*spin-off* Next Gen Gameplay and/or Graphics Differentiators

[Prophecy2k]

The concept is a library which is available to all, and for all to contribute to and/or extend/improve for their own uses. But MS/Sony should be stepping up and saying "this is the benchmark, take it from here".

This is an interesting concept which i could see having some merit. It wouldn't necessarily be useful to everyone, as some would prefer to use their own in-house developed technologies so as to maintain a competitive advantage.

In many ways i see stuff like Sony & MS' efforts with their in-house tech teams this gen, in particular Sony with ICE-team and Phyre Engine as a good starting point. Allowing devs to submit updates for approval in a similar way to other open source software would help things along too. If they expanded their libraries for their "free" first-party developed technologies, allowing devs to freely access and carve out chunks of useful code, then I think that would be a sterling commitment on the part of the first parties and imho them really going over and above the COD (pun intended).

If they did this, along with the subsidies or marketing support ideas that i suggested, then I think they would have much more success in assisting devs to get their projects up to the quality bar that they individually wish to. I mean lets face it, it's not as if devs don't want to have high quality AI and animation systems in their games. If Sony and MS, along with 3rd party middleware providers can grant devs a leg-up from the beginning of the generation then i do believe that it would be beneficial to both developers and gamers in terms of the quality of games produced. I also think it would matter much less though as you got further into next-gen, for the reasons ERP very succintly explained.

I do feel the "passive" approach is the much better one. Enforcement which tries to dictate game design in an indirect way will never work. I agree with ERP on that most definitely.

 

[TheChefO]

A big portion of the issue here is buying into the middleware hype.
This stuff doesn't just work, and it many case it's a very far cry from what's advertised.

Even Havok which has extraordinary adoption, if you go to a meeting with dev lead all you here is them bitch about there tools and Havok.

I have seen products on which Havok was used and almost certainly did far more harm than good, a game specific solution for character control can IMO be at least 20x faster than Havok's solution (and I might be understating that).

Havok (and I'll extend this to most middle ware) makes it easy to get things working and takes IME an enormous amount of work to ship something that actually works with acceptable quality in your game.

I'm not knocking Havok it's a well put together piece of tech, otherwise people wouldn't use it, and there is a value to having things working earlier in development, but don't assume that adopting complex technology is easier than building it.

As to Euphoria, there is a small class of animation problems it does very well, and a very large class where there are better solutions. Unless that small class of animations is of great value to the game, or is disproportionately represented in the game, the adoption cost would have to be relatively low. Everything I've heard says it isn't.

As to using TRC/TCR's to dictate technology, they already place an enormous burden on developers, go ask a PS3 team how many man hours go into implementing and testing save game. And I'm not talking about serializing game state, I'm just talking about writing that state to the media, and the associated UI.
Plus all your really asking todo is dictate game design choices and I have a problem with any restriction on that.

Two things:

1) the general direction of gaming is toward greater interactivity and realism. With that, custom solutions which used to apply for one instance, but not another are becoming fewer and fewer.

Game design used to be that you had a character that could do x, y, and z ... and enemies could do a, b, c (with boss battles bringing in new abilities/behaviors) that was the game.

As time goes on, the list of abilities grows for the character and the enemies. The list of behavior grows for the enemies. The interactivity grows.

The end result is a game which interacts with everything and can do things a person can do. There are fringe cases still, but there will be fewer and fewer niche cases and the niche cases are generally based on human+ abilities which if one studies every game on the market, there are a lot of similarities.

Human or human-like forms and functions will present the majority of game design scenarios and so it makes sense to focus on this framework.

The majority of game worlds will also be reality based (past, present or future) so it also makes sense to focus on this framework.

This makes sense as the organisms buying and playing videogames are humans living in reality.

The entertainment comes from what one does in the world, the story around them, and the characters along the way.

*Note* this isn't to say there will bo no games which do not follow this framework, but the majority going forward will be human or human+ characters in reality based worlds.

With that said, it makes sense to focus on methods which can efficiently replicate the human form, behavior, and environment with designer variables to adjust per the current use case.

So the idea of a method which doesn't account for x,y, or z and therefore can accomplish the same thing that another method does but the first can do it 20x faster starts to become irrelevant because that speed comes at the loss of physics, interactivity, or ability which should be there (majority case).

Granted, this will require more power, and ideally, the most efficient middleware being used. But the notion of being a developer group off doing their own thing is a dwindling notion. Pushing the envelope in certain realms will lead to some joining dev groups which are currently involved in their area of interest, and others joining middleware teams optimizing the code.

______________

2) As I said, Euphoria is simply a good representation of the concept. Specifically using it or not is irrelevant. Having a set of libraries and tools which enable developers to use which uphold a standard at or better than Euphoria should be something Sony/MS are working to provide.

And yes, setting this as a standard for new games on nextgen consoles. Simple question for game approval: Do you have Bipeds in your game? Use the library.

Tweak or build on the library, fine, but the straight use of mocap animation is dated and should be long forgotten. It had a nice run since the ps1 in 1995, but it's time to progress and evolve.

 

[ERP]

1) the general direction of gaming is toward greater interactivity and realism.

See I disagree with the premise.
There is a certain class of currently popular games aiming at realism, but a lot of games aren't going for that at all.

With that said, it makes sense to focus on methods which can efficiently replicate the human form, behavior, and environment with designer variables to adjust per the current use case.

As a research project sure, but today procedural mechanisms for base human animations ARE NOT OF ACCEPTABLE QUALITY. For specifics types of interaction there things can be of use, but you are over-valuing a single small part of a game.
You should talk to Chris Hecker about this, he's probably spent more time working on procedural animation than anyone else and he'll be the first one to tell you that it's not good enough for human like forms where there is an expectation of how they move.

Honestly if it's a problem that can be solved with a reasonable degree of quality, and there is demand for it, you'll see middleware and adoption of it.
Historically Games evolve as new games set expectations of future games, you can't do less than your competitor if your going to compete in the same genre.
It'll become the norm when someone actually manages to do it well and demonstrate the value.

FWIW your also underestimating how much current games can and do do with animation, no one just plays canned animations. I would say they are often some of the most complex pieces of technology in a game.

As I said earlier with next generation, I do think you'll see more interaction in games, more populous "living worlds", I do think that procedural animation has it's place, but I think it's one part and a small part of a much bigger picture.

Hell I'd like to see games without broken pathing, and brain dead stupid AI before we start worrying about procedural animation.

Going back to an earlier point, not walking through NPC's, playing Skyrim, I often wished Lydia had no collision. There is a fine line between pushing the realism mantra at the cost of gameplay, they don't always align.

 

[function]

For a landscape, vertical resolution is pretty important. On the other hand, horizontal resolution is more important for things like characters, which have limited pixel/screen width to resolve anything. Though it's quite possible that a single dimension scaling just looks better during the software upscale since they'd be using as cheap a filter as possible.

If they're using RSX to scale, then the HUD would be scaled too (don't recall if that's the case). On 360, dynamic res is necessarily done in software because of a quirk in how the scaler hardware works - it blanks the screen when changing input res.

Looking at Eurogamer it looks like the HUD (PS3) isn't scaled, so there goes the idea of the HW scaler being the issue. And as it can't be the HW scaler on the 360 either that means its probably a cheap scale there too.

Certainly. It's the reason why that MSAA remapping trick was used in several games on PS3.

Oh yeah. Kind of missed the obvious there ...

hm... keep in mind that devs perform shading and other work on the aliased render target that has all the samples mapped to a pixel (post-resolve). So I'd guess you'd have a pretty plain looking game.

But otherwise, when you're using the MSAA samples to fill in for the sub-HD frame, you'll get a weird pattern on edges since your information is from one of the two sample positions, not the final resolved sample. For non-edge pixels, the two samples would be identical for a particular pixel, so you might get something weird or something better depending on what resolution you're dealing with and what you do with the larger buffer. It'd be different between say... starting with 640x720 and upscaling vs 1120x720 and upscaling/downscaling.

I was thinking of a case where you had, say 1280 x 720 with 4X MSAA so you could always guarantee a downscale from the unresolved buffer. I guess this leads me back to an earlier thought about what a unresolved 4X MSAA, 960 x 540 image mapped 1:1 on a 1080p tv would look like. It's not that much lower than CoD ...

BTW, I have to say that Xbox 360 games downscaled on a SD widescreen CRT, run through RGB scart, can look amazingly not-actually-that-bad. And there's no LCD blur - you forget how annoying that can be actually. On the CRT you can still see well when things are moving fast, where as you lose a lot of information when a cheap (and sometimes not so cheap) LCD starts blurring. CRTs are the gift that keeps on giving.

Well, it'll work fine on plain textures, but once you start using non-linear representations (normal maps, parallax, displacement etc), you'll need something fancier. I get the feeling it might be something that is dealt with early on in the generation when devs have no clue what to do with the excess power, but as time goes on, there are other rendering features that take precedence.

Just use super detailed meshes and lots of LOD levels. And give the machine 16GB of ram to store them all. And use a power brick with a 12cm fan in it. It's the one thing I would pay any price to see gone - I'd burn down my own house to kill visible mip-map transitions.

Try playing Daytona USA on the 360, and looking at the track infront of you (hard not to tbh). Always there, you can never overtake them. Turn around and try and run off the other way and they're still there. It's an omnipresent tide of rolling IQ filth; a part of the 1990s that the Voodoo 1 killed but that respawned.

I'll stop now.

On a related note, what's holding back AF on console is basically texture bandwidth and texture caches.

I do wonder about manual/shader filtering anyway (ala Rage for example). TBH, even on PC I find >4xAF a bit hard to discern in the general case (playing @ 1200p also mitigates the need to an extent). Performance on PC is, of course, in abundance, but I think it'd be a pretty fair compromise in a more limited hardware situation. At least in a lot of games that I play, it's not often I'm having to look down an empty corridor. Devs can always select higher filtering on particular textures for a given scene if they have that much time to do so as well.

Funnily enough, on my 1600 x 1200 I've found a similar thing with AF, and it's not that the monitor's pixels are too small to make them out - supersampling (even just 2X) is godlike.

 

[TheChefO]

See I disagree with the premise.
There is a certain class of currently popular games aiming at realism, but a lot of games aren't going for that at all.

COD
BF3
GTA4
Red Dead
LA Noir
All sports games
Uncharted series
Assassins Creed series


All of these are looking to replicate a believable, reality-based world.

Even in cases where it is fantasy/scifi based:
Oblivion
Skyrim
Gears series
Halo series
Resistance
Killzone series
Fallout series
Bioshock series
Batman AA/AC

...they are looking to create believable worlds, which share reality based characteristics with physics and AI behaviors.

I think that covers the majority of the "hit" games of this generation.

As a research project sure, but today procedural mechanisms for base human animations ARE NOT OF ACCEPTABLE QUALITY. For specifics types of interaction there things can be of use, but you are over-valuing a single small part of a game.
You should talk to Chris Hecker about this, he's probably spent more time working on procedural animation than anyone else and he'll be the first one to tell you that it's not good enough for human like forms where there is an expectation of how they move.

From what I understand, Euphoria doesn't use a 100% procedural solution either.
My point is the advancements made in interactive characters like Euphoria has done, should be adopted as a standard. From that point, build, tweak, modify, whatever. But characters should react in kind to interaction. Characters should place their feet on the ground, even when it isn't flat. Characters should not spin in place without adjusting their feet. Running in place against a wall should also not see the light of day nextgen.

Honestly if it's a problem that can be solved with a reasonable degree of quality, and there is demand for it, you'll see middleware and adoption of it.
Historically Games evolve as new games set expectations of future games, you can't do less than your competitor if your going to compete in the same genre.
It'll become the norm when someone actually manages to do it well and demonstrate the value.

I think Euphoria is by far the furthest along in bringing believable characters to interactive environments, but as we see with their licensing arrangement, it isn't cutting it. We need Sony/MS to step up in this regard.

FWIW your also underestimating how much current games can and do do with animation, no one just plays canned animations. I would say they are often some of the most complex pieces of technology in a game.

Some more than others, but again, I'm talking about a standard level of quality to expect. A baseline if you will.

As I said earlier with next generation, I do think you'll see more interaction in games, more populous "living worlds", I do think that procedural animation has it's place, but I think it's one part and a small part of a much bigger picture.

Interaction, animation, and physics are all part of a bigger picture of nextgen standards that I hope to see.

Hell I'd like to see games without broken pathing, and brain dead stupid AI before we start worrying about procedural animation.

AI is part of the animation tree as well. Canned animation can never account for every scenario and so it will always limit the AI:
*identify weapon
*grab weapon
- weapon on floor
- weapon on table
- weapon on stairs
- weapon on wall
- weapon on couch
- weapon on rubble

(weapon_type: Pistol, Rifle, Frying pan, Bottle, etc)

Attack enemy[a]
- aim for head
- aim for leg
- aim for hand/weapon

etc.


Procedural animation accounts for variables of the situation which enable better AI to be visually represented instead of the stick figure one-size-fits-all event animation that most games use atm.

Going back to an earlier point, not walking through NPC's, playing Skyrim, I often wished Lydia had no collision. There is a fine line between pushing the realism mantra at the cost of gameplay, they don't always align.

That's an issue of AI, and an issue of not having characters react to interaction.

 

[Silent_Buddha]

Not a lot of time for replies today so just short comments of the most important things.

There are rare games these days which are not based in reality where physics and realistic behavior don't matter, but they are few and far between.

There are a ton of games that don't aim for realistic behavior. They far outnumber the games that DO aim for realism.

I disagree.

Both are proven methods which have been utilized in shipping games since 2007.

That quote is with regards to monetary and hardware budgets. The monetary budget you have a little wriggle room, but even that has a max. Unless you really think we should have games costing a few hundred dollars per copy?

And hardware budget limitations are very real and very inflexible. That's why so much time is spent on finding approximations that can be done cheaply (with regards to computational resources). Again I'll use Crysis 2 here as an example. The Dx11 Ultra PC version still doesn't employ all features the developer wishes they could do. And that can bring even the most power PC available today to it's knees.

Unfortunately it will be up to Sony/MS to mandate such a standard, and they likely will not have the balls to enforce such a standard.

Wait, we should now start to dictate to developers how they should be making their games without even knowing what their game even needs? What if someone wants to make a retro-styled game with Humans and is purposely NOT using physics/real human motions?

Just select the Euphoria option in Visual Studio and bingo, job done! The switch is in the "Project -> Art and Professionalism" menu, practically next door to the "Build -> Next Gen Port Auto Upgrade" option.

Except Euphoria doesn't work that way. When a company decides to use them. They have to submit a detailed design document outlining their needs and what they are trying to accomplish.

Euphoria's company then attempts to implement what they "think" is the best way to achieve this. They then send that back to the developer making the game. The developer takes that makes another detailed report outlining what they got right and what they got wrong for the project at hand the sends that back.

The process is repeated however many times needed until the end product is close enough to what the developer needs, or the developer runs out of money for developement and they just go with what they currently have.

If you can afford the time and resources for not only Euphoria but the whole process of getting back the animations quality you want then it can be a decent solution.

Most companies with limited budget will be far FAR better off doing their own system as they are intimately aware of what their project requires. I believe ERP already touched on this.

Middleware can be good but it's not and never will be the "best" solution for a product. Blizzard recently dropped support of Havok, for instance, as they could better serve their projects by developing an in house physics package more suited to their game types.

Regards,
SB

 

[TheChefO]

There are a ton of games that don't aim for realistic behavior. They far outnumber the games that DO aim for realism.

I provided a list of some of the most popular games this gen which in one form or another, are aiming for a believable world.
As I said, if the game isn't aiming for realism, no problem, don't use it.

That quote is with regards to monetary and hardware budgets. The monetary budget you have a little wriggle room, but even that has a max. Unless you really think we should have games costing a few hundred dollars per copy?

Again, please see the multiple posts where I said, MS/Sony should step up to the plate here.

And hardware budget limitations are very real and very inflexible. That's why so much time is spent on finding approximations that can be done cheaply (with regards to computational resources). Again I'll use Crysis 2 here as an example. The Dx11 Ultra PC version still doesn't employ all features the developer wishes they could do. And that can bring even the most power PC available today to it's knees.

nextgen hardware.

If Rockstar can put this tech to work in every title they do on xb360/ps3, I'm sure nextgen machines can manage the workload for most games too.

What if someone wants to make a retro-styled game with Humans and is purposely NOT using physics/real human motions?

What would be the point?

But yeah I suppose they could always target xb360/ps3 if they want to stay in the past ...

Middleware can be good but it's not and never will be the "best" solution for a product. Blizzard recently dropped support of Havok, for instance, as they could better serve their projects by developing an in house physics package more suited to their game types.

Sure, I get that. But at the same time, (as you said) developers have limited time/money. The effort should be put in to improve the middleware either for the individual teams use, or for the betterment of the field by submitting more efficient code/routines back to MS/Sony to be used as part of the standard library/SDK.

 

[function]

Except Euphoria doesn't work that way. When a company decides to use them. They have to submit a detailed design document outlining their needs and what they are trying to accomplish.

Euphoria's company then attempts to implement what they "think" is the best way to achieve this. They then send that back to the developer making the game. The developer takes that makes another detailed report outlining what they got right and what they got wrong for the project at hand the sends that back.

The process is repeated however many times needed until the end product is close enough to what the developer needs, or the developer runs out of money for developement and they just go with what they currently have.

If you can afford the time and resources for not only Euphoria but the whole process of getting back the animations quality you want then it can be a decent solution.

Most companies with limited budget will be far FAR better off doing their own system as they are intimately aware of what their project requires. I believe ERP already touched on this.

Middleware can be good but it's not and never will be the "best" solution for a product. Blizzard recently dropped support of Havok, for instance, as they could better serve their projects by developing an in house physics package more suited to their game types.

I was trying to be humorous.

 

[ERP]

Again, please see the multiple posts where I said, MS/Sony should step up to the plate here.

OK so I'm not going to argue anymore on cost/value propositions, you have an incredibly simplified view of what games currently do, what value a technology like euphoria is actually providing and what's involved in developing one.

So what your suggesting is that Sony/MS provide a library for human motion?
Sure if it can be done right, which I don't believe today it can.

The problem is it would have to be a complete animation solution, you're asking a developer to turn ove a huge chunk of his game to whoever is providing the solution, losing for the most part the ability to optimize that code. If that code isn't a direct fit, your stuck working around it.
Now you're going to say, but they will have the source code, to which I say and it does you no good, because in general what you are looking to optimize involves an understanding of the code you are working with.

Lets take a look back and look at Euphoria's process for a moment, it involves close interaction with the developer, in no way is a plug and play library.
Now what has Euphoria spent on developing it? man hours? $?
How much would is cost MS or Sony to match that work?
How much more to make it plug and play?

If MS or Sony did provide a workable solution, you'd effectively stop everyone else working on it, so you'd also lose inovation in the space.

I'm not saying ignore the problem, I'm saying let the solutions emerge, having Sony or MS throw shit loads of money at the problem presupposes that it can be solved by just throwing more man hours at it.
I suspect the problem is waiting for the right piece of insight, not more man hours trying to fix Euphoria approach.

If I were attempting to "solve" the problem I'd start with MS's Kinect work, I think much of it is applicable, i.e. building a model for human posture by example. It's a completely different approach than Euphoria, might be another dead end.

 

[Shifty Geezer]

This is the thread for "Next Gen Gameplay ..." I'm pretty sure there will be additional computational resources. Hence, elevated expectations.

You were complaining about this gen, saying GTA4 had such-and-such a feature and so everyone else should have it too. Battlefield 1943 had destructible environments, so I guess Rockstar just weren't trying hard enough with their solid buildings...

Uncharted series

Uncharted already saturates the PS3, and you're saying that they're not trying hard enough or have their priorities wrong? Adding grenade damage to the terrain would have been impossible within the limits of the hardware and the rest of the game, and would have ruined the maps with their vertical gameplay. The invisible walls, though unnatural, are essential to being able to control the player view and cram those gorgeous visuals in. And you're saying they're not trying hard enough, or just aren't ballsy enough, shunning other technologies?? They already have a sophisticated animation technology with environment response, but it's well short of the Euphoria stuff and to go with a full behaviour-model would also have been impossible. Uncharted suffers most, IMO, from enemies that are clones. That's the first problem that needs to be addressed in many titles next-gen - repetition of content that destroys the illusion. Some games could compromise the visual content in favour of interactive content, but those would be obvious compromises. You greatly underestimate what it takes to implement features, both in terms of effort and performance requirements. This gen began with all sorts of hopes like Digital Molecular Matter and procedural content, but reality showed, even with perhaps the largest performance increase we'd ever seen between console generations, that achieving reasonable run-of-the-mill results at a constant decent-IQ 720p has been something of a struggle. There's not going to be enough power to run all the desired features every game next-gen. Devs will have to make choices as to what features get supported, same as every gen. And this magical middleware future where devs can buy off-the-shelf components that plug effortlessly together to build a game isn't going to happen, which I thought you'd have picked up on by now with every dev on this board who talks about middleware telling us that it doesn't work that way.

 

[Scott_Arm]

Also, there are games like God Of War 3 that really don't have any physics interactions between bodies at all. There may be some ragdoll when Kratos breaks free of hose pile-ons, but otherwise all of the animation is completely mo-capped or hand animated. The game does not suffer for it at all, and they would have had to make some sacrifices to fit in a more complex physical simulation between bodies and the environment. Probably not worth it, for what the game is.

And this is coming from a guy that is drooling over the Max Payne 3 vids.

 

[Shifty Geezer]

If I were attempting to "solve" the problem I'd start with MS's Kinect work, I think much of it is applicable, i.e. building a model for human posture by example. It's a completely different approach than Euphoria, might be another dead end.

I don't know what Euphoria's approach is, and for things like adaptive walk cycles (trip, stumble and recover, or walking up steps) it must be very complex. I would have thought a fairly simple system of inverse kinematics would work for a good number of cases though, fairly roughly. Instead of a fixed reaching animation, have the arm bones to the mesh and direct the hand towards the object. Of course, as Laa-Yosh would point out, it wouldn't look right if the rest of the body doesn't adapt to shift centre of gravity etc., but as an improvement over static animations, it'd perhaps be a step forwards. The main issue I see with this, only from what developers on this board have described over the years, is the art quality and creating meshed characters that are suitably flexible and versatile without issues. In the same way in theory you could have naturally animated human faces boned and skinned, but in reality even Hollywood offline rendering can't automate that effectively and animators have to touch-up digital sources. What sounds good on paper or in theory is always massively harder and more complex in practice.

Natural walking would also be a huge gameplay issue for many games. Many, many games have completely unrealistic avatar control, probably most shooters and all 3rd person action games, with turning on the spot and so forth. A realistic implementation of physics in their behaviour would make most impossible to play because the player would want the avatar to move in such-and-such a way and it'd just be physically impossible and the avatar'd fall over, or stumble around regaining balance while getting walloped. Heavenly Sword and similar action games only work because of predetermined animations that tightly knit around the control scheme. A realistic combat game with such mechanics would be awesome, but it'd be pretty niche I expect in the same way people couldn't get on with Lair's realistic approach to flying a heavy dragon. You'd have to adapt to its play-style. Thankfully devs have the option to implement whatever approach they want and it's not being mandated, so if your game is to have instant 180 degree turns to keep up the fast shooter experience, you're free to do that...

 

[TheChefO]

You were complaining about this gen, saying GTA4 had such-and-such a feature and so everyone else should have it too. Battlefield 1943 had destructible environments, so I guess Rockstar just weren't trying hard enough with their solid buildings...

I suppose this comes from not reading the posts in context, but the concept is take the best of what devs have done this gen, and let's establish a standard going forward for nextgen.

Then it's a matter of how to get to this standard and how to implement it.

Uncharted already saturates the PS3, and you're saying that they're not trying hard enough or have their priorities wrong?

Context.

If you read my post in context, you'd see it was a response to "realistic games". Or games which are attempting to replicate a realistic game world.

I'm not bagging on your dear Uncharted series (nor any other in the list) simply listing games which could benefit from standard libraries to improve human interaction, animation, behavior, and AI. And in some cases, real world environments...

 

[Shifty Geezer]

Not to have this thread bogged down with TheChefO's remarks about devs not trying, I'd expect to see realistic dynamic lighting next-gen. The progress made in CryEngine leads me to believe it'll be pretty standard, and it should save a lot of effort. Devs won't need to prebake light, and texturing won't need a lightmap. It'll allow more flexible levels and graphically be a huge leap in believability.

I hope to see more varied content, elliminating clones which plague current games. This needs some form of procedural content I think, mixing and matching parts. I'd like to see some procedural texturing too that varies content, but there's not much history of that.

AI needs progress, although I'm not quite sure what can be done in that regard. Path finding isn't new but it's still often broken. Is that just because of a lack of processing cycles? Or are current approaches too simplistic and we need a rethink of the behaviour routines? I also would like to see 'intelligent' player characters that don't respond to the player's buttons directly but which infer the player's intentions accurately and execute them correctly. eg. Just playing War in the North there'd be times when I hit the triangle button to finish a prone orc and the character has swiped to the side or something. Avatars are pretty dumb. And in shooters, for the game to understand when a player is shooting their gun towards an enemy, that the character actually executes the command. I know many existing gamers would be up in arms about such ideas, but for gaming to expand it needs to be accessible to people without awesome gaming skills. A decent amount of processing power could be given over to supporting them with 'soft' games that have plenty of aids to make the experience easily enjoyable and not frustrating for them.

Would like more dynamic worlds in games where it's appropriate, but that tends to lead to more bugs.

Core gameplay has been pretty good I think such that there's not obviously a good place to go.

Ulimately though I hope to see more creative use of gameplay. That's a software design thing and nothing whatsoever to do with technical limits. Games have felt very conservative this gen for whatever reason. A lot is a bit over-the-top too, and we could do with more art and less garishness.

Oh, and NO MORE DAMNED BUGS. That's not a gameplay feature really, but this gen has been abominable for bugs and broken games, which I find plain disgraceful. It's as if there's no pride in products any more. I know it's rare for the engineers on the ground to be happy with releasing broken products and its the powers on high that command it, but we gamers have suffered plenty of frustrations due to lack of QA and then shoddy follow-ups with missing fixes etc. Chances are though that more complex games will result in more errors.

 

[Scott_Arm]

...
Natural walking would also be a huge gameplay issue for many games. Many, many games have completely unrealistic avatar control, probably most shooters and all 3rd person action games, with turning on the spot and so forth. A realistic implementation of physics in their behaviour would make most impossible to play because the player would want the avatar to move in such-and-such a way and it'd just be physically impossible and the avatar'd fall over, or stumble around regaining balance while getting walloped. Heavenly Sword and similar action games only work because of predetermined animations that tightly knit around the control scheme. A realistic combat game with such mechanics would be awesome, but it'd be pretty niche I expect in the same way people couldn't get on with Lair's realistic approach to flying a heavy dragon. You'd have to adapt to its play-style. Thankfully devs have the option to implement whatever approach they want and it's not being mandated, so if your game is to have instant 180 degree turns to keep up the fast shooter experience, you're free to do that...

Check out the tech vids for Max Payne 3 and focus on his feet, how he shifts his weight and moves his upper body in relation to his lower body. They're promising responsive controls. We'll see!

But yeah, a shooter is one thing. If God of War had realistic reactions to everything, Kratos would spend mos to the time on his ass because just about everything he fights is a hell of a lot bigger than him. In that game you want response to input at all times. You can accept that he doesn't stumble around and get knocked all over the place, in fact barely responding (to being hit) at all, because he's a magically-strong godly roided freak. Realism isn't always a good thing.

 

[ERP]

I don't know what Euphoria's approach is

I believe they use a physical model with constraints and "muscles", then I believe the "magic" is based on an evolutionary model that teach neural nets how to react by applying forces. Again I believe it's based on Carl Sims work.

Simple IK has been a part of game animation for years, it looks very robotic, but it's the easiest way to deal with things like reaching for things.

I just think if you have the right model, you can reduce the problem to a reasonably sized multi dimensional space, you set constraints and select close points in that space,combine them to produce a final animation frame. It's very close to what Kinect is doing with it's model, only utilizing it differently. You'd probably still need some sort of physical model to set realistic constraints or targets.
Like all learning problems having the right model is hard, and creating a small enough state space to search is likely equally challenging.
I've seen papers on similar approaches, going back 9 or 10 years, but they were limited to specific tasks, notably reaching.

It's an interesting technical problem, if someone wants to throw shit loads of money at me I'd be happy to work on it ;P

 

[Scott_Arm]

...
It's an interesting technical problem, if someone wants to throw shit loads of money at me I'd be happy to work on it ;P

I think I have a spare ten-dollar bill floating around here somewhere. You start tomorrow.

 

[Shifty Geezer]

I suppose this comes from not reading the posts in context, but the concept is take the best of what devs have done this gen, and let's establish a standard going forward for nextgen.

If there's an issue with context than it's your ability to express it.
Your very first lines in this thread:

The interesting thing is we get certain "standards" to be adopted visually, but rarely do we get these standards adopted for gameplay.

The closest I can think of recently would be the widespread adoption of the cover system in FPS.

Other things though like facial & walking animation quality(LA Noir, GTA4, Red Dead), game-world interactivity(Oblivian, Fallout, Skyrim), Destructibility (Red faction) all seem to be limited to a franchise or individual company.

The reason the facial animation of LA Noir and and the destructibility of Red Faction aren't standard in more games is because they don't fit in with the game that are already stressing systems. They only work within a franchise or company that's designed itself around those key ideas. If they were adopted as standards than the rest of the game would suffer due to lack of resources.

Then you said:

As for destructible/deformable worlds, waaaay back when the original Pentium came out, a little game called Magic Carpet came out which had deformable terrain (along with support for 3D).

Yet we still don't see this as a standard feature. It can't be that demanding as it was running on the original Pentium architecture almost 20 years ago (typing that makes me feel old )

...as if most games these days should have it deformable terrain because it's old, despite that fact it's both technically demanding in a more varied game and has gameplay impact.

You weren't talking about next-gen, but what this gen has failed to do by consolidating an impossibly wide range of demanding features into the common game.

 

[TheChefO]

So what your suggesting is that Sony/MS provide a library for human motion?
Sure if it can be done right, which I don't believe today it can.

Euphoria isn't perfect, but that's the closest we have so far...

I'd suggest both Sony and MS both buy Euphoria in a joint venture (no point in trying to corner the market with one or the other as most games are multiplat so whatever technology is on one console WRT euphoria will find its way to the other console).

It is in both Sony and MS' interests to increase believability of both their nextgen consoles and the current methodology of developer involvement with Euphoria is limiting the adoption and progression of the medium.

The problem is it would have to be a complete animation solution, you're asking a developer to turn ove a huge chunk of his game to whoever is providing the solution ...

It would be part of the SDK. Integration with popular modeling packages would be ideal.

Lets take a look back and look at Euphoria's process for a moment, it involves close interaction with the developer, in no way is a plug and play library.
Now what has Euphoria spent on developing it? man hours? $?
How much would is cost MS or Sony to match that work?
How much more to make it plug and play?

It would likely be more beneficial to just buy euphoria out in a joint venture than to start from scratch.

If MS or Sony did provide a workable solution, you'd effectively stop everyone else working on it, so you'd also lose inovation in the space.

I don't see why this would be the case if devs could also contribute to the library (and/or modify it internall for their own selfish needs ).

The other aspect of what I'm proposing is that a dev could use their own fully custom solution. I don't think it would be worth it from a cost standpoint, but the bottom line would be a higher elevated standard for character interaction within a game world.

If the end result of the custom solution is equal to or better than the standard SDK library, go for it.

Just saying static animation cycles should be in the trash bin nextgen.

 

[TheChefO]

If there's an issue with context than it's your ability to express it.
Your very first lines in this thread:

...

You're right, I'm wrong.

If I haven't clearly identified my points by now, I'm not sure I ever will.

//