• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

EA Fight Night: Next Gen Render Tests

I still think the pictures are bull. I'll be happy to be proved wrong, but for the meantime you can call me on it. I don't think that this quality of images will be possible on inexpensive hardware in a realtime in-game environment.
 
radioheadrule83 said:
I still think the pictures are bull. I'll be happy to be proved wrong, but for the meantime you can call me on it. I don't think that this quality of images will be possible on inexpensive hardware in a realtime in-game environment.


they are renders, not actual gameplay pics, these are just to give an idea of what to expect. And why not? I'm expecting this
 
I think the hardware will be capable of doing what EA's claiming it will be able to do; the real x-factor will be the development teams, more specifically, the artists and character modellers. For example, there has only been one developer this gen to make even slightly convincing hair IMO, being Team Ninja (just look at DOA3). Others, like Yukes (look at any Smackdown! game) are way behind the 8-ball.

At some point, the issue stops being the machine and it starts being the operator.
 
Am I the first person to say that these probably qon't look quite so good in direct captures?! These images look stunning but they probably look more like an "actual" rendering job when they're not captured and blurred down in this manner.

These are bound to be based on Xenon specs, since PS3 doesn't actually exist.
 
I can't see the pics? Can anyone else see them now?

The link says "Images have been removed due to recent abuse."

EDIT: NVM, I just saw them on page 2.
 
radioheadrule83 said:
I still think the pictures are bull. I'll be happy to be proved wrong, but for the meantime you can call me on it. I don't think that this quality of images will be possible on inexpensive hardware in a realtime in-game environment.

Game hardware is rarely cheap to make, especially when it's launched/announced. Games consoles are often the most powerful pieces of consumer-level kit you can buy when first announced and released.
 
Kobold said:
I geuss there is some different ways to AA, but that's what I mean!

However, of they can render the game in a HDTV resolution, doesnt that mean you can still display it on a normal tv? (Sure not with the sharpness) I don't understand how that takes more VRAM or anything else? Your saying it costs more resources to show a HDTV picture on a normal TV then it costs to show it on HDTV?

How inefficient is that...you show a picture, but to show it in smaller form you need MORE memory then before. (this is what your saying?)

Nah, what i'm saying is that AA in general takes up more system memory than people can really afford with the current gen hardware.

HDTV resolutions aren't really 'above a normal tv' this generation. Most games are running in 480p which is 640x480 but is displaying a full 60 frames per second. 'normal' tv's run at 480i which is 640x480 but are only showing 60 fields (half frames) per second.

The actual resolution is the same, but a 480p image has twice the visual information to show (which is why it takes more work to render).

To get AA on that sort of image you'd have to render it at something like 1024x768 and then downsample it to display at 480.

Better AA routines take even more memory as they render the image several times (quincux makes 4 copies of the same image, but at the same resolution and shifts each image 1 pixel over from the original image and blends the 4 of them together.)

The simple fact is that all AA routines take more ram, and ram is a rare commidity in modern concoles (and likely will be in latter ones too).

In the next round of consoles if the game is meant to output in 1080p, feasibly you could downsample that same image out to 480i and it would AA the image on a normal TV, but it might do more to make the image 'fuzzier' than is intended due to the fact that a normal tv with a composite connection will tend to blur an image anyway. That's nothing current gen consoles can handle however.
 
Will phsyics be inproved? It would be awesome if you were playing an action game running through the streets and everything that applies to our law of gravity would be applied in next gen games.


p.s. I know there are games like that this generation, but I hope some of you know what I'm talking about.
 
AssMan said:
Will phsyics be inproved? It would be awesome if you were playing an action game running through the streets and everything that applies to our law of gravity would be applied in next gen games.


p.s. I know there are games like that this generation, but I hope some of you know what I'm talking about.

Physics should be more widely applied to incidental/ambient detail etc. I'm sure the implementation will be better quality too. From an animation perspective, Havok are actually working on an animation tool with a focus on physically based animation, so hopefully that might help in "stepping up" realistic animation for next gen games. I'm sure they're working on their next-gen Havok engine too.
 
Will phsyics be inproved? It would be awesome if you were playing an action game running through the streets and everything that applies to our law of gravity would be applied in next gen games.

On top of what gofreak said, just take a look at Havok's implementation in 3ds max 6 and 7. It's completely scalable, so the more raw processing power, the more accurate everything is. Even further, we're barely at the point where we can handle complex constraint systems (like brick walls that can be fractured and bridges with supports that can be taken out) in a complex real-time environment, so it'll be interesting to see what happens this gen.
 
radioheadrule83 said:
I still think the pictures are bull. I'll be happy to be proved wrong, but for the meantime you can call me on it. I don't think that this quality of images will be possible on inexpensive hardware in a realtime in-game environment.

Within the realm of shader technology, this level of rendering is not only possible but what you should expect. However it would be better to see actual gameplay shots and not a head in isolation because when you have to 'spread the love' around the arena and handle the fully 'game' pipeline you may not get anything that good.
 
At first, I thought that was a real person. Wow... next-gen graphics will look like that? Now I see why small independent developers will have to ally with big publishers.

It has nothing to do with "big developers". I expect to see SEGA's games look far better than EA's, when it comes to sports titles, despite being developed by developers such as Kush Games.
 
Vark said:
Nah, what i'm saying is that AA in general takes up more system memory than people can really afford with the current gen hardware.

HDTV resolutions aren't really 'above a normal tv' this generation. Most games are running in 480p which is 640x480 but is displaying a full 60 frames per second. 'normal' tv's run at 480i which is 640x480 but are only showing 60 fields (half frames) per second.

The actual resolution is the same, but a 480p image has twice the visual information to show (which is why it takes more work to render).

To get AA on that sort of image you'd have to render it at something like 1024x768 and then downsample it to display at 480.

Better AA routines take even more memory as they render the image several times (quincux makes 4 copies of the same image, but at the same resolution and shifts each image 1 pixel over from the original image and blends the 4 of them together.)

The simple fact is that all AA routines take more ram, and ram is a rare commidity in modern concoles (and likely will be in latter ones too).

In the next round of consoles if the game is meant to output in 1080p, feasibly you could downsample that same image out to 480i and it would AA the image on a normal TV, but it might do more to make the image 'fuzzier' than is intended due to the fact that a normal tv with a composite connection will tend to blur an image anyway. That's nothing current gen consoles can handle however.
Thanks Vark! :)

By the way, who's so sure this is a render? Maybe they beheaded the poor man and left his rotting head to dry!!!!
 
:lol

Seriously, this is the only thing to keep me excited about gaming now that halo2 has been released. It's time for the next gen to start.
 
Phoenix said:
Within the realm of shader technology, this level of rendering is not only possible but what you should expect. However it would be better to see actual gameplay shots and not a head in isolation because when you have to 'spread the love' around the arena and handle the fully 'game' pipeline you may not get anything that good.

That head's supposedly only 10,000 polys, no? That's only a tiny fraction of few million polygons per frame @60fps!

I'm so geeked thinking about it. :D
 
Shogmaster said:
That head's supposedly only 10,000 polys, no? That's only a tiny fraction of few million polygons per frame @60fps!

I'm so geeked thinking about it. :D

Polygons won't be the problem, but shading might. Doing that kind of shading on 100s of characters (as you might think with a simple vertex performance extrapolation) is probably not going to happen. Not to mention, the lighting model may be constrained to specific types of environments etc.
 
gofreak said:
Polygons won't be the problem, but shading might. Doing that kind of shading on 100s of characters (as you might think with a simple vertex performance extrapolation) is probably not going to happen. Not to mention, the lighting model may be constrained to specific types of environments etc.

In this case, we're expecting the two combatants to be of this shading quality, and the rest of the scenary/crowd etc to be rendered in less "intense" manner, so I still think this is well within the realm of a Xenon launch game in the form of a Fight Night game.

But then again, with the leaked specifications of the R500, maybe Xenon can render 100 of such detailed and shaded characters in another game! I really don't see anything that outrageous about that render test! Seriously.
 
If this truly is close to how games will look next gen, then the jump, visually speaking, will be the biggest i can recall.

Honestly, this is what i always wished games would look like when i was a little kid playing nes.
 
Kleegamefan said:
That *cant* be a realtime render.....can it??

Yes - it can and probably is. With dynamic bump maps and per pixel texturing and lighting you can get incredible results in a VERY low polygon count. Expect stuff on this level in the next generation - easily.
 
Shogmaster said:
That head's supposedly only 10,000 polys, no? That's only a tiny fraction of few million polygons per frame @60fps!

I'm so geeked thinking about it. :D

Indeed poly's aren't the problem anymore. Throwing polygon counts around is pretty pointless these days. The much larger issues are memory bandwidth, latency and availability. Throwing a metric assload of polygons on the screen doesn't mean anything if you can't actually "texture" all of them.
 
Kleegamefan said:
That *cant* be a realtime render.....can it??

I don't think this specfically is a realtime render. I would hesitate to say that it represents 100% what we will see next-gen either. Maybe come close..I'd be bowled over if this is exactly what we'll see though. This is as good if not better than a lot of Hollywood's attempts at digitising humans (though, admittedly, they took some "easy" routes, like the lack of hair etc.)

The other next-gen teasers were prerendered, AFAIK. I doubt this is any different. I don't think we even have 100% confirmation that this is legit EA material (I know they insisted on it being taken down, but that could easily be as much because it's fake - and thus heightening fan expectations too much - as because of it being real).
 
Top Bottom