Categories
Journals

Some exploratory sculpts and sketches of Ktish, the main character of The Quiet One.

Yeah, drawing is not my strongest skill. 🙂

Categories
Journals

Constructing languages

So. A fake language to sell a fake universe. How does one go about doing that?

Well, it helps to have a script and a few sentences to translate. I’ve got a script and a few sentences to translate.

What doesn’t help is to make up a bunch of arcane unfollowable self-contradictory rules of syntax, a spelling system that is fully regular but horribly ugly, and a bank of phonemes (the distinct sounds of the language) which lead to unpronounceable car crashes.

Unless of course you’re doing what I did and doing that sort of thing as a doomed first draft. Then it’s fine. It helps to get the nutbag craziness out of the system, and throwing it out to start again feels utterly cleansing. I hit that blissful point yesterday evening.

So. Any given language is the solution to the universal problem: how do I convey something to this other person with an idea-level degree of precision by flapping my mouth, tongue and larynx at him? Every language on Earth solves this problem uniquely, in fact, every person on Earth solves this problem uniquely with versions of languages that solve the problem uniquely. The solutions are unique based on words (lexicon), word structure (morphology), sentence structure (syntax), the meanings of those words and sentences (semantics), the sounds that make those words up (phonetics) and how those sounds interact (phonology), who is speaking and being spoken to (sociolinguistics).. and that’s just predominantly synchronic aspects of language – aspects that don’t take history into account so much.

All a constructed language has to do in the context of a film is fake the audience out and sell the idea of a broader universe. In making a genuine language as opposed to faking it, this means many problems in the above paragraph have to be solved manually – either you make up novel solutions, or you borrow solutions freely from other languages (the saner option by far). So far I can say that I’ve consciously researched Kaytetye, Noongar, Inuktitut and Putonghua for their solutions. Probably Finnish and Japanese as well. I’m always stealing stuff from those languages.

I’m putting lots of attention towards constructing the language of the main character because there’s a ton of dialogue in it in the first screenplay (fifty-two whole sentences to translate), and lots more to come. There’s lots of that language to build, which means it’ll get somewhat unmanageable pretty fast, so I’m better off getting to know it really well as soon as I can.

I’ve started from two points – phonology and Roman alphabet orthography (so I can write the words down) and syntactic structure.

Picking the right sounds is important – the sounds are the surface level of the language, the most obvious interface. The main character doesn’t have fleshy lips to pucker in order to make sounds like w and m and f and v, but he can pull in the corners of his mouth to create an open o and a sort of ö sound like in German. This means there’s a lot more mid-mouth and guttural sounds.

The best way I know to figure out what direction to go with the phonetic side of things is just to babble randomly, in character. Sometimes I come upon a non-word I like: ktullu, nkul, trrh, derche. These fit my instinctive expectation of how this main character sounds – short, staccato, gruff.

I tend to do this part of it alone, if you’re wondering. 🙂

The syntactic structure work usually involves taking sentences and breaking them down into placeholder units in constructed language order. Knowing how that order goes first helps immensely.

One of the sentences from the screenplay that needs translating is: “They never miss a chance to make us look stupid." It’s a deceptively complicated sentence – there’s "us” looking stupid, then “we” are causally being made to look stupid, then there’s a specific kind of chance in which “we’re” made to look stupid, and then on top of that there’s “them” missing that chance. It’s highly embedded and frankly not the best sentence to kick off with. Eventually I glossed it as:

stupid-ATTRIB [they]_ABS, always stupid+appear-1p_EXCL-CC_STATE result+cause-CLAUSE_CLITIC happy-HABITUAL-3p_SENTIENT

Or, glossed back into English:

stupid them, always stupid-looking-us occurring-making happy they are

Eh, it’s a work in progress,. Ts`ikla yododoklé-rza. (It looks a bit stupid.) Glosses are never flattering either.

But I know I’m on the right track when speaking in it flows off the tongue naturally, when it invites expressive intonation, when it rolls along with the style of its non-artificial cousins like a speech system that could have been around for thousands of years – instead of sounding like a crazy unpronounceable experiment dreamt up by a mad linguist with too much time on his hands. Occasionally, in bursts, it’s sounding that way.

Incidentally, “the quiet one” in this language is ktii’rsah – literally “he is taciturn” – which makes the main character’s name something like Kteesha in English spelling. For now I’m just calling him Ktish.

Categories
Journals

I had a hankering to try building a test character using implicit surfaces since that’s what that game Spore used.

Basically instead of describing a surface with verts or NURBS, you use of fields which generate spheres, capsules, boxes and whatnot. When the fields get close enough together, they start to gloop together to create a continuous surface. These objects are known in Blender (and probably in many other 3D apps) as metaballs whether or not they’re ball-shaped.

In Blender at least, metaballs are something of a pain in the arse to turn into characters – you have to parent individual metaballs to individual bones instead of just the usual Armature modifier, and there seem to be weird little rules about what you can and can’t do with them.

There’s an especially weird trap where they are influence-grouped by their name (“family”). One ball becomes the master ball for that “family” and other balls named after it will interact with other members of that family.

Metaballs of the same family all have the same material and threshold of influence. If you want to put a metaball in a different family, you have to rename the object. You can have multiple metaballs per metaball object – the nose and horns on this guy are all one metaball object.

Metaballs can be positive or negative. Negative balls can sculpt away positive balls. As with many things in Blender, extra fun is often to be found by hitting TAB with objects selected. 🙂

So. Not as nice to work with as verts, but as you can see you can make metaballs move around and look vaguely characterish. Even if they’ve got no arms. (The problem with walk cycles is that you can spend all day tweaking them and I have to go to bed at some point. Hence why this sucks a bit.)

Also, IK necks – how good are they? Damn good. Definitely doing more of those.

Barcelona Rooftop HDR background image from here. (sIBL Archives.)

The constructed language has had a first pass in interlinear text. So instead of rushing into writing down actual words, I’m writing down glossed grammatical structures like ART-3s_SENT [planet]-GOAL [far away] fly-PASTPERF-3s_SENT. This lets me sketch the language at a high level without having to invent words or even choose the range of sounds that make up the words first.

I don’t know how this first language fits together well enough to actually translate all the dialogue in the screenplay yet, so I have to go back and decide that some time this week. I’d probably have figured it out tonight if I hadn’t been messing around with metaballs. 🙂

Categories
Journals

This weekend I worked on the main character’s abugida a bit as well as doing some technical tests in Blender around replacement animation.

The abugida as you can see is mostly straight lines. I got the idea that it needs to be carved into unwilling surfaces with brute force so sweeping curves didn’t seem to really suit it. So far no part of any glyph takes more than five strokes to write. It’s feeling a lot more like the Korean alphabet than a genuine abugida at this point but then I guess the language I’m going to be spelling out in it isn’t anywhere close to ready yet. I want to start working up a grammar over the next few days based on the dialogue that’s already in the screenplay, figure out the translation in the abstract, then finally drop some actual words into placeholders and see how they all sound.

That’s really the end goal with the conlangs in this series – being both believable and cool to see/hear. In context, the artificial languages and scripts have a single job to do: sell the notion of richness in the universe of the story. Once they sell that notion, further polish is pretty much unnecessary. If I don’t need a word for “microchip” or a compact syntactic device to indicate that something is possibly to blame for something else, I won’t invent it. No point. Too much else to do. 🙂

So that was yesterday. Today I was feeling a bit under the weather but I loaded up Blender and had a mess around anyway.

The technical tests were mainly to get an idea of how I might go about doing replacement animation. Replacement animation is used extensively in limited animation and stop-motion but not so much in CG. The classic use for it is lipsync in stop-motion with a non-malleable medium – instead of moving the character’s face on a tiny little armature, you just whip one head/face off and put on another to get a different mouth shape for the next frame. In Blender.. well.. I’ll get back to you how it’s possible. But it is. Probably. 🙂

And if it isn’t, well.. I’ll figure something out. It’s certainly not coming easily.

I tried using the Boolean modifier to link two shapes together but it’s honestly unpredictable and didn’t really play well with the Armature modifier. The objects would randomly disappear one at a time based on rotation. Very weird.

I tried hooking vertices together as an experiment but that didn’t form a nice continuous seam, especially not when the Subsurf modifier put its two cents in.

Finally i tried Shrinkwrapping one set of verts to another but there’s not a continuous mesh anymore – annoying joins ahoy. You can see one pretty clearly starting at the bottom corner of the eye and curving around the cheek.

The end result of these tests is that I need to have more tests. That’s what tests are for.

Categories
Journals

A refined game plan and progress report

There’s a slight tweak to the plan now: it’s still going to be a series of shorts made more or less in story order. I’ll generally be attempting to work in ascending order of difficulty with chosen technical targets to achieve as I go.

I’m in pre-production on the first short now. I’ve got soft script lock – the assets (characters, sets, etc) are pretty much set even if the dialogue isn’t. It’s far enough along to allow me think about production design, conlangs and conscripts, character design and other development/pre-production concerns alongside dialogue polish.

Production design and art direction in particular are kind of daunting. I’m not a trained designer so instinctively that makes me want to work with as few elements as possible – at least at first. The coming to me in disjointed abstractions – concrete, dug-out rock and frosted glass/plastic are a start. Rounded corners. I want something iridescent in there as well.

I’ve been researching architecture, specifically high-density living. I’m casting around for somewhere high-tech but not extravagant but not derelict to put my main character as a living space. I might find more inspiration in Japan or maybe Korea than Hong Kong – the clutter in some of those little apartments in HK is utterly amazing.

To give myself a better sense of what virtual materials I’ve got to work with and what sort of rendering times I can expect if I use them, I’m jumping into the Cycles rendering engine and either brushing up on procedural texturing or just generally playing around. (Speaking of Cycles, Blender 2.69 is in release candidate phase.) I guess materials and textures fall under post-production moreso than pre-production, but if something’s utterly impractical because it means I’m waiting 30+ minutes per FHD frame for rendering, I’d prefer to know ahead of time. Also, discovering cool new things that the technology is capable of is often inspiring in and of itself.

Like knowing I can get nifty procedural iridescent insect chitin without a huge amount of waiting, that’s pretty cool.

The conlanging – language construction – is off to a fairly casual start. The first word I made up for the main character to say is trr. It’s his “contemptful noise” equivalent of pft because he can’t say pft because he hasn’t got lips that work like that. The nice thing about doing the voice work myself is that I don’t have to train an actor how to do a uvularised voiceless alveolar plosive with lateral release – basically a cross between a T, an L and hocking up phlegm. It’s hard to even write using IPA – probably tχˡ comes close – so a tailor-made abugida is in the works too. So far I have a general idea of what I’m doing – I have a symbol for the letter r in onset, nucleus and coda positions. I can thus write the hypothetical word *rrr. Very useful.

So yeah, the next couple of days will be reading up some more about production design, doing another pass on the script to make it funnier, researching compact living spaces, sussing out procedural materials/textures in Cycles some more, and generally trying not to lose too much sleep.

Finally, have a picture of a mulgara.