Here's What the Designers of SL 2's Avatars Can Learn from Second Life's Own Dreadful Defaults

Janine "Iris Ophelia" Hawkins' ongoing review of gaming and virtual world style

A few days ago, Hamlet and I were discussing Second Life's avatars and, more specifically, how SL 2 could do potentially do better. Second Life's current avatar is so unpleasant that players have been working against it and its limitations almost from the very beginning, so there must certainly be lessons to learn from it, right? A key to how Linden Lab might "get it right" this time?

Well there is, and it's surprisingly simple...

Believe it or not, I don't think the current Second Life avatar is bad. Not on it's own. The problem is that it's very old, probably about 12 or even 13 years depending on when in the development cycle it was created. That's old enough to be in middle school. If you compare it to avatars from other 12 year old games, MMOs in particular, Second Life's default avatar is actually quite nice. Look up some vanilla Morrowind characters if you don't believe me when I say that Second Life did okay considering how ambitious and flexible the systems they implemented were. The original hair is and always was hideous, I suppose, but I'm not even willing to call that outright bad. If you can find another game with a customizable, shapeable hair system for comparison then we'll talk, but until then... Let's just call it unflattering and move on.

So the problem as far as I'm concerned is not some inherent flaw in the default avatars built-in by a designer who didn't know any better -- it's not any on feature that we can point to as proof that they were phoning it in. User avatar innovation, from skins to prim hair and on, has not been driven on because the avatar is bad, but because it's old. Game graphics move fast, and the cutting-edge processor-shattering glory of today is the garbagey eyesore of tomorrow. This change happens on such an accelerated scale that a game like Skyrim, released just three years ago, can't quite hold up without a plethora of player-made mods to bring it up to snuff. Much like Second Life.

So what's the key to making SL 2's avatar a step up from its predecessor? Flexibility. Flexibility for users to modify and innovate, and just as importantly flexibility for the developers themselves to make updates without breaking all existing user content. What content creators can make is all well and good, but the ephemeral potential for a gorgeous avatar doesn't carry much weight for the new player faced with a seemingly insurmountable lump of hideous clay. If SL 2 is committed to mesh clothing then permanent changes to the default avatar will be much easier than they would be in Second Life right now, simply because it's easier to update a model without changing its skeleton (the structure that allows a model to be animated and that mesh clothes are rigged to) than it is to update a model without changing its UV map (the texture overlay that makes up its "skin".) 

Technical talk aside, it boils down to this: Any avatar left untouched for years will only slip further and further away from the graphical standards it was designed to meet. We're reaching an unprecedented point where online games are being actively run for decades, and developers should prepare for the inevitability of visual updates if they want to stay relevant and appealing to new players and not just a dwindling number of nostalgic oldbies. Even World of Warcraft is embracing this approach.

Realistically, the developers of Second Life's default avatars probably weren't thinking about how they would be seen in 12 years, but the developers of SL 2's avatars would be wise to take the outcome of that oversight into account.

Tweet

Iris Ophelia (@bleatingheart, Janine Hawkins IRL) has been featured in the New York Times, and has spoken about SL-based design at the Fashion Institute of Technology in Manhattan and with pop culture/fashion maven Johanna Blakley.

Comments

[Autumn Ashdene]

Awesome thoughts here, Iris! Here's hoping that SL 2 learns from what many mistakes SL has made along the way!

[Pussycat Catnap]

I can pull out Possette from Poser 4 in the late 1990s and show you all the ways, in detail, that the SL avatar has been bad from day one.

And I can take Possette into a 3D modeling application and scale her polygon count down to whatever SL's is - and she will STILL be worlds better.

And Possette is considered a BAD 3D model...

SL's avatar has anatomy problems from head to toe. Poor rigging. and absurdly bad morphing range choices (the fact that a female avatar's arms are too short unless the dial is set to 100, and that is still too short if she is over 5'6" tall... this is just WRONG).

Time is no excuse.

Primitive technology is no excuse.

There simply was no excuse.

The people behind the SL 1.0 avatar deserve to be called out for decades to come. Its a model that should be brought out in 'modeling for game design' classes in a special 1-week long lesson of 'this is how to ruin your platform's dreams of being significant, this is how to ensure your IPO will tank, this is how to ensure people will never respect you, this is how you ensure even 40-year-virgins addicted to anime-pron will have someone to look down on: you'...

The lessons to learn from it are simple: Look at that. Laugh. Now do something very different.

[Pussycat Catnap]

Sigh...

I should be more cautious about hitting submit on my over-the-top analogies.

Ok... the SL 1.0 avatar is bad, and was bad even when it was made.

The mistakes there can be avoided by hiring actual 3D-modeling professionals who have specific game-design experience. People who will know how to rig for real-time animation, how to model for it, and how to model morph points for the needed levels of variety.

The mistakes can be further avoided by making 3 completely different meshes: male, female, child.

- This also lets you put in ways to force the child one into 'G-rated' skins, and prevent the adult ones from being 'morphed' into too-childish of appearances.

You might then need some 'swap points' - specific cut points in the mesh where limbs can be removed and replaced with other 3D models, including the head, and in the adult models the 'pubic region'.
- so that your community can easily customize the avatar later on.

(I should be able to go into appearance, and use a pull down menu or something to simply select what head to show, what kind of left upper limb to show, etc... from among those I have made or bought then linked to that point).

- this kind of customization was seen in MMOs from 2004 on (think of City of Heroes as an ideal example - some of which can be seen in the modern replacement MMO Champions).

From the get-go, every model should have the ability to bake at least 3 kinds of textures onto it: a skin, a shadow map/tattoo layer, and an alpha-mask.
- Even for rezzed objects. There is no reason alpha maps should be limited to the base "nude" avatar, and the "tattoo layer" is more or less akin to a "shadow layer" on a rezzed object so those should be treated as such in a uniform sort of way.

There is no need to ever add support for texture based clothing in today's modern 3D - just start out doing "mesh worn clothing" right from day one... Using a comform to shape system akin to some of what is now seen in Daz3D's auto-fit system... but more locked in because you're only dealing with 3 meshes...

Finally...

A system to swap out the ENTIRE mesh, and its rigging, for a custom one... so that users can later add in animals and whatever... But with some tight limits.

(alternatively, you start with a 4th mesh, that is just balls connected by sticks, in a quadruped form, and then have a rigging for that - thus helping boost consistency for animators.)

[Nathan Adored]

I remember some years back when one of the now-gone opensim worlds was working on a new avatar system. The particular world was mostly kid-avs, and so they were very conscious of how out of wack the existing av system was when you tried to create a realistic kid-shape out of it, so they were designing theirs under the theory it was better to start with a base shape that was kid-like and smaller, rather than adult-like and larger, knowing that it was easier to morph UPward into adult-shape than it was to morph DOWNward into kid-like without distortions appearing.

I HAVE seen another avatar system, for some 3D art rendering system whose name escapes me, where it was literally possibly to dial it all the way down to infant or all the way up to old age and back *without* distortion, and the clothes morphed to match the body along with it.

That said, I really do NOT think it is appropriate to make THREE different base avatar models. Two is probably better. Basically, stay with male and female. At that, Pussycat here was talking about *deliberately* *hamstringing* the child base under some misguided desire to prevent someone from using it certain ways...

Do NOT try to design out things you think might be misused, because that will likely come back to bite us in all sorts of "law of unintended consequences" ways and probably WON'T stop someone from "using them the wrong way" ANYway, but will simply, needlessly complicate the design when simpler is ALWAYS better. If there's three bases, they should be designed to be as similar in functionality and expandability as possible. Do NOT change one of them "just because." IF someone misuses it, it can be handled on a case by case basis by the powers that be. I DO NOT TRUST the "protect us from ourselves," nanny-state software design mindset, where they try to "second guess" every double damned thing we do in case it MIGHT be the WRONG thing! This sort of misguided policy will ALWAYS, ALWAYS backfire and the software will WAY TOO OFTEN guess wrong, screwing things up for those of us just trying to do *harmless* things with that tech, or when trying to do off the wall things the designers did NOT factor in when devising their nanny-state horse-hockey programming. Designers and programmers should check busybody knee-jerk garbage at the door, it has NO double-damned business in tech design! PERIOD!

[Ananda]

Why this talk of three or two standards? Why not just start with one? One of my continual aggravations with the current mesh regime is just how thoroughly gender-dimorphic it is, and how difficult it is to find *any* mesh clothes that can work well for an androgynous form. I realize most people like playing the "more well endowed than thou" game and making hypermasculine or hyperfeminine forms but it's also not very reflective of how variation between the genders is nowhere near as big as the current avatar dichotomy. Starting with one mesh that can go in multiple directions would serve a lot of populations that are not currently well represented, including the androgynous, the furries, and the child avatars.

[Nathan Adored]

Personally, what I want to see in a new, mesh avatar and mesh clothes system is one that is as flexible as the old legacy avatar and related system-clothes system was. We need to have a system again where one can put on long, wintery underpants, and then put on knee-length pants over the top of that, and have the pants conform themselves *around* the underpants, and to be able to *see* the legs of those long-johns coming down *past* the pants legs, to then put on a v-neck shirt *over* a regular t-shirt and have the upper part of the t-shirt be visible through the v-neck part of the outer shirt, and then put on a long jacket over all of that and leave the front of it hanging open. All of this with NO poke-through, because it all conforms itself AROUND the shape of the avatar, and where each piece of clothing conforms ITself AROUND the shape of every piece of clothing underneath it.

[Pussycat Catnap]

"Why this talk of three or two standards? Why not just start with one? One of my continual aggravations with the current mesh regime is just how thoroughly gender-dimorphic it is, and how difficult it is to find *any* mesh clothes that can work well for an androgynous form."
********

Daz3D tried that in Genesis 1. Its a good model. But in Genesis 2 they went back to two different models: a male and a female.

Why?

Because you cannot achieve a wide enough range of possible looks if you make a mesh that uses the same polygonal structure for two radically different bodies.

In a world of meeting the largest need - making it ideal for a body with no sex meets the need of a vastly smaller audience than making it ideal for bodies that have a sex.

[Pussycat Catnap]

I took my thoughts into a larger blog post response:

http://catnapkitty.wordpress.com/2014/07/23/second-life-2-0-my-initial-hopes-for-doing-the-new-base-avatars/

Where I'll be expanding that out to further ideas. There is just so much wrong in the current avatar, but what needs to be done is really all pretty old technology, and pretty accepted artistry (i.e.: actually hiring a 3D artist to begin with...)

[Loki]

One avatar to rule them all?

This is going to be a debate that has pros and cons for and against it. One good avatar would be easy for users to get to grips with , edit, design clothes for, but it 's limiting.

Some creators want the freedom to create truly unique avatars using the latest tech possible , not have to settle for working around an avatar base that gets old and does not change for fear of breaking old crap.

Perhaps in SL2.0 LL don't make any avatars, they simply supply a structure for creators to import and offer their own. Democracy would decide which avatars become the most supported and the doors would remain open for disruptive innovation.

[Ananda]

Hmm, I'll assume that Pussycat is not actively trying to be insulting by referring to relatively androgynous bodies as "no sex". What we are discussing here are videogame situations and the problem is that they currently leave out body types that the majority of the actual human race has in pursuit of these hypertrophied male and female forms. I call myself "androgynous" because I actually have a well-proportioned female shape that happens to conform to a human athletic body.

Or more to the point, establishing a completely different baseline for each would finish the reduction in flexibility we've already experienced with mesh. It used to be you could buy clothes from any department of the stores and be pretty sure you could make them fit. Non-morphing mesh made this quite a bit harder.

What I'm afraid of is rendering it completely impossible to cross gender lines at all.

[A.J.]

If you've ever worked with the current problems of the SL male avatar shape, starting from a point of androgyny is a very depressing thought.

I also agree with Nathan Adored about a need for something similar to the system clothing layer. Mesh clothing is good, but it's wiped out some good options for clothing layering. Some times you want to wear a shirt that's not a liberty bell.

[Arcadia Codesmith]

I'm with Anada on this one. The human body is not nearly as dimorphic in anatomy as our cultural expectations lead us to believe. Apart from the obvious (and superficial) primary and secondary sexual characteristics, telling the sexes apart at a muscular/skeletal level is a matter of relatively subtle differences of degree (and is remarkably difficult with pre-pubescent humans).

It's the cultural baggage of graphic designers that makes it next to impossible in most virtual worlds to create a good female bodybuilder or a bishōnen (slender, beautiful man).

Gender is not a binary. Gender is a spectrum. Creating a model or models and customization interface that work well for all points of the spectrum (not just the most extreme polarities) is a challenge, but it's a challenge well worth undertaking to capture the full range and diversity of the human form.

[Pussycat Catnap]

I'm not at all sure how my comment could have even been perceived as offensive.

We're talking form and shape here, not identity politics.

This goes beyond just genitalia. There is musclulature and proportions to also consider.

I know identity politics want to blur lines - but the forms are different.

And they are especially different when dealing with a half-million tiny triangles you are trying to trick an observer into thinking resemble a human body because of the position you have dropped them into on their screen.

The avatar is not this:
https://tse1.mm.bing.net/th?&id=HN.608015941671914374&w=305&h=300&c=0&pid=1.9&rs=0&p=0

It is this:
http://0.s3.envato.com/files/23464775/preview%20master.jpg

Just look at that last image... notice how radically different the polygons are in placement.
- That is VITAL to being able to achieve proper animation.

Identity politics plays no role in this. Polygon position for animation and morphing potential is what counts.

[Pussycat Catnap]

My above example image is even not ideal. Its differences are not really -ENOUGH- between the male and female. The modeler got it in the chest area, but not in the hips - so walking / sitting will not be ideal, and some more curvy female shapes (the actual norm - a very curvy body) will be 'broken'.

There is also a lack of detail in the limbs of the male - meaning a more muscular male will not be possible.

This is why Daz3D moved away from one model when they did Genesis 2. One model for both sexes is just not flawed when it comes to being able to fully capture the range of body shapes each sex can achieve.

[Arcadia Codesmith]

Morphological differences between human males and females, beyond superficial primary and secondary sexual characteristics, are predominantly bell-curve distributions. There may be significant differences between the averages in things such as body-fat distribution and muscle mass, but the overlap between the ranges is more significant than the differences -- we are more alike than we are different.

A pair of 3D models that does not allow females to achieve the "cut" look of a professional bodybuilder or precludes males from a delicate, "feminine" presentation is a reinforcement of the cultural aesthetic ideal of a specific culture, and not a realistic representation of the full range of human morphological variation.

That's not to say that you can't go with a multi-model character customization system, or that such a system can't include average templates for "male" and "female". There's no reason not to include factors relating to cultural aesthetics when you're marketing to a specific culture.

But it should also be possible for a character between the two polarities to achieve an androgynous presentation and allow the player to choose whatever gender identifier they please. Not only does that address marginalized individuals in your target culture, it broadens your product's appeal in other cultures where gender aesthetics may be markedly different.

[Ananda]

It's pretty hard to escape from identity politics, when it is encoded at the foundational level in nearly every online game. Second Life is one of the few that does currently allow mixing and matching of variously gendered goods. Here's a bit more detailed look at what could happen if we go with entirely different polygonal models.

A different wireframe for male and female generates a different UV map. Different UV maps means making different textures for each. This means skins, tattoos, clothing layers and alphas will all be incompatible if you've got the wrong shape on. Additionally, mesh clothing and attachments will morph based on assumptions about how the underlying body sliders morph. This would mean that once again, trying to put mesh items on the "wrong" body type would just not work out too well. Now, what are the chances that the makers of fashion and skins and such will be willing to devote the extra time to double up all their products? It sure doesn't happen now. The name "Ubisoft" comes to mind.

It is absolutely identity politics if your proposed system not only discourages the boys from shopping in the pink toy aisle and the girls from shopping in the blue but just full-on prevents it.

[Ananda]

As a P.S. nothing in what I'm saying is to suggest that resident-designers shouldn't be free to come up with as many Pro Wrestlers and Venus of Willendorf mesh figures as they please, just that I think the initial project should be as accommodating as possible of how actual bodies can vary without regard to gender.

[Jaqua]

I agree with Pussycat, I'm from the Poser community, and Possette is considered to be a bargain basement avatar, but what I think SL 2.0 should do is buy out Victoria and Michael 4 from Daz3d, you know the avatar (Victoria) that was featured in Time Magazine some years back as the most popular avatar in the world, because both Michael and Victoria are easily the most versatile avies out there, and they're free. Don't let me get started on the accessories and fashions available for these two, also, you can design for them the same way you would for any mesh based avatar, I learned out to UV map and skin from using them.