Development Info Wasteland 2 Crowdsourcing Experiment

[shihonage]

So, the art... submitted by hundreds of people of wildly fluctuating proficiency levels... will be magically aligned with W2 art style and polygon count requirements...

Now, you may say, "it's just filler assets like rocks and radar dishes and shit"... but I would expect that in terms of man-hours/money (the concepts sort of bleed into each other, considering someone at inXile would have to be reviewing and correcting even the "best" models), IMO it would be more efficient for inXile to just buy filler assets from sites that already sell them.

The less eccentricities Fargo produces, the more confidence there is in what he's doing.

 

[Infinitron]

Hundreds of people? That's optimistic. I'm pretty sure only people with reasonable expertise will even bother to try participating in this. Also, they don't have to buy any of them.

 

[Brother None]

Now, you may say, "it's just filler assets like rocks and radar dishes and shit"... but I would expect that in terms of man-hours/money (the concepts sort of bleed into each other, considering someone at inXile would have to be reviewing and correcting even the "best" models), IMO it would be more efficient for inXile to just buy filler assets from sites that already sell them.

This *is* a site that already sells it, the Unity Asset store. Unity does the first reviewing and packaging of the assets, and inXile then selects any and all assets they feel they can use. They provided some basic requirements and a scene for you to test your assets in. If it doesn't fit into that scene, it doesn't fit into Wasteland 2. If it'd take too much correction work I doubt they'll pick it up, it's a basic cost-benefit picture.

But you're saying buying filler assets from the Unity asset store without making any specific crowd requests explaining what you're looking for is better? I don't get it. How?

Hundreds of people? That's optimistic. I'm pretty sure only people with reasonable expertise will even bother to try participating in this. Also, they don't have to buy any of them.

Doesn't matter if it's hundreds. Like I said, Unity is working with inXile on this and does the first basic reviews for quality and suitability to make sure it meets standard requirements before they're even submitted to inXile.
Sadly, this means inXile will never see Prosper's submissions

 

[skuphundaku]

I think object oriented languages like Smalltalk and C++ originally had just this sort of thing in mind though. Inspired by books like the Mythical Man Month they envisioned a future where you wouldn't be constantly reinventing the wheel every time you coded and could just search a database of shared objects for many tasks. Such a database could even rate the objects based on execution time / efficiency and how well documented and well organized the code was. I suppose there are some C++ object libraries available, but AFAIK not for every task imaginable and not in the sense of something that nearly every programmer could use and share.

The object oriented paradigm is still too fiddly to even come close to how physical components work. The state of the art in software modularity are component-based systems. Components are self-contained software constructs that expose a standard interface and, as long as other components respect that interface, they can interact. Such a system, that you may be familiar with and not even know that it is component based, is Android (what you perceive as applications are, in fact, constructs made out of four types of components held together by metainformation).

 

[shihonage]

But you're saying buying filler assets from the Unity asset store without making any specific crowd requests explaining what you're looking for is better? I don't get it. How?

If Unity is really shielding inXile from the waste of man-hours on filtering out submissions insuitable for W2, then indeed the current approach may be beneficial.

 

[jewboy]

The object oriented paradigm is still too fiddly to even come close to how physical components work. The state of the art in software modularity are component-based systems. Components are self-contained software constructs that expose a standard interface and, as long as other components respect that interface, they can interact. Such a system, that you may be familiar with and not even know that it is component based, is Android (what you perceive as applications are, in fact, constructs made out of four types of components held together by metainformation).

Can you give another example of what you are calling component based systems? I don't know much about android except that it is Java based. What about Microsoft COM Objects such as are used in DirectX? I've always felt that objects, while useful, are not as independent and modular as I would have liked. Not enough like true black boxes with just an input and an output or communication method (truly sending messages to each other). OTOH, I haven't heard of any specific system that does a better job. I like the idea of programs and sub-programs communicating with each other directly.

Another possibility I like with keeping program components truly independent /standalone is the idea of self modifying code. It would be nice of an independent program unit could do some learning and decide that a subprogram could be more efficient and rewrite and recompile that code. Of course that isn't true recursive self-modification, but it seems more manageable.

 

[Apexeon]

This might interest the 3D asset-making machine, Apexeon.

I was going to make a ray traced version of there demo movie a while back for fun.
I wanted to do the demo in the IE style (ray trace style) to see if the game gets a more gitty vibe.
I will have a crack at the WL2 list and post them in my mega spam thread.



Or I could just spam a few pictures from my failed indie 2007 pitch.
It was a turn based design .
I am a big Masters of Orion 1 fan.
Publishers and coders are not .

 

[skuphundaku]

The object oriented paradigm is still too fiddly to even come close to how physical components work. The state of the art in software modularity are component-based systems. Components are self-contained software constructs that expose a standard interface and, as long as other components respect that interface, they can interact. Such a system, that you may be familiar with and not even know that it is component based, is Android (what you perceive as applications are, in fact, constructs made out of four types of components held together by metainformation).

Can you give another example of what you are calling component based systems? I don't know much about android except that it is Java based. What about Microsoft COM Objects such as are used in DirectX? I've always felt that objects, while useful, are not as independent and modular as I would have liked. Not enough like true black boxes with just an input and an output or communication method (truly sending messages to each other). OTOH, I haven't heard of any specific system that does a better job. I like the idea of programs and sub-programs communicating with each other directly.

Another possibility I like with keeping program components truly independent /standalone is the idea of self modifying code. It would be nice of an independent program unit could do some learning and decide that a subprogram could be more efficient and rewrite and recompile that code. Of course that isn't true recursive self-modification, but it seems more manageable.

Yes, COM is another example of a component-based system. It's older and, IMHO, messier because of that, than Android, but the principle is the same. The OpenNI API that was developed for Kinect and other depth cameras is yet another. Now, .txters ready your .txting tools, the implementation of the system I developed for my PhD had a component based architecture exactly because I wanted to easily be able to change the the way data flows through the system, to allow for plug-and-play interchangeability of components that have the same interface, to allow for components that, conceptually, can run in parallel, to easily run in parallel without the programmer having to micromanage everything. One of the further research goals (stretch goals) I listed at the end of my thesis is to add an AI overlay that would be able to dynamically create whole component setups for a given goal based on some user input and the metainformation provided by the available components. And I wasn't even considering recompiling/rebuilding stuff here, just working with the components that you already have. Going into recompiling/rebuilding components on the fly opens up a whole other can of worms. After all, the strength of a well designed component based system is that you can keep recompiling/rebuilding stuff to a minimum, if at all.

 

[DwarvenFood]

What's interesting about this sort of thing is it levels the playing field for the whole world. I've lived in places where people would work hard all day long doing things like heavy construction or demolition work for $2/day. I've paid guys less than $10 to work all day long doing construction work on my house and they considered that good pay. I don't know what they are paying for this sort of modeling, but you could earn a real living this way in some places. And, since English is the most popular second language for nearly the whole world the field is even more open than it would be otherwise. This is really cool.

Have you not witnessed how this process, and how it's been used, has contributed to ruining the economies and cultures of pretty much every country in the world?

Going a bit off topic, but what do you mean here, because in this case the pay would not differ per region/country and I think what you are talking about is outsourcing of work to cheaper countries so costs can be lowered ?

 

[General Maxson]

How come nobody mentioned the train screenshot? It looks great to me.

 

[TwinkieGorilla]

How come nobody mentioned the train screenshot? It looks great to me.

The fuck you talkin' about?

 

[General Maxson]

How come nobody mentioned the train screenshot? It looks great to me.

The fuck you talkin' about?

Train screenshot from here: http://www.nma-fallout.com/, first news topic, last pic. Only one in color. I don't know how to link it here unfortunately.

 

[Infinitron]

 

[nihil]

I really hope they provide solid guidelines for art direction and such. Inconsistent graphics are worse than mediocre graphics.

 

[Oesophagus]

Prosper should submit the codexian in-game statue.

On that note, someone with skills should actually submit a codex troll

 

[grotsnik]

That train shot's pretty neat. Looks a good bit more in the vein of that STASIS dev's atmospheric, vegetation-clinging-to-decaying-ruins post-apocalyptic screenshots and a bit less Looney Tunes desert.

 

[Infinitron]

Yo, Brother None, any chance of a higher-res version of that image? It's all blurry.

 

[General Maxson]

I don't know why but it reminds me of Troika's PA tech demo (and that's a good thing).

 

[tuluse]

I don't know why but it reminds me of Troika's PA tech demo (and that's a good thing).

It has the same Sun-baked glowiness to it.

 

[TwinkieGorilla]

Troika's PA tech demo

Goddammit.

Every time somebody reminds me of that thing a small part of me dies.

 

[Infinitron]

RPS commenters are very upset about this: http://www.rockpapershotgun.com/2012/12/06/build-your-own-wasteland-2/

 

[ColCol]

RPS commenters are very upset about this: http://www.rockpapershotgun.com/2012/12/06/build-your-own-wasteland-2/

And not a fuck was given that day.

 

[Jaesun]

 

[Micmu]

Why do I always misread the word "crowdsourcing" as "crowdfucking" ?

 

[Brother None]

Yo, Brother None, any chance of a higher-res version of that image? It's all blurry.

Maybe. I suggested the same when I first saw how blurry the shot was. No idea if they will. But it's about time for more screenshots.