like, say, True Love, splice management strategy with a visual novel.
What's left that I missed?

Originally Posted by Haeleth

It's easy to find people who can produce decent raster graphics in any style you care to mention; it's harder to find people who can produce decent anime-style vector graphics.

Actually, I've seen success with manual vector-tracing of inked images, which is less of a skilled work than drawing them that way in the first place and opens up many possibilities, but that's beside the point.

Originally Posted by Haeleth

The only thing I can really think of that matches what you're looking for is the recent trend towards engine customisability.

Wasn't there a trend for making in-house engines since about 2003, though?

The trend for customisability allows for more innovations by the specific developers, but is not an innovation all by itself, actually - they have to decide to use this customisability to do something different, first.

Originally Posted by Carl

Sounds like the early days of anime in the West.

Better than that, usually, since the anime was pushed into the mainstream market by purist fans over here. (...and yours truly has much karmic debt to answer for by being there in the beginning of the whole mess eight years ago. Doing less than I should have still comes back to haunt me.) Translations are normally contracted out to long-time fans and nothing is released without subtitles. The publishers get more flak for messing up DVD covers than anything else, really.
Yet, this is not so with the very few visual novels released to date, for some reason, and I don't like the trend.

Originally Posted by Carl

Bishoujo games grew up on a very cheap, mainstream computer system made by NEC called the PC88xx series.

Oi. I remember picking a dead one apart when working in a lab and wondering what the heck it was. :) Still have the manual lying around somewhere.
Traditions of the medium forming on a specifically limited but very popular machine explains a lot, actually.

Originally Posted by Agilis

Looking at Muvluv Alternative, the most beautiful CGs, backgrounds, etc, still tend to be rasters, but a great deal of the character art during the normal parts of the game involves textures and overlays, and the characters zoom in and out from motion and panning too well (near zero 'rough edge' problems) for them to be generic pixel-based standup character art.

Precisely what I would call 'wise use of vector-based art'. Doing the game entirely in vectors would be counterproductive, since a nicely-shaded raster image converted to a vector representation quadruples storage and memory requirements. :)

[Haeleth]

Originally Posted by Agilis

Looking at Muvluv Alternative, the most beautiful CGs, backgrounds, etc, still tend to be rasters, but a great deal of the character art during the normal parts of the game involves textures and overlays, and the characters zoom in and out from motion and panning too well (near zero 'rough edge' problems) for them to be generic pixel-based standup character art.

I haven't played that, so I can't really comment, but I have to say I'd be surprised if it used vector art; I can't easily find any Japanese sources that suggest it. Raster images can be scaled smoothly - indeed, this is trivial if the engine uses DirectX, since it could just render the raster image as a texture and automatically take advantage of any 3D hardware your computer has to handle scaling and panning.

Which would still be a considerable step forward over the traditional software-based scaling most engines use, of course. :)

[Mihara]

Originally Posted by Haeleth

Raster images can be scaled smoothly...

Scaled down - yes.
Scaled up - not really. :)

[Haeleth]

Originally Posted by Mihara

Scaled down - yes.
Scaled up - not really. :)

That depends on how you define scaling up. Relative to the native resolution of the picture, sure, but there's nothing stopping them including high-resolution versions of the images they wish to zoom in on (and indeed I know for a fact that some games do just that); depending on the exact effects desired, it could easily be simpler than attempting to seamlessly combine raster and vector artwork.

[Mihara]

Originally Posted by Haeleth

That depends on how you define scaling up. Relative to the native resolution of the picture, sure, but there's nothing stopping them including high-resolution versions of the images they wish to zoom in on; it might well prove simpler than attempting to seamlessly combine raster and vector artwork.

It's a processor/memory tradeoff thing, since you have to write the code and get it right only once. Vector artwork will require less memory regardless of what zoom level you want to display it at, but more processing time, high-resolution raster images will require less processing time (especially if you can use a GPU to blit and otherwise mess with it) but much more memory, since you'd have to keep them in memory uncompressed to do anything useful with them. Vectors have one other advantage, though, in that GPUs allow you to do vector calculations much quicker as well. (GPU-using Flash player is about 20 times faster than the original Macromedia one...)

Raster has the advantage of being much easier to port to slow mobile platforms though, which is a major incentive to keep using it. But still, I'm pretty sure at least some images in, say, Yami to Boushi to Hon no Tabibito have been vectors at some stage of production, even though the game exclusively displays rasters.

Actually, I'm surprised that vector-based artwork did not flourish in PS1 era. PS1 had little memory (was it 2 or 4 megs? I forgot. :) ) but a quick GPU.

[Shii]

Originally Posted by Mihara

I doubt there have been serious published studies on the subject of interactive visual novels as a medium, at least, I couldn't find anything offhand. Nevertheless, they are just as much a viable subject for such studies as music (which has trends, genres, styles, all heavily interconnected and independent) or any other kind of art.

Harpers has a round-table article this month about using video games for a literary education, but none of the commentators could imagine the idea of a visual novel so they were slowly circling around it. It was kinda amusing to read actually.

AVRICH: But any game that would teach students real literary writing, with real characters, would have to include some real literary reading. There’s no other way to learn style, tone, the uses of irony.

[Agilis]

Originally Posted by Haeleth

I haven't played that, so I can't really comment, but I have to say I'd be surprised if it used vector art; I can't easily find any Japanese sources that suggest it. Raster images can be scaled smoothly - indeed, this is trivial if the engine uses DirectX, since it could just render the raster image as a texture and automatically take advantage of any 3D hardware your computer has to handle scaling and panning.

Which would still be a considerable step forward over the traditional software-based scaling most engines use, of course. :)

Well, I only conjecture that they're using vectors. From what I remember, in Age's home page staff room, around the time the CD version of Alternative was being released, one of the staffers mentioned that "the textures are of lesser quality than the DVD version". If you'll note the DVD spec version takes up a few hundred more MB than the CD version specs. At least, it did when I last looked around may. The post is still there probably if anyone thinks it's important enough to analyze.

It Could mean that they build characters with really really big 'textures' of raster pieces. like a face is a stack of eyes, eyebrows, hair, mouth. it just seems rather counter productive to do so without vectors.

^_^ The way that the senjutsuki (the giant robots) move around on screen, much like stop-motion animation with paper cutouts certainly suggests vectoring of some sort.

[Haeleth]

It's quite possible, and would certainly be a sensible approach to the issue for the reasons Mihara has noted. I'm only skeptical because I haven't seen it elsewhere, but that's not to say Age aren't doing it. Sadly there doesn't seem to be a demo, so I can't easily see for myself... :/

[Mihara]

Originally Posted by Shii

Harpers has a round-table article this month about using video games for a literary education, but none of the commentators could imagine the idea of a visual novel so they were slowly circling around it.

...you make me think I could actually get away with a few papers about it, or possibly, get a grant. :)