Could there be a AAA game industry apocalypse?

Exactly. That is why comparisons don't have any value. They made what appears to be a beautiful game, with a pretty small team. It took 3-4 years, and several of the team were high-level AAA vets. It is unique conceptually driven by a strong art/tech vision built by very knowledgeable and experienced folks. But all games are small teams compared to AAA, those by definition are going to be largest staffed. But they are also a small segment (AAA) of the games market by title. No Man's Sky's team is 1/3 the size of my team. However, out game is being developed in just over a year. Size comparison are very difficult to place value on, they are deeply affected by skill levels and timeline. Budgets are also a factor, No Man's Sky's wasn't built on a typical indie budget either. A solo indie can't look at NMS any more than CoC for a comparison as to what they can do is recognize that polish, strong art/visuals and strong vision can pay off.

 

Being an indie is like being a small business, you ride the wave of new technology and techniques (and risk) and hope that at the top of the wave some publisher plucks you out of the water. All this jazz about AAA or not doesn't really factor into my planning and preparation. All I know is that I have to do something that hasn't quite been done before if I'm to succeed at doing something that I can't really say I've seen done yet.

That means tools, tools, tools - and workflow. I know some people have the time, patience, connections, and money to do something at 50% efficiency but I don't. As a solo indie if I can't create 5 or 6 good quality models a day for a couple of years straight, and have time left over for programming and level design, I won't even bother to try making it all myself. Which means that the traditional highpoly - retopology - unwrapping - texturing is going to have to be highly automated.

 

There are tools -- not being an artist (but wanting to be familiar with the tools), I don't know how good they are but while dabbling in Blender and realizing just how awful that process is, I found two you may already know: RetopoFlow, Topogun. I've also been meaning to check this out:

http://www.cgchannel.com/2015/11/download-free-auto-retopology-tool-instant-meshes/

 

I checked out RetopoFlow a while back but it doesn't automate it anywhere near enough, it looks like it would cut down the topology time a fair bit but is still basically a manual operation.

Now Instant Meshes looks very interesting. Its methodology is basically what I worked out when brainstorming ideas for a possible tool - coverting the mesh to uniform quads and snapping edges to sharp features. I would add a feature to then add one or two bevels to sharp edges (since I'm going for a no-normal-bake approach) and automatically add seams and unwrap - that would be my ideal workflow.

Also it would be good to know if it can create a single continuous lowpoly mesh from many embedded highpoly meshes by disregarding geometry that's inside the hull of another object, that's something I would definitely like to have since my approach to highpoly involves many embedded objects that are stuck into one another.

 

@Billy4184
Automatic Mesh retopo is an active field of research, I have seen some paper or two that you might be interested in if you don't know already. I'll have to hunt them down, but just a quick googling give me that, so tell me if this is the kind of stuff you are after.

http://polycount.com/discussion/152430/sketch-retopo

The reason I keep spamming NMS is because they did nothing somebody didn't made before, and in fact you can find more ambitious and more complex project made by team of non industry programmer.

Planet explorer and planet nomad are two example that come into mind (and they might not be the best example I would have to redo a survey of the genre lol, that's out of the scope of this post), there is many other. So that remove the experience factor, in fact they are building on other's innovations. In fact there was a river of salt from these people because NMS took the spotlight for stuff they did years ago.

So what made NMS special? I think it's first TIMING, that VGA reveal happened really at a great time. The second is the right visual, most stuff in this genre took inspiration in the militaristic and realist style of space exploration, going for the modernized version of the underused 70s SCIFI style, and aiming at the same adventure feel of these pulp, was a great way to differentiate and introduce a breath of fresh air when visuals tended to gather around only a set number of small style (realism, cartoony, atmospheric, TF2).

But really, where I want to steer the discussion is toward defining the precise point where we have problem. To innovate is to understand and define a problem first. Lumping everything under generic term, for example under scope, do nothing for the moving the needle. For example scope can be breadth or depth, gta 5 and unchartred 4 have very different type of scope, Uncharted has a smaller scope but has high density, Gta5 has a lower density of details but a broader scope.

The question is, is there a place where we can innovate and push further, we don't have to be good at everything, and that's the appeal of these discussion to me, not to actually replicate AAA, but to identify an aspect we can actually push that they don't out of habit and convenience (don't fix what's not broken with lot of resources).

 

If I knew, I definitely wouldn't be sharing it.

I share so much stuff, it's not appreciated, so it's understandable that if there was a holy grail, why would anyone want to dilute the contents?

 

Oh I don't know; the countless "we must have xxx feature or we can't release our game" posts and their soap box ilk? I think there's a lot of people manage to convince themselves that water isn't wet under the full moon. Just look at the U.S. primaries.

 

im surprised theres not a apocalypse with all the garbage games with in-game $ garbage
idk i avoid all those games like the plague
.. its really disgusting and a shame soo many games are like that nowadays

lotta suckers out there i guess

 

Actually... I apologize. I completely misinterpreted what you were saying. My bad. I completely agree with what you were saying.

 

Hehe it's alright I do that a lot - I even edited my post to trying to say something lighthearted but it's hard at 3am

 

For simple objects, yes. People are already having good results creating stones and rocks with it. Though don't underestimate the amount of postprocess work needed to create a quality end result, as scanning is bound to come with some artefacts. Maybe there are tools that handle that automagically, I don't know...

Then there is the thing that simple objects are actually pretty simple to sculpt if you have some skills and good tools. Creating a good looking stone in ZBrush takes maybe an hour, maybe half that. Scanning that stone might take slightly less time... but then you still need to retopo and texturize it. That is also not so timeconsuming, all in all the amount of time saved with 3D scanning for a SKILLED ARTIST is negligible next to having less artistic freedom (unless you start changing the scanned stone, and then you could work from scratch anyway).


For less simple objects scanning might make more sense... but you now get quickly into needing complicated scanners, very big ones (ever tried fitting a tank or big rock into those scan-at-home devices?), and no tool in the world being able to accuratly postprocess noise and artefacts out of the end result.

The amount of input (# of images) needed for good results in photogrammetry is actually quite staggering, point clouds are cool but then still need special hardware to scan.


Just as an example, for World of Tanks Wargaming is trying to create "HD Models" for all their ingame vehicles (which means around 100k tris per model and high resolution textures)... thus they need to redo 450+ vehicles to an extremly high standart. We are talking about having every damn visible part of the airfilters as a separate mesh.

Now, they tried everything in the book to cut cost and speed up the work, including photogrammetry and portable scanners to get the base for their outsourced artists to work on (remark: yes, just the base, because all thos details do not separate themselves neatly into separate meshes).

They did that for their new IS-3 model (Soviet heavy tank, produced in large numbers, so plenty of examples around). First they seemed to have a hard time getting access to such a tank, AND convincing the owners (guess a museum) that they wouldn't break anything (as said, its not such a rare tank).
Then they spent days taking more than 3000 pictures from the vehicle. IDK how much time was spent on postprocessing.

The result was impressive, but clearly nowhere ready to be put into a game engine. Everything was one big monolithic model. Not ideal for anything besides 3D printing really

And then there was the noise... weee, even on the model they felt clean enough to show to the masses there was still some noise and artefacts visible in hard to clean up places.

Lastly, it happens that this example of the tank wasn't necessarily exactly what they wanted. See, in a tank as often built as the IS-3, there are many differences between production runs. Chances are good that the specimens in good condition today are later production runs, OR have been upgraded during their lifetime.
So with their plan to show the IS-3 in their initial config from just after WW2, that scanned example was hardly ideal.

They ended up redoing the model. IDK really, but going by their current standards I guess they scrapped the photogrammetry model and went with a hand modelled one. This would have allowed them to work with original factory drawings, instead of trying to rework a scanned example into the standard of the factory drawing (which kind of defeats the purpose of doing a scan in the first place).


AFAIK most HD models so far have been hand modelled by skilled artists with very exact plans. Because of the amount of work of even getting access to an example to scan, all the variations between examples, the amount of work to scan, and finally the amount of vehicles that have never been built or where no example has survived, going with handmodelling was most probably just more economic than scanning (keep in mind the studio is working from belorussia, and thus most probably employs most of their artists in the CIS region, thus wages are most probably way lower than in the US or worse, europe).


TL;DR:
3D Scanning is the shizzle, if you use it to its strengths. It will never replace a good artist for the complex stuff, and a good artist might get better results in about the same time for the less complex stuff...
But you CAN substitute a good with a less decent artist for the less complex stuff by using 3D Scanning to make up for missing sculpting skills.

 

@gian-reto-alig for the time being, it's going to be very hard to hand off the creative side of art to computers, so sculpting and creating highpoly meshes will probably be around for a while.

But retopo, unwrapping, baking and to a large extent texturing should pretty much be an automated process by now. I actually find it much, much faster to make a highpoly than I do all these other things - all in all probably 75% of my time is spent after the highpoly ... and that's without going anywhere near hand-painting which I consider to be the main time-drainer.

 

Well, I find there are many tools that can help bring down the time needed, but in general, I agree with you.


Having had good results lately for retopoing my rock/stone sculpts done in ZBrush with the Blender Decimator Modifier. After having mucked around with 3D Coat Autopo for some time and never getting good results without a lot of manual cleanup, I found that for simple meshes were topology wasn't so important, and tris were actually desirable like a rock, the decimator mod does the work in no time, and the end result needs almost no cleanup afterwards (just some weird thin tris to get rid of).
Just export the tool from ZBrush, import to Blender (takes a while, because millions of polys), add the decimator mod, find out the right ratio to get to your desired tris count, apply the modifier (takes some time again), export or UV directly in Blender.


Of course that doesn't work for your animated character that needs carefully chosen topology, or your hard surface sculpt that needs non-random topology.

 

We are already seeing a push to higher and higher budgets to create AAA level games, with fewer and fewer studios working at this level.

I would be amazed if AAA studios themselves are not pushing to make tools that reduce their costs and their need for staff.

This could be seen in tools or middleware adoption all the way to using an off the shelf engine.

What if No Mans Sky released their tech as a World / Universe Building toolkit for other studios or on the asset store?

 

Then you'll see tons of vaguely NMS-like knock-offs and generally nothing else will change. Have you watched any of the presentations about the engine? It's all highly custom stuff and very specific to NMS, it isn't some sort of generic universe-making tool, and it's running inside an entirely custom engine built in-house start-to-finish.

 

Uhmm. Yes. They have pretty much always done that. Where do you think engines come from? Tools are a big part of game development in any medium or large studio.

Though it doesn't really result in less staff, or practically speaking, lower costs. As those costs and resources are usually applied to other areas.

 

This might be the key difference between AAA and indie that people are missing in this thread.

The AAA makes big, expensive tools so that they can do bigger, more expensive games. The indie can use those same tools to make small games with less resources.

 

...like, oh, I dunno... Unity itself?

 

Unity is kind of the oddity here... most other engines have been built by game devs for a project, have been reused and grew from there to become an engine offered to other devs for money.


Sooo... zombiegorilla still has a point.

 

Storks. Very technically skilled storks.

 

Unity did start by making games. They just weren't good at it. (Thier words)

 

Oh, okay.... didn't knew. Will have to read into that part of their story one day.

 

I haven't read the other comments, but...

Just because you can buy high quality assets doesn't mean your game will be good, or "AAA" quality. You can't buy unique, moving stories with unique characters. Assets rarely come in one matching, consistent mood or visual style. You can't buy innovative gameplay, only clones that other people also have access to. You can't buy good marketing or networking skills. You can't buy good, thoughtful design.

What you'll have is a tonne of mish mash games made by inexperienced people, which might look nice visually but will most likely be pretty crap as a whole.

 

https://en.wikipedia.org/wiki/GooBall

 

This AAA vs “good” argument assumes visual fidelity is going to equal “good”. But that’s not so. Even if, in the deep learning future, AAA absolutely nailed simulating all the senses (holy grail), and humans could just stand back and admire the AAA VR wonder, after flicking a switch – even that wouldn’t guarantee the experience would necessarily be “good”. That doesn’t make sense to me. Fidelity does not equal good. Art = good.

 

AAA games, industry will not die. People always need drugs. And also, indie is not yet apocalypse. Still there are someone indies achieving success.