I wasn’t talking about knocking over BIG things like tables… I meant little things, like glasses or walking through leaves and small stones, small delicately placed things that simple do not budge. Ever. Bumping a table might spill a beer, and that isn’t going to mess with pathfinding.
But, that said, I really do appreciate your in depth response. That was generous and interesting. (I know that sounds a little internet sarcastic, but I’m sincere, I do appreciate it)
Could some of it be the basic math?
Cubic splines are great for smoothly interpolating mechanical movements, but people and other animals are … twitchy
It’s not realistic to smooth things out TOO much
For some reason I am reminded of what Scott McCloud wrote about in Understanding Comics when discussing Japanese style: details make something or someone background, or illustration, but simpler styles convey identity. The less details, the easier it is to put yourself in the pictured protagonist’s shoes.
Maybe games could go with something similar, the Pixar approach of getting the environment perfect, but exaggerating the humans and animals enough to bring them juuuust enough on the brink of the Uncanny Valley without falling in?
This is primarily a function of scale. Branching stories can’t be Q&Aed properly in large games so they are necessarily shrinking as games get bigger. “Bigger” in this sense means the larger teams and production schedules required to develop for ever more complex and powerful hardware. This is why stories have gotten worse as hardware has gotten better, which seems counterintuitive. It’s all about only building what you can QA. It’s also why every Bethesda game is a goddamn hot mess bugs-wise.
Financial risk plays a huge role as well. Everything is a clone of GTA or Call Of Duty now because when you’re risking hundreds of millions of dollars and five years of a team’s time, the product has to be reasonably successful or the company goes under. When those are the stakes, you’re forced to settle for a reliable formula that is likely to do okay.
All this is why, much like movies, the only interesting stuff now happens in the indie space where people can afford to take risks and scale is smaller. The challenge there is curation because places like Steam are overwhelmingly full of junk, with some really innovative gems hiding within.
Ah, in that case the reason is twofold- production scale and performance overhead.
Making all the little bottles and dishes simulated with physics is really expensive, and games have to make hard choices about where to spend those resources. A lot of games start out wanting to do this kind of thing (and I worked on many that did) but when there’s six months left to ship and you have to find 5 fps or the project fails, rigid body simulation of bottles on the tables is the first thing you cut.
Second, each of those bottles and dishes has to be modeled by an artist. If you also have to add rigid body collision volumes, inertia tensor tuning, and ray cast manifolds for every model, production time for props triples. This is the kind of luxury no team can afford. This is before we even get into the QA nightmare when bottles start jumping through walls because of raycast optimizations in Havok or whatever.
Yah, Ubisoft and everyone else are full of shit on that one. If that were really the reason, they could make only a female protagonist. If, as they claim, both genders are equal and production costs are the only reason, then just make women instead of only making men. What’s that? More excuses, Ubisoft? Yah, thought so.
Oh damn it. I got distracted and forgot the biggest point I wanted to make.
My BIGGGG peeve is that all this work is just going into another murder sim fps. The only genre that will support this much investment. We could be doing so much more with the tech and yet… It’s just gophers and mallets all over again.
Also, Veronica, I’m an old game dev with decades of experience. I understand the tech and the industry. I’m enjoying your writing so much I didn’t want to mention it, because there are other people here reading your really great info. I hope you don’t mind.
Apologies if I was ‘splaining there. I just like talking about this stuff and I appreciate your patience. 
I agree, the tech gets ever more amazing, but the content never changes. Seems like such a waste. I left the industry because I got tired of devoting years of my life to yet another game for the same 4 million emotionally stunted young men. I never even liked any of the games I worked on in my whole career.
How close are we to having virtual environments like this being cloud-based assets that can be easily licensed and used by a variety of different types of games? (Racing and flight simulators, family friendly games like Splatoon, VR hangouts, etc)
In Microsoft Flight Simulator they literally recreated the entire world using a variety of data from Bing Maps and other sources (with some areas more “hand built” and accurate than others) and it’s slowly getting better and more accurate over time. Eventually I would think that an investment like that would be very valuable to game developers and you’re going to have far fewer companies attempting to create their own environments rather than licensing something that’s already out there.
That makes sense; once you have a meticulously rendered simulation of a city out there different developers could just adapt and re-skin it for different purposes like how Hollywood re-uses old sets. Say, tweak this sim to an early 1940s version of the city to make a covert ops game set in Nazi-occupied Amsterdam.
I can’t imagine spending any time looking at those goulish robots, let alone paying $$ for the privilege.
I’ll bet that you could have AI do some pretty wacky re-skinning automatically, like giving any given city a “Simpsons”-style look if a game doesn’t call for photorealism.
No apologies needed. I’m really enjoying your breakdowns and I’m sure other people are too. I left the biz to make toys after being badly burned by messed up social dynamics.
AI doesn’t even have to “reskin” in-engine. The style transfer GANs can use the output image and change it on the fly. There’s an amazing vid on YouTube of GTA5 being modified to real-world dash cam video.
“Man, if I had a dollar for every time had a dollar for every time someone at Ubisoft tried to bullshit me on animation tech,” Jonathan Cooper, an animator at Naughty Dog wrote on Twitter. Cooper, who previously worked on Assassin’s Creed III, Mass Effect and more, called out Ubisoft’s comments on Assassin’s Creed: Unity writing: “In my educated opinion, I would estimate this to be a day or two’s work. Not a replacement of 8000 animations.”
Including female assassins, according to Cooper, wouldn’t necessarily need additional motion capturing. “I think what you want to do is just replace a handful of animations,” he said. Using Assassin’s Creed III: Liberation as an example, Cooper says that characters can be created from models of either gender, pointing out that of the games characters “Aveline de Grandpré shares more of Connor Kenway’s animations than Edward Kenway does.”
The issue is not that playable women characters are particularly hard to animate, it’s that too many studios only try to wrench them in as an afterthought. With good writing and planning, there is never an excuse for not having a variety of playable characters unless the creators simply didn’t want to include them.
They probably started work on “female avatars” way too late, so by the time crunchtime came around instead of being an aspect of the game that needed bugfixes and polish, it became a project that would need this many thousands of hours of work, and this many staff.
Actually, the image of Zuckerberg in Facebook’s VR world looks eerily similar to Zuckerberg’s real life plasticized face. So maybe Facebook is closer to being finished than we think.
NPCs are hard, and something that is generally not worth the investment because most of the time they are just environmental window dressing.
For realistic NPCs not only would you have to basically map an entire “life” for every single one (like Shenmue and others have tried to do with varying levels of success), but you’d also have to invest significantly in kinematics, processing, and of course staffing — all of which @VeronicaConnor discussed with much more eloquence and authority than I ever could. All this for something few will ever just stop and look at for more than a few seconds.
I look at this video and am pretty damn impressed. Slightly janky NPCs barely even register to me.
Also let’s not forget: this is a Call of Duty title. Most of it is focused on shooty/stabby/explodey things.
I think it might even be an old game with updated graphics.
Can’t wait to see a raytraced version of “No Russian”
This is really key. Game teams have to put the time into things people are likely to see because there isn’t enough time to do it all. It’s the same reason branching story lines are rare. Putting all that effort and QA into building a branch that only 20% of players may see is not good production resource management.
I’m not in game dev, but I am in software development and I know these struggles all too well.
A wise manager once told me “shipping is a feature”, and I’ve witnessed first hand Brooks’ Law far too many times to count.
That’s not to say you should set out to ship garbage, but you also have to know when “good enough” is good enough.
You can tell it’s not real because the PoV character isn’t getting run over by bicycles.
