The Nintendo 3Ds

[klikklak]

Yes and my point was that if the 3DS wasn't powerful enough to comfortably handle this, Nintendo would have known by now, so what exactly is the problem here?

Yes, but when looking at images of games and evaluating them based on polygon counts, you're ignoring the fact that it draws it twice. Of course it can handle whatever is specifically made for it. If they made a tamagotchi machine and it could comfortably run 16 colours, it is powerful enough to handle it. But when you compare it with other machines, it doesn't look that underpowered, but rather relatively impressive.

EDIT: ehh, posted before reading all the new posts. We agree.

EDIT2: Since they've put the 3D in the name and the screen there, I'm guessing non 3D games will be as rare as non-wiimote games on the Wii. So it doesn't matter that it could render twice as pretty games in flat 3D, if all the developers just do S-3D.

 

[riumplus]

From my experiences with nVidia's 3D Vision, the performance doesn't have to halve to maintain the same image quality when you go 3D. Say you're playing a game at 1024x768 with 8x standard antialiasing. When you render it twice from two different angles, this essentially acts as supersampled antialiasing on any jagged lines. A similar thing happens with anisotropic filtering but not as much. If you keep it at the same quality settings then the framerate halves but the resulting image looks MUCH better. If you want to keep the same image quality you can turn down a lot of the quality settings.

Plus if you've coded the game for 3D-ness then any UI elements at 0 depth get drawn only once across both eyes.

The extra horsepower required for the same image quality is only around something like 70% more, not a full doubling... And it's not like the 3DS' resolution is that high. 400x240 per eye? That's a quarter the Pandora's pixel count! Chuck on a huge R&D budget with gobs of employees to design the system, and even with Nintendo's track record of using slightly dated gaming hardware it's exactly what I expected.

Easy for handheld hardware to do this IMO. Chuck a lenticular lens on the Pandora and you've got handheld hardware capable of doing fluid 3D at 400x480. And this is 400x240.

 

[quote]

Yeah klikklak I guess you're right. All games developed for the 3DS will be in 3D, or won't sell very well.

How is that guy moving the camera around Snake without a second nub?

It's beyond me why they just don't add another nub so that FPS's can be played properly.

n00b dumb c*nt!

Quoted to prevent editing.

 

[Exophase]

Plus if you've coded the game for 3D-ness then any UI elements at 0 depth get drawn only once across both eyes.

I doubt it works this way. You either have two framebuffers (or partial framebuffers, if it's augmenting the DS design, which I have a feeling it will be) or one framebuffer. You can't share a framebuffer for one element. You could have multiple render targets and in theory could have something render to both framebuffers, but I doubt 3DS has this feature.

The extra horsepower required for the same image quality is only around something like 70% more, not a full doubling...

How do you figure this? We're talking strictly graphics acceleration, right? Naturally your high level scene culling, physics and game logic wouldn't be doubled, but I'm scratching my head in trying to determine what would make it only cost 1.7x on the GPU. The only place I can think of where you save anything is that a single geometry pipeline could output results for both scenes - it'd still have to do 2x the amount of actual work but there wouldn't have to be duplicated input. But if this is handled asynchronously (fixed geometry engine like on DS, or non-unified vertex shaders) then I don't think it'll make a big difference.

If this is an agumented DS like I think it might be then it'll be render locked at 60Hz anyway, with a fixed polygon budget... you may end up seeing that you can't get a boost by turning off the 3D display.

Yeah klikklak I guess you're right. All games developed for the 3DS will be in 3D, or won't sell very well.

I believe this about as much as I'd have believed that all DS games would have to be 3D (you know, the classic kind, not the display stuff) or not sell very well. In reality many high selling DS games were 2D, so we'll see.

On the same token, there's little reason for 3D games to not use the 3D display.

 

[klikklak]

I believe this about as much as I'd have believed that all DS games would have to be 3D (you know, the classic kind, not the display stuff) or not sell very well. In reality many high selling DS games were 2D, so we'll see.

On the same token, there's little reason for 3D games to not use the 3D display.

I think even brain training programs will use the screen in 3D since it doesn't require glasses. Actually seeing how they use it is going to be quite interesting.

 

[puppydee]

Oooooo, another handheld that I'll never play but will buy anyway just so I can say I've got one. How long before someone asks for an Ds3d emulator for the pandora NO incase anyone wonders

 

[Nova]

I believe this about as much as I'd have believed that all DS games would have to be 3D (you know, the classic kind, not the display stuff) or not sell very well. In reality many high selling DS games were 2D, so we'll see.

On the same token, there's little reason for 3D games to not use the 3D display.

I think even brain training programs will use the screen in 3D since it doesn't require glasses. Actually seeing how they use it is going to be quite interesting.

 

[riumplus]

The extra horsepower required for the same image quality is only around something like 70% more, not a full doubling...

How do you figure this? We're talking strictly graphics acceleration, right? Naturally your high level scene culling, physics and game logic wouldn't be doubled, but I'm scratching my head in trying to determine what would make it only cost 1.7x on the GPU. The only place I can think of where you save anything is that a single geometry pipeline could output results for both scenes - it'd still have to do 2x the amount of actual work but there wouldn't have to be duplicated input. But if this is handled asynchronously (fixed geometry engine like on DS, or non-unified vertex shaders) then I don't think it'll make a big difference.

If this is an agumented DS like I think it might be then it'll be render locked at 60Hz anyway, with a fixed polygon budget... you may end up seeing that you can't get a boost by turning off the 3D display.

The moment you turn on 3D you've essentially got built-in 2x supersampled antialiasing from having two differently-rendered images overlapping in your brain. This lets you have the same perceived image quality while doling back the rendering complexity. From my own experience on my 120Hz screen & a nVidia 3D Vision set, 4x MSAA/4x AF in 3D looks equivalent to 16x MSAA/16x AF in 2D. Run the game in 2D and I need 16x, run it in 2D and I'm fine with 4x. May be at different points for other people but that's where it is for me. The difference that provides in framerate would vary greatly on the hardware's weak points but I thought 1.7x over 2x is a reasonable guesstimate. It's about what I get with my GTX 260+. On something limited by memory bandwidth, that decrease in required AA/AF could do much more.


Wish there was more details on its innards! I predict there'll be some interesting 3D augmented reality uses with those dual cameras, whether official or homebrew.

 

[1magus]

No matter what they do, it'll be impossible to out show the 3DS. What are you going to do? Dual analog? Meh. Higher res screen and more graphics power? Nobody cares. While the 3DS isn't perfect, it's so far ahead of the game that the only option Sony has is to more or less clone it. Even then, people will prefer the original. Look at the list of 3DS games! You simply cannot beat that. I think Sony should just give up handheld gaming...

A full PS2 download service for PS3 and PSP2 would get some people to rethink on the 3DS, if it had Dual Analog sticks than you are going to see a lot of people interested in the PSP2. PS2 games on the go would be amazing, it would probably drive the price up however.

 

[Goity]

Craig, you've been saying this stuff ever since Sony sued you. It's understandable you feel this way, but it's clear you have a bias as a result of this.

What? That the PSP2 won't be open? I also doubt that will happen...

I mean the general Sony hate / Sony is surely doomed speak. Of course the PSP2 won't be open.

 

[migo]

You can't really cut corners with it. I mean, if you need to render a scene from two angles you need to render twice. I fail to see how this could be skipped.

There's still the difference between rendering everything and only what's visible, I remember when I was comparing graphics cards for PCs around 1998 that was an issue for some cards and not others. If they're rendering for each eye to get 3D they're certainly not going to render the whole scene twice, only what's visible. There's also the thing with S3TC that the Savage4 and Savage2000 had, which while being less powerful than the TNT2 and GeForce 256 put out better image quality and higher performance for games specifically coded for them. Obviously, with the same hardware rendering one image instead of two is going to be easier, but I'm sure Nintendo and developers will find a way to get good looking graphics out of the system with the hardware that's included.

 

[Exophase]

The moment you turn on 3D you've essentially got built-in 2x supersampled antialiasing from having two differently-rendered images overlapping in your brain. This lets you have the same perceived image quality while doling back the rendering complexity. From my own experience on my 120Hz screen & a nVidia 3D Vision set, 4x MSAA/4x AF in 3D looks equivalent to 16x MSAA/16x AF in 2D. Run the game in 2D and I need 16x, run it in 2D and I'm fine with 4x. May be at different points for other people but that's where it is for me. The difference that provides in framerate would vary greatly on the hardware's weak points but I thought 1.7x over 2x is a reasonable guesstimate. It's about what I get with my GTX 260+. On something limited by memory bandwidth, that decrease in required AA/AF could do much more.

Wish there was more details on its innards! I predict there'll be some interesting 3D augmented reality uses with those dual cameras, whether official or homebrew.

I understand your argument now, but on 3DS I doubt there'll be any kind of option for super or multi-sampling AA at all (you'll notice almost none of the shots had any AA - DS had it for free with coverage based edge-AA, but it wasn't full scene obviously). So I think it's the whole comparison is moot.. what really matters is how the same geometry compares with 3D vs not with everything else kept the same, and the load on the GPU will probably be close to 2x.

I will like to see how things pan out for no-AA, how well the merged images overcomes the problem like you say. In practice I think it won't quite work out like 2x supersampling, because the variance will be dependent on depth of field.

I mean the general Sony hate / Sony is surely doomed speak. Of course the PSP2 won't be open.

Thinking that the odds are against Sony doesn't mean you hate them. Although they have given a few reasons to hate them too, not that I'm personally bothered about anything.

IMO PSP did well - about as well as I expected it to do from the offset, and much better than I expected it to do after a couple years in when it was clear UMD movies were a failure and the faults were more evident. The graphical advantage over DS was very real, the two screens weren't a huge deal (and it had much higher resolution), and the touchscreen wasn't really everyone's bag, and was even kind of offputting for some games. But I think Sony is in worse shape here, because:

(oh god it's another one of exophase's bulleted lists look out)

- 3DS is announced first and has tons of footage to show for itself. No sign of PSP2 in sight - you'll recall PSP was announced long before DS was, and this gave it a marketing advantage. Sony usually announces its platforms long before release and we haven't heard good rumors of actual hardware in the field (just some bit about what IP they're going to want to use), so chances are good that PSP2 will come out much later than 3DS. Granted, the way I see it, if you weren't in first place coming out first can actually hurt you, since people will be waiting for the next thing, but at this point no one even knows for sure if there will be a PSP2.
- There's no way that Sony can show nearly as much vendor support for a new platform, Nintendo's got way too much of a lock on this already. PSP didn't show a lot of vendor support early on either, but 3DS is showing way more than DS was.
- 3DS has backwards compatibility for sure, and I think the chances of PSP2 having it are slim. From PSP-Go it's clear Sony doesn't want to support UMD again (for good reason), nor a way to transition from UMD for free or reduced cost. There'd still be some value in being able to purchase downloadable PSP games for PSP2, but I think that PSP2 won't have the capability to play them in hardware because we're unlikely to see MIPS again on a Sony handheld and they probably wouldn't want to carry a 3D design that's pretty inefficient by handheld standards. There could potentially be software emulation, but I think Sony isn't that well prepared to do such a thing either (if PS3 is any example).
- 3DS brings a lot to the table that's new to their handheld line and really addresses the weak points of DS. PSP just has less room to grow without outright copying 3DS, which wouldn't be well received.
- If the rumors of 4x SGX543 are true then PSP2 is going to lose substantially on cost and power consumption again.
- Sony's just been hurting badly in general, with all the lawsuits, bad advertising, and PSP-Go they don't have a good reputation right now. PS3 has rebounded nicely since the slim came out, but I don't think that'll help a new handheld.

I think dual analog would help them. Some other innovative input method could as well, although I'm at a loss as to what.

One interesting thing about 3DS's design is that it's actually detracting from the touchscreen, since you can't use it on their nice screen. Touch was the biggest feature of DS, IMO, and without being able to use it with the 3D display (even when not used as 3D) you'll mainly just be using it as passive input than touching active things. It's probably just as well, since I imagine it doesn't really mix with 3D vision.

 

[Goity]

A PSP emulator on the theoretical PSP2 isn't that unfeasible. After all, Sony are already emulating the PSP on PS3 (specifically the minis, which are coded in the same way as regular PSP games), the PSP is mostly virtualised hardware anyway from a development point of view, and Sony have made emulation efforts on the PSP with PS1 emulation. I suppose it just depends how powerful the hardware would be.

A dual analog could make a PSP2 an attractive proposition. After all, certain genres do sell very well even on the PSP (though the attach rate is disturbingly low), and a decent input method could allow better conversions from console games. Of course, it depends how powerful this console would actually be as to whether it could differentiate itself from the 3DS enough for people to want to buy it.

I wonder what the battery life will be on the 3DS? The original DS' battery life wasn't awesome, and I hope the 3DS doesn't have similar issues.

 

[Exophase]

A PSP emulator on the theoretical PSP2 isn't that unfeasible. After all, Sony are already emulating the PSP on PS3 (specifically the minis, which are coded in the same way as regular PSP games),

I hadn't heard about this, but:

http://www.eurogamer.net/articles/digitalfoundry-psp-emulator-analysis-blog-entry

Get back to me when there's something really generally compatible and fully featured and not just playing a subset of games specifically designed to run on the emulator. I also don't think PSP2 is going to have processing power even remotely comparable to PS3, and the PSP emulation there isn't even pulling full speed for 3D games that are not exactly top tier in what they're pushing. If anything this is a good example of how Sony would struggle with PSP emulation on PSP2.

the PSP is mostly virtualised hardware anyway from a development point of view,

It's not really, in the sense that the three computationally heaviest components basically aren't: the CPU, the VFPU, and the GU. The interface to the GU is basically a low level hardware interface involving directly writing command lists to the thing and directly managing its VRAM, and the VFPU is a CPU coprocessor. The Media Engine and other peripherals are abstracted, so from a "what do you have to implement" point of view a lot is saved, but from a "what is going to steal your performance" point of view things are different. I will admit that it's possible to do faster PSP emulators by virtue of the abstraction in at least some ways.

Something like say, 80% compatibility with emulation sounds great as a technical accomplishment but those kind of numbers get you slammed by consumers, especially if it misses on any of the really top tier games. When buyers see that they see the thing as broken, full stop. Performance problems only make things worse.

and Sony have made emulation efforts on the PSP with PS1 emulation. I suppose it just depends how powerful the hardware would be.

Don't compare PS1 emulation to PSP emulation. Or if you do, first compare the progression of PS1 emulation on PC with that of PSP emulation.

A dual analog could make a PSP2 an attractive proposition. After all, certain genres do sell very well even on the PSP (though the attach rate is disturbingly low), and a decent input method could allow better conversions from console games. Of course, it depends how powerful this console would actually be as to whether it could differentiate itself from the 3DS enough for people to want to buy it.

You're assuming that performance gap could be interesting at all - my opinion is that 3DS may already be reaching a "good enough" saturation point for handhelds that'll cut out a lot of the appeal.

I wonder what the battery life will be on the 3DS? The original DS' battery life wasn't awesome, and I hope the 3DS doesn't have similar issues.

Wasn't awesome relative to what? It was better than PSP's..

 

[Goity]

Well, the PSP minis aren't designed specifically to run on the PS3 emulator. The minis came out several months before they became compatible with PS3, and all minis work, with the only real difference between minis and retail games being lack of scope and no use of network hardware. This suggests the PSP emulator must be pretty reasonable on PS3. Added to this, the emulator used for the Digital Foundry analysis is quite old now; Sony may well have improved it since (as they have done repeatedly with their PS1 emulator).

And even with 80% compatibility, they could realistically just limit available games on the PSN store (assuming the PSP2 doesn't use UMD). This way nobody would really know the difference.

I was using the PS1 point to show that Sony are willing to invest in emulation on their consoles, not as a direct comparison of difficulty.

As for the performance gap? That's a very valid point. The 3Ds should have enough power to do most types of games, so is there a requirement to go further? However, I think the answer is probably yes, mirroring the situations visible in the console space where people are interested in the best graphics.

The DS' battery life is better than the PSP's (damn thing drains itself after 2 weeks!), but compared to all of Nintendo's previous consoles and the DS lite, it was a marked step back. I wonder if with its first iteration of the 3DS it will take the same philosophy of low battery life?

 

[Nova]

I want this one:

 

[Exophase]

Well, the PSP minis aren't designed specifically to run on the PS3 emulator. The minis came out several months before they became compatible with PS3, and all minis work, with the only real difference between minis and retail games being lack of scope and no use of network hardware. This suggests the PSP emulator must be pretty reasonable on PS3. Added to this, the emulator used for the Digital Foundry analysis is quite old now; Sony may well have improved it since (as they have done repeatedly with their PS1 emulator).

You don't know that the minis weren't designed with the goal of their software being emulated in mind. Even if that isn't the case, it's still much easier to reliably emulate a much smaller set of games, especially when there are barely any 3D ones. The trickier games can have emulation hacks designed just for them - you can't feasibly pull this off on a library of several hundred.

That analysis wasn't that old. Maybe Sony will improve it like they have their PS2 emulation in PS3, by removing it entirely.

And even with 80% compatibility, they could realistically just limit available games on the PSN store (assuming the PSP2 doesn't use UMD). This way nobody would really know the difference.

Yes and no. Then users are just going to ask why Sony isn't selling their favorite games. Any restrictions will hurt them - selling the games in the first place will put the feature at a huge disadvantage anyway. I really doubt PSP2 will use UMD, I'd actually bet money on that.

I was using the PS1 point to show that Sony are willing to invest in emulation on their consoles, not as a direct comparison of difficulty.

Every current manufacturer is willing to emulate, that doesn't mean anything regarding what they can do.

[quote name='Goity' date='18 June 2010 - 09:25 PM' timestamp='1276892746' post='877729]As for the performance gap? That's a very valid point. The 3Ds should have enough power to do most types of games, so is there a requirement to go further? However, I think the answer is probably yes, mirroring the situations visible in the console space where people are interested in the best graphics.[/quote]

I think people care less about graphics quality on the small handheld screens, but I can't really substantiate this. It would be naive to think no one wanted better than 3DS, of course, it's just a question of how many and how much.

[quote name='Goity' date='18 June 2010 - 09:25 PM' timestamp='1276892746' post='877729]The DS' battery life is better than the PSP's (damn thing drains itself after 2 weeks!), but compared to all of Nintendo's previous consoles and the DS lite, it was a marked step back. I wonder if with its first iteration of the 3DS it will take the same philosophy of low battery life?[/quote]

Oh okay, you weren't talking about DSL. I'd say Nintendo realized they were cutting corners too much on DS's battery capacity, but the DSi's return to the same capacity puzzles me. So I don't know which direction Nintendo will go in - I think it'll be purely a question of price to them, not size. They could go with < 1000mAh and save a bit or ~1500mAh like current high end phones. I expect the latter is more likely. I don't think we'll be seeing any monster batteries, nothing like Pandora size obviously, that's just way too big.

 

[Goity]

The minis clearly weren't designed with emulation in mind. Most of them are essentially indie titles of the scope seen on the iphone, and the development tools, at least according to Digital Foundry, are more or less the same as with full psp games. If Sony had such a scheme, we'd know.

And no, the analysis isn't that old, but it was done when the minis emulator was first new. Since then it has noticeably improved; initially the emulator was compatible with around 6 games, with that extending to the whole library within a month.

As for the PS2 emulation? That was a choice of cost on Sony's part. Even in the later BC models, the PS3 still had the PS2 graphics chip in, which cost too much money when the PS3 was losing money and was hideously priced. There's no feasible reason for Sony to remove a software-only solution.

A limited range of software available hasn't hurt consoles his generation; the full Nintendo backlog isn't available on the VC, and only a limited number of PS1 games are on PSN. I don't see why Sony couldn't do the same next gen.

 

[migo]

I think dual analog would help them. Some other innovative input method could as well, although I'm at a loss as to what.

They could go with touchless interface with the 3D image. I've seen some tech demos where you're looking at the screen, the image pops up and you have your hand over it manipulating the image that way. It would be killer with a game like Echochrome or Zen Bound. They could also do something like Kinectimals with that too. One of the 3DS devs mentioned that he wanted to touch the image, so actually giving that capability would be a logical step.

One interesting thing about 3DS's design is that it's actually detracting from the touchscreen, since you can't use it on their nice screen. Touch was the biggest feature of DS, IMO, and without being able to use it with the 3D display (even when not used as 3D) you'll mainly just be using it as passive input than touching active things. It's probably just as well, since I imagine it doesn't really mix with 3D vision.

I think the strength of the touch screen is like when the GBA was used as a controller for one of the GameCube games, showing stats and such on the GBA so you don't have to go on the main screen. Having say a FPS where most of the HUD stuff would go on the touch screen would be a good use of it.

 

[Javacat]

I think the strength of the touch screen is like when the GBA was used as a controller for one of the GameCube games, showing stats and such on the GBA so you don't have to go on the main screen. Having say a FPS where most of the HUD stuff would go on the touch screen would be a good use of it.

I've got to agree with this. I actually hate games on the DS which spread the action over both screens. There aren't that many games which do it, but the ones that do, I just usually give up on after a few minutes of playing.

I like that the DS3 has clearly defined the screens as different entities now (by not making them the same size, or reducing the space between them). It now makes the touch screen as an additional input or a 'peripheral screen' (or whatever the word is), used for stats, maps and misc info.



I've also been thinking that the 3D screen could be used to make some interesting 2D/3D hybrids. Like you could actually define some distance between the background and foreground in something like Sonic 2 or Mario Bros. 3, whilst keeping everything 2D.