What defines the next generation?
polycounter lvl 10
So a bunch of friends and i were wondering.. What are the new big fancy techs/techniques that will define the next generation of games?
In the past we saw the rise of normal maps, SSS attempts, DoF, bloom, dynamic tonemapping, etc. but what are we expecting to see in the coming few years? Which new techniques are we expected to take on as 3d artists? Which new tech would we like to see?
The list so far:
Simulations (cloth, hair, water, etc.)
Tesselation
GPU particles
Realtime GI and better lighting solutions
Improved AI
Advanced Animation, more procedural animation
More advanced Post Processing?
Improved tools
More shader headroom and compute shaders
Physically based lighting
Social media integration on everything
Forward+ or different render technique
Physically based shader models
Advanced skin shading
More custom user content
Possible Cone tracing solutions for lighting, reflection and refraction
Cats
*3d scanning
what else can you guys think of? I'll keep the list updated.
In the past we saw the rise of normal maps, SSS attempts, DoF, bloom, dynamic tonemapping, etc. but what are we expecting to see in the coming few years? Which new techniques are we expected to take on as 3d artists? Which new tech would we like to see?
The list so far:
Simulations (cloth, hair, water, etc.)
Tesselation
GPU particles
Realtime GI and better lighting solutions
Improved AI
Advanced Animation, more procedural animation
More advanced Post Processing?
Improved tools
More shader headroom and compute shaders
Physically based lighting
Social media integration on everything
Forward+ or different render technique
Physically based shader models
Advanced skin shading
More custom user content
Possible Cone tracing solutions for lighting, reflection and refraction
Cats
*3d scanning
what else can you guys think of? I'll keep the list updated.
Replies
Forward+ may eventually replace deferred rendering this generation, too. Especially as both of the big consoles use AMD cards (I think?).
Graphically, just look at what's being done on mid-high end PCs today and you're probably not far off.
Mostly I'd like to see better AI and non-graphics related things.
That reminds me of a meme!
But yes next gen ps4 coming 4 years or so next gen
Farfarer, i agree. Non-graphical improvements are probably what's needed more. Things like behavioral simulation that makes the world believable rather than the picture.
Interesting games allowing interesting interactions with an interesting world will shift titles in my direction. Or an unique "feel", like how Driver felt a million miles from Colin McRae which felt a million miles from Gran Turismo. The middleware, performance, experience and man-hours just weren't there so the developers couldn't use a fully comprehensive simulation, they had to choose what they wanted to include in their physics model and build it around that... made it feel a bit more defined as to what they were going for by contrast to EVERYTHING FEELS SOMEWHERE BETWEEN NEED FOR SPEED AND GRAN TURISMO we had this generation.
As above, I'd like to see the performance spent on things non-graphics. Moving gameplay code up into higher level languages should cut iteration time down dramatically making larger scope and better tuning easier than ever before.
Give it time, we're at the transistion still. Compare some ps3 launch titles to some of the recent games, huge difference.
Although (seriously) I think its going to be an increased reliance in shader functionality, better lighting, tessellation, etc.
I can see the demand for technical artists increasing substantially next gen although the role of your regular artist might not change as much.
The links appear to be chromatic aberration, shitty gpu particles, god rays fucking everywhere.
I think this will be the generation of the 'realistic stylised'. So stylised art direction for designs, but with realistic rendering. That Tangled look.
What do you mean, PS3/360 = next-gen?
Anyhow, the big one I think we're going to finally see is self collision. Up until now it wasn't feasible due to the low poly counts games had to adhere to, but thanks to tessellation thats all changed. If VR really takes off with the Rift then it'll happen even faster, as when you see characters own limbs clipping through the body or through enemies (which is completely normal in games currently) it really breaks a lot of the immersion.
[ame]www.youtube.com/watch?v=KDvfFzFIruQ[/ame]
Cloth physics will also become much more common, same with hair and possibly water. AI will only improve as a response to VR if anything, its just not important enough otherwise compared to all the other areas that need improving.
Well, going via a naive approach using VRAM alone:
PS3 = 256Mb
PS4 = 8Gb, shared. Let's give this a 3:1 GPU-CPU split, so 6Gb.
6144Mb/256Mb = 24x bigger.
Bearing in mind that each double in texture size quadruples the memory overhead, that leaves you with an average of two doubles of texture size, leaving 1/3 of the increase unused. As we'll be using more maps, but discarding precomputed lighting's memory overhead and using larger models, I'd argue that to be a fair estimate.
But, I doubt increasing polycount or texture size will improve the end game much.
- Lighting solutions
- AI
- Animation/blending
- Particle systems
- Post Processing
Artwork for games really can't get more advanced than it already is, unless we're dropping straight zsculpts, which would be stupid. We'll get a bit more wiggle room for more geometry and textures but really, our processes won't change. The tools will improve a hell of a lot though, which is the exciting thing.
OP said next gen, not this gen
Motion/Touch screen control methods
Social media integration on everything
Cameras/sensors under (or above) your TV
4k rend....no wait next time
Bunch of improved crap (lighting, effects, blah, blah) too
What won't define next gen:
Faceplates on game systems
One thing that is lacking in the released and announced stuff so far is a lack of new IP. Usually at console launches companies are more willing to gamble on creating new IPs more so than at the end of a console's lifecycle. I know Microsoft has yet to announce and e3 is coming. But I hope there is more new IP and not next gen sequels coming. There are some great franchises created this gen and previous gens. But I really would love to see team x or company y do something new with all the new tech and not iterate on previous designs. That said I'm super excited about Infamous Second Son (which is a sequel) and a few others. But all in all I'd love to see some new IP.
same thing. worse probably.
No, right now would be current-gen.
Seriously lads we're 8 months away from having PS4's and nextbox's under our tellys. Next-gen is those consoles; not the ones we have now. No need to confuse the issue. :thumbup:
Hey.....I already said that.:)
What's this Forward+ tech you mentioned? I've never heard of it. Why is it better?
Performance might be better than deferred because deferred can be pretty heavy on memory bandwidth due to needing to write to many G-buffers at once rather than a final scene colour as this would allow.
edit: here's the original video without commentary
[ame="
Basically if you want better results you need more skill and more time.
So if you wanna see some awesome non graphics stuff you need a studio that has a huge amount of programmers or just really good ones.
Improved shaders in a physically based shader model, better lighting, more resolution and anti alisting will be what sets this generation apart from the last in terms of graphics.
More emphasis on micro transactions, social media inner connection and user created content is really going to be the main cornerstone of this generation if you are talking about things that are not graphical specifically.
Geometry Tessellation wont be used as people believe it will be. Pixel displacement shaders will instead be used and very sparingly. Terrain would be the only real thing to use tessellation or tessellating characters during cut scenes.
Everything wont be dynamically lit and lightmaps will still heavily be used.
Some games will sticking to 720p to have superb anti alisting over 1080p. The human eye cannot even detect the difference between 720p and 1080p from 5 feet away.
The tech they show off cannot be extrapolated to a full game. They are not taking level size, number of characters, animations, sounds, scripted events, full level of textures and shaders, streaming of levels or anything related to an actual game into account beyond showcasing amazing art tech in the smallest of slivers to help sell whatever there product is.
Example. The UDK Samaritan demo was a "real time" demo that had to use 3 un-released nVida cards in tandem to run.
To get a real sense of what next gen games will be like dont look at tech demos. They give you a false impression. Instead look to the first slew of games coming out and the studios developing them to get a real sense of what is possible.
And right now that is the new Killzone game. That is the bar for high quality first generation next gen games. Guerrilla is one of the more talented developers in terms of graphical quality than the majority of studios. So expect most first gen games too look worse with your more tallented developers creating games around that quality level. And to extrapolate the jump in quality over the life time of the console look at the quality of Uncharted 1 to Uncharted 3.
From your top tier developers that is the kind of improvement you should expect over the console life cycle from a first generation next gen games to the end.
At least thats how I see things.
Visually, sure. But as far as actual experiences:
VR
[ame="
Galvanic Vestibular Stimulation
[ame="
tactile sensation feedback
[ame="
Mixed reality
[ame="
Nothing kills a joke like having to explain it
Lens flares (realistic ones only)
Cats
Procedurally generated AI path-finding
Realtime displacement animation
Physically accurate ambiance occluding
Lomography filters
Heart shaped bokeh
Sorry but screen based AO really bothers me.
Yeah, ok serious response time then:
Voxel cone radiosity, this shit is awesome.
https://www.youtube.com/watch?v=fAsg_xNzhcQ
playable demo here: http://www.altdevblogaday.com/2013/01/31/implementing-voxel-cone-tracing/
Also, this is what UE4 and the latest Crytek stuff is using, afaik.
[ame="
I love seeing videos like this and then researching everything they talk about so I understand better. I remember seeing something about how NVidia and AMD usually share their new techniques, so does this mean NVidia cards will eventually get this feature as well?
Also dynamic environments/world, not only destructible but construction without having to following a blueprint. Imagine a DayZ game, where you use a sledgehammer and breaks a desk apart and use the smaller pieces for fireplace and the larger parts to hammer on the windows to block them, without the whole "Can only place them here and there" issue. Hell, might break a hole in a wall and the border it up the hole again.
I believe that was our very own commander_keen
Btw the Samaritan demo runs completely fine on a normal PC, even back then. The only reason it used 3 powerful cards was because they turned up the AA very high and the transparency AA killed it.