Quoting the posts in the order they were posted
Top Hat said:Quite obviously, there's a big difference between what the mass-market is offering as role-playing games and what the general consensus of the RPG Codex seems to be. Naturally, game companies want to make games that have a lot of flashy features to sell their games, whereas we want games which are a little more meaty.
So, what ways are there for us to ensure that (1) we have games that are marketable to a wide audience but (2) these features don't take valuable time/money/other resources away from making a good game.
The first thing that comes to mind is voice acting. As it stands now, this severely limits the amount of dialogue that can appear in a game. Voice acting also limits future modding potential and makes it harder to make expansions (since the cost of production will increase). However, having selective voice acting for only main people isn't a big draw that companies can use.
The only way out of this problem, I feel, is to have completely computer-generated voices. Now, I haven't been following this too closely, but from my limited experience, the voices often sound too "robotic" to reasonably resemble human voices. One way of working around this is to, well, make all the characters robots.
This also gives a reasonable explanation in regards to the "immortal character" nonsense, where characters can "die" in battle, but are brought back to life if their side wins: essentially, a character can "back themselves up" before any big battles and simply download their memories/skills/etc into a new shell.
You can also add the much-touted character customization, which will work in a much easier way: instead of having to design a good-looking person (when tweaking can destroy the aesthetics of a face if something is a bit off) you merely have to plunk down bits onto a flat surface. You also get more customization: give your character five eyes or two noses or something - then this can improve/detract from the relevant skills.
Naturally, of course, computerized voices will improve. Then we can have voices that sound exactly like the designers want them to sound, instead of exactly like some actor. If named actors are a big draw, then just have them voice a few roles, letting the computer generate the rest.
Well, that's one idea. I'm sure you've got your own.
Lesifoere said:Top Hat said:The only way out of this problem, I feel, is to have completely computer-generated voices. Now, I haven't been following this too closely, but from my limited experience, the voices often sound too "robotic" to reasonably resemble human voices. One way of working around this is to, well, make all the characters robots.
I'm intrigued by your ideas and wish to subscribe to your newsletter.
cardtrick said:Top Hat said:The only way out of this problem, I feel, is to have completely computer-generated voices. Now, I haven't been following this too closely, but from my limited experience, the voices often sound too "robotic" to reasonably resemble human voices. One way of working around this is to, well, make all the characters robots.
This is one of the best paragraphs on the internet.
skyway said:This is one of the best paragraphs on the internet.
yes, I LOL'd too. Though there is an AT&T's software that generates human voice very precisely with emotions and all that. I actually played some game, where a voicing part was done using this technology (only a small part though).
Nog Robbin said:Voice synthesis is improving. I read an article I believe in the Focus magazine (checked their site but it has no search facility) about how close it could be to being able to simulate voices, inflection and all, from a person after recording just a few words. Even if it's close to human sounding I'm sure it would be good enough for gaming purposes. However, you still have the issue of generating the speech - I'm sure they wouldn't include the tools to produce the speech with a construction kit and I would have thought for the time being that sound files would still be used rather than simple speech to text (where inflection and emotion is harder to encode simply.)
The robot idea is really not bad at all though and gets around current limitations.
TalesFromTheCrypt said:This is fucking sig material
Top Hat said:Musing #2, or Graphics - You've been PVN'D
Quite obviously, the big problem with modern games is the push for high-level graphics, most of the time at the expense of the part of the game that makes it, well, a game.
I'm going to pick on a game - Heroes of Might and Magic V. Apart from the utterly atrocious storyline, one thing bugs me (and not only me - a relative who has the game complained about this as well): because of the three-dimensional engine used in the game, it has become very difficult in some instances to see a treasure chest, etc. Granted, you can move the camera. But really, I shouldn't have to. The earlier games (at least the first four (including King's Bounty)) managed to convey all pertinent information without being bogged down in graphical splendour. This is an example of how NOT to implement fancy graphics.
However, this isn't a rant about what's wrong with modern games. So, the big question is: how can we make elaborate graphics *work* for us? How can we implement high-level graphics in a way that enhances, instead of detracting from, a game? How can we make Bloom sing for its supper, as far as quality gaming is concerned?
I suppose the first question to ask is: can you have a game that is both graphics-intensive and pushes you intellectually? The answer to this is yes: Myst was a game that sold ridiculous numbers of copies on its graphics alone; yet I doubt whether a majority of that game's buyers have actually finished it, given the fact that you actually have to think to finish the game.
I suppose the next argument is: so graphics and puzzles work well. But what about story? Well, to be honest, the Myst series isn't known for its intricate storylines - they are quite linear (until the end, where tere are often a number of choices to make - although there's usually one obvious optimal ending). However, since there aren't so many characters standing around telling you about things, we often get the story through the various books written by the characters.
This set me on to the idea of Passive Visual Narration (PVN): essentially, we use the awesome graphics engines to tell our story without having to resort to character exposition. Let's try an example: and, yes, it will involve robots again. Bear with me.
Now, unless there's some alternative universe or biomechanic basis responsible for our robots, they're going to have to be built as some point in the past, by some kind of intelligent life. Let's suppose that they were built by humans who all died or vanished somehow - yet the robots are still running and they've left artifacts of their passing behind.
For those who've played the game, it's kind of like the Dwemer in Morrowind. However, let's "level up" the idea a bit. First, let's forget about the "Dwemer ex Machina" (he he he) and make it so that there aren't any people around. Let's also suppose that there isn't any "official robot records" of what happened to their creators. So, basically, the only way to learn about their creators is to study what they left behind.
For instance, old newspapers and magazines left behind might report on the rampant supervirus/space exodus from a polluted world/global warming (!) horror that drove away/wiped out mankind. But how do the robots learn what this gibberish means? Maybe they have to jerry-rig some kind of scanner to convert the text to some computationally understood format. Then, once they learned the language, they might be able to decipher less scanner-friendly texts from personal diaries, video and audio clips and so forth.
Analyzing the rubbish and other remnants can tell our robot friends what their creators ate and drank, what environments they liked, what they did for fun, etc. Then maybe some robots think it might bring their creators back if they remade the world in this way, setting off some quests or something.
So, instead of having to just go around, chopping your way through dialogue trees until only the dead stump of "Rumors" is left, you actually learn about the world in the way that archaeologists learn about ancient civilizations. This way, if you were insisting on voicing all your characters with prerecordings, you could save some exposition by letting a picture tell a thousand words. You also cut down on the work of the writers (not the people that come up with the story, but the people who actually write all the stuff), which means that their stuff should be better. And, since you're paying the graphics people anyway, you might as well make their work more integral to the story than just representing who's where and what they're doing.
denizsi said:About voice acting, I've been thinking about prerecorded voice samples re-synthesized real-time. There are already software doing this (gaining different and realistic voice samples from the same source samples) to acceptable extents. The idea is that 10-50 (or more or less, depending on the words) different pronounciations of thousands of single words would be recorded, like "you, you!, you?", including characteristic variations, ie. there can be difference on voice characteristics between two people giving the same verbal reactions.
A research to determine which words to have different accents, stressing and emphasis as used in common sentences would be in order obviously.
A text interpretation / parser engine with grammar rules with its own coded language would be used to voice the text, so you could either write with the codes and test voices as you write or import a complete text and work word by word for right samples.
For instance, say you want to write this:
"So, of all people in the world, you are the only wise one to suggest his, eh?"
As you write it, program auto-inserts suggested source sample numbers in front of the word, which dynamically change as you keep writing. Eg:
So², of³ all² people¹² in¹ the¹² world³², you¹ are¹¹ the¹² only¹ wise¹³ one¹ to¹ suggest²³ his¹³, eh?¹
You can add further details like accent, emphasis, stressing, stretching etc., eg:
So²., of³ ++all²^ people¹² in¹ the¹² world³², *you¹* are¹¹ the¹² +only¹^ !wise¹³ one¹ to¹ suggest²³ this¹³, eh?¹%
to get it voiced like however you imagine the following might have sounded:
"So.. of AAALL people in the world, *YOU* are the ONLY one to suggest this, ehhh?"
The game would only display the text itself without codes, and you could immediately test any text for voicing at any moment to optimize it efficiently as you go. In fact, depending on the game API, you could modify, test and save voice resampling even in-game.
Additionally, you could import this text into the framework, where the text would be parsed for each word with respect to punctuations, noting the grammar rules, and then you could click on every word to bring a pop-up / pull-down menu to choose sample variations suggested by the program with options for accent, stressing, emphasis etc., or experiment with your own settings.
Finally, before or after you write any text for any specific character, would play with the individual synthesizer settings to come up with a unique or fitting voice type for that character. Game then, would resample the source samples real-time in respect to the coded text and the re-synthesizing
As far as I know, this is completely doable with today's tech. I don't know how efficient and presentable it might turn out in the end, but judging by the available professional sound software, it should be pretty good, and the kind of examples I offered look quite functional to me.