PERSONAL

0

2,004

Project Progress and Making Faces

5 years ago3,083 words

I've done some work on one of the ideas I described last time!

I've had eight radiotherapy sessions now... 22 left. They're still making me anxious, and tired, and the journey is annoyingly eating up chunks of my time that I'd rather be spending on games. I won't go on about that this time since it's much the same as it was before. No major symptoms yet though.

Oh, I will say though that the radiology staff are doing all they can to accommodate my anxiety, which is nice... though it feels patronising at times, like when they congratulate me for "doing really well" like a child or dog. I suppose that's the risk of making your struggles clear.

Also, one of the guys on the staff today was anxious too, and I could tell as soon as I saw him. He confirmed it after the scan - he told me he struggles with anxiety as well - and while I found that memorable and nice since I'd rather be around someone who gets it from personal experience, it also made me think about how hidden it isn't... How a person's anxiety is detectable from first impressions, and affects further impressions. Not really something you can just magically change though, especially when the anxiety occurs despite your thoughts rather than as a result of them. Feels like something not within your control.



Last time, I talked about wanting to make some shorter games, and three ideas for things I could make. I refrained from reading the comments for several days in case they were discouraging... They rarely are these days, but I suppose that's a remnant of the past, that fear, and when the various physical and mental brain issues are weighing me down, the idea of any potential discouragement feels daunting. I can't exactly exist in a peaceful bubble, blocking out all criticism, and hope to succeed... I get that. And it bothers me that I take so long to get around to other people's words. Stupid mental issues. Grumble.

I was surprised though to see a couple of you showing interest in my idea for a game based on associative visual memory. I thought it wouldn't really work because interaction is minimal (you're mostly just watching), but perhaps I should experiment with it and see what I can come up with. I might do that over the coming days.

For now, though, I've been working on the first idea I talked about, the one where you're presented random people, who you befriend (or comically annoy) by talking to them. I suppose it's sort of like a choose-your-own-adventure thing, though each individual person is a micro-story rather than there being one grand overarching plot.

I've spent two or three days working on it (though not as much of those days as I'd like due to aforementioned brain issues, but I'm getting somewhere at least), and I've already got the major mechanics working. It's playable, to the extent that it'd need to be. There are three major things I still need to focus on though, which are going to take a lot longer to establish: the point of the game, the writing, and the graphics.

Why exactly are you talking to these people? People don't talk in a vacuum; there always has to be a reason for meetings to occur at all. And if you as the player are meeting with all these random people who talk about their personalities to you, why?

I wondered whether to have the player be some kind of therapist or something similar, and each person is telling you about their problems, and you have to help them by talking... but what, would they all have mental disorders, then? They could be invented, silly disorders, perhaps. Or maybe because loneliness is so rife in this internetly-interconnected era, you as the player have set up a 'make a new friend!' stall somewhere, and random strangers approach you to try and achieve that. Or maybe that, but it's set a couple of decades from now, commenting on how if we continue in this direction, people will be that desperate for companionship. Or maybe you're choosing a date. Or maybe you're a boss choosing employees. Or maybe you're interviewing people about some kind of overarching story involving some paranormal goings-on... and you need to win their trust to earn clues. Or maybe you're deciding whether or not to let people into the afterlife, trying to understand them at their core. They could tell you about the life they just lived, which could be amusing if life events are randomly generated; just hearing about a ridiculous life could add an extra layer of humour to the experience. There are many possibilities. I need to decide on one, since it'll determine everything that comes after.

Perhaps it could be a comment about relatable social media, and how the 'friends' people have on Facebook aren't really real friends, much of the time. You can have people you've never even talked to on there, just to get your count up. In this game, it could be the future (or an alternate reality), where in order to make it so the friends counter only applies to real friends, pressing a 'Like' button on someone's profile instantly teleports them to you, from wherever they were, at which point they're expected to strike up a conversation with you, to get to a point where you both agree you know and like each other and can be 'true friends', increasing both of your Friends counters. It's silly, of course, but I feel like it'd have some relatability to it, like being funny on the surface while hinting at a deeper issue a lot of people might have given thought to. Also, the people you randomly warp to you could be alarmed by that, could be in a towel after having just got out of the shower, or asleep in their pyjamas, or they could be a surgeon in the middle of an operation, covered in blood; silly things like that. Could be commentary about how phones allow us to instantly message people regardless of what they're doing, and how silly that'd be if we could communicate face-to-face with them just as instantly and at any time. (Teleporting in would also solve a technical issue as to how the conversations could visually start and end, and increasing your Friends count might be a relatable goal. Perhaps the higher it got, the more people would immediately respect you. Perhaps you could also choose between randomised profiles, giving some greater degree of choice and giving people expectations - which could be comically broken - before they meet each person. Or maybe I could even add a selection of pre-written characters rather than doing the randomiser thing? Though that seems to go against the whole point of the thing...)

So yes, I have many ideas for the setting, but have yet to settle on one. (The one I went into more detail about seems most appealing to me at this point though.)

The writing will probably be a bit of an ordeal. Currently it works, but... Well, each person has three personality 'runes', using the system that I've used in previous games. I was going to use ∞ the Big Five ∞, but I felt that conscientiousness wouldn't really determine much in conversation anyway, and introversion and neuroticism can be blended to a degree to form the Grave/Jolly dichotomy. It's not perfectly accurate to real life (there are neurotic extroverts and calm introverts), but it's easier to make. Players wouldn't even see the runes, really, so no explicit, confusing explanation of them is necessary; they'd have to use their judgement to guess how the person might respond to things based on what they've already said. (Sort of like how most people can play Pokemon without an awareness of EVs and IVs.)

It's interesting that procedural generation is common in games these days, but it typically focuses on the superficial, the body; I've not seen other examples that try to generate distinct trait-based personalities randomly like this. I think things like The Sims allow for personalities, but I think that's more things like "is Evil" rather than "20% introverted"... But I don't know.

All the dialogue will exist in a single text file, and here's a bit of how it looks at the moment:



Simple, right?! That's just for them to say 'hello'... and the way in which they say it is determined by their runes. I've already got the code working, as I said (it involved a whole lot of ∞ Regular Expressions ∞), and I've got a working version where I can move through a sample conversation - including selectable options - and the person's runes determine the outcomes properly. But to write the entire thing will require quite a bit of work, to account for all the possibilities! It seems fun though, and probably will be once I get started. The main issue now is determining the setting, since that'll determine what I need to write.

The graphics are my main focus at the moment, though. You'll be interacting with humans in this, so it's important that I get them looking right, and it's so easy to get them looking not-quite-right!

...Originally, I'd written loooooads here about my experience making faces, both in 2D and 3D, because I personally think about that kind of thing all the time, and find it interesting looking back at my progress, relearning lessons along the way. I had dozens of screenshots and everything! It was taking too long to write though, and I don't know how interesting it'd be to read, since it took me ages to get to the main point. Most of what I showed, I've shown elsewhere before, but I've been synthesising a lot of it in my mind recently so I tried to put that down on digital paper. But it was getting too long and eating up too much time. For now, I'll just focus on the graphics I've got at the moment, without delving deeply into their origins.

What I have at the moment looks like this:



...That's four screenshots side by side, obviously.

The character can emote with their face, but I've yet to add different body poses.

I've only got this one ("naked") human female model so far, which I feel okay about, I think... It's not especially stylised, but it's not exactly realistic either. It's the latest product of my journey through the realm of 3D modelling, and I think it's an improvement on a lot of stuff that came before.



I'm quite happy with the topology and arrangement of polygons, though obviously it's just a step in that journey, and I'm a long way away from the mastery of people who do this professionally, and have done for years.

The shoulders were a particular concern, and I was determined to get them as acceptable as I could just because my previous models suffered in that area. Shoulders are an absolute nightmare to get right in 3D though because of how the bones and muscles move, and how vertices shift as a result of 3D skinning and bone movement. They're okay like this, but when the arm lifts up while facing forward, the vertices become a tangled mess and the form warps badly. Usually a bunch of different bones and blend shapes / shape keys or whatever are used to counteract this, but I don't want to include too many controller bones because I'm trying to be efficient. I'm using an extra deltoid bone on top of the bicep one, but... this is all unnecessarily technical! The whole thing is a pain anyway, though I suppose the person who's just looking at the model rather than making it won't have to think or worry about any of that. I've been looking at shoulders in every other 3D thing I see though, and I notice distortion even in those quite often, and the attempts to compensate for it... I could write an essay about all that, but I doubt it's interesting to anyone but me!

(Originally I made models in a T pose - which seems to have become memetic among the kiddiwinks, the successor of dabbing, or something; strange - but that pose is so unlike common standing poses that shoulders modelled in that way are certain to be distorted in the majority of poses. I've had my best results modelling with the arms mostly by the side, as they would be most of the time anyway, though I imagine other modellers would have different experiences.)

My main concern here though is the face.



Does that look okay to you? I suppose it's difficult to tell with the wireframe on top...



Attractiveness is a huge deal, but it's based around subtle proportions which, if done ever-so-slightly incorrectly, can create an ugly or unnerving effect. Learning the proportions of the face is something any artist worth their salt has to do if they want to do portraits, and I've gone through all that myself over the past few years, but even knowing how faces 'should' look doesn't mean you can get them right all the time. Or I can't, anyway. Since I'll be using this as a base from which to eventually derive genetic variation (all the faces look the same now, but won't eventually), it feels important to get it right, but it's difficult to know whether I have or not. It looks slightly strange to me.

Average faces are known to be particularly attractive (but they're not the most attractive; there's a bunch of research about this), and these averages can be derived by taking the composite of a bunch of different face photos. There's a website ∞ here ∞ with a selection of photos, which allows you to generate an average based on whichever photos you choose. I chose the female faces I found most appealing, and this average was generated:



I'm amazed at how that technology is able to do that; I don't understand how it does!

Also, I raised an eyebrow at how... interesting most of the faces on that site look. I've recently written about incels, since I've been reading the content in their domains from time to time, and as a group, they're obsessed with looks (it's one of the reasons I find them interesting, actually, because I'm curious to learn as much as I can about faces). But they speak as if the majority of people are good-looking, or close to it, and I think that samples of photos of people like this show that that's very much not the case. I don't find any of them especially alluring myself. Perhaps it's all about context, though; robbing these faces of context or expression probably has a big impact on how they're perceived.

Anyway, I used that composite photo as a base for this head I made, though the eyes are deliberately bigger. The result matches the source to a large degree...



...But as I said, something seems strange about it to me, off. Like the head is too tall and narrow or something. I suspect it's because the eyes are too large and too high up, and the photo's jaw is wider in an asymmetric way (the model's is obviously narrower in this image, but on the hidden side, the jawline matches).

Neotenous features are attractive in females, so perhaps I'll need to veer away from real proportions in that direction (which tends to better match the result you get if you ask people to create their ideal face anyway, studies have found, and most art goes in this direction).



The difference here is subtle - I've reduced the size of the lower face by about 10-20% - but I think the perception of the left one feels more natural, easier on the eyes, even though it's less realistic in terms of the supposed 'ideal' or averaged proportions. Interesting, how that works.

Also, I imagine the paint job - the texture - makes an enormous difference to perception. For example:



This is just a rudimentary texture that I'm using for testing, but already it makes a huge difference compared to the other mesh, which is identical in terms of vertices but just has the texture turned off. Much of how a model feels is down to its texture, and if I put more effort into this one, I'll hopefully be able to produce better results. Lighting makes a big difference too, and I'll need to play around with that in the Unity project.

Things like this are on my mind constantly, which is why I'm talking about them, but I suppose they're largely irrelevant to the players. Do most people consider the arrangement of lights or the topology of shoulders in, say, Breath of the Wild? I wouldn't at all naturally since it wouldn't be necessary and distracts from the immersion, but I've been on the lookout for how more experienced developers handle these issues while playing any games myself recently, to see if I can learn anything to improve my own abilities.



But yes, enough rambling about technicalities for now. I'm glad that project is coming along, to a degree; I just need to decide on the setting, write the writing, and improve the graphics.

I might put that project on hold for a bit, though, and focus on the memory idea for a bit. "Dreamember", "Dreamemory", "Dreamind", something like that, maybe (I think I prefer Dreamind). I'm unsure whether to do it in 2D or 3D (I'd been picturing 2D in my mind, but it might actually be more difficult), but... I'll do some brainstorming and see what I can come up with.

I'll keep you posted about how it's coming along, anyway!