Support NeoGAF

[krizzx]

No it doesn't.Sorry, There's probably not more geometry on Zelda TP Link's face either.

Yes, it does. Sorry. This is one thing I will definitely have to say you are wrong on.

That was actually a bad choice to post to me to being with because I was on the scene when people analyzed that when it first released. I already knew that all of the DS Mario 64 models had more polygons by heart. Its even on the wikipedia page and a lot of other places around the web if you want to verify.


http://upload.wikimedia.org/wikipedia/en/d/d1/Super_Mario_64_DS-Graphics_comparison.jpg


Can't remember the exact number(they actually had the count), but Mario's modal has about 20% more polygons compared to his n64 modal. Analyzing levels of graphical details is one of my specialties. I can quicly analyze the level of animation, "extent" of graphical effects in use and the amount of detail that is drawn on a texture vs real geometry.

 

[phosphor112]

It's slower for sure, but that might not be an issue.

As Shin'en put it:

I don't see why you guys always quote Shin'en. Everything that isn't the Wii U is limited by bandwidth, NOT latency. Latency is still important, yes. That's why Durango has eSRAM and why PS4 has heavy cache and bus customizations to avoid GDDR5 problems, but when it comes to games. Bandwidth is the biggest issue because most of the processing is graphics.

 

[Schnozberry]

What exactly are the benefits of lower latency exactly?

Hopefully less wasted CPU/GPU cycles waiting for data. When you're trying to get maximum performance out of your hardware while running memory intensive applications like games, latency is more important than it would be for doing more general computing like browsing or spreadsheets, for instance. Coupled with the EDRAM and Caches, the Wii U memory hierarchy is pretty well designed. I don't think the system is bandwidth starved as many have claimed.

 

[lostinblue]

I don't see why you guys always quote Shin'en. Everything that isn't the Wii U is limited by bandwidth, NOT latency. Latency is still important, yes. That's why Durango has eSRAM and why PS4 has heavy cache and bus customizations to avoid GDDR5 problems, but when it comes to games. Bandwidth is the biggest issue because most of the processing is graphics.

I could have not quoted them. What I said can stand on by it's own.

Nobody knows how much impact lacking bandwidth has on a real world scenario such as this; but I'm sure those were taken into account. Still this console it's a balancing act so who knows.

Thing we know for sure is that we've heard developers complaining about other stuff in there (CPU) and not bandwidth; might not be an issue, and even it it is it might be a matter of doing things a little bit differently (like using the extra eDRAM as texture cache like the GC did)

Also, we're comparing apples to oranges, PS3 and X360 were bottlenecked to hell and back, this one makes an effort to remove those bottlenecks; eDRAM is embedded into the MCM, that makes it be less cycles away, and they embedded extra memory on both CPU and GPU. Even CPU/GPU communication, on the X360 when switching to an MCM they had to insert a delay into the interface block because the parts were so much closer now. This thing is supposed to be tight.

All the losses of performance here and there might just make it so that less bandwidth on paper doesn't amount to much on a real world scenario. wouldn't be the first time too, I reckon GC looked unimpressive in paper compared to xbox.

Yes, it does. Sorry. This is one area I will definitely have to say you are wrong on.

That was actually a bad choice to post to me to being with because I was on the scene when people analyzed that when it first released. I already knew that all of the DS Mario 64 models had more polygons by heart. Its even on the wikipedia page and a lot of other places around the web if you want to verify.

Can't remember the exact number(they actually had the count), but Mario's modal has about 20% more polygons compared to his n64 modal. Analyzing geometry details one of my specialties.

I actually know what I'm talking about.

Wait a while, I'll have to go through old backed up files on the external disc.

Also, since we're pulling out being good at pointing something polycounts, I should say I'm a 3D modeler (not strictly, but I do 3D modeling and was trained to do so on an academic level on various software suites, I think I know a thing or two most don't).

EDIT: I'll prove you're wrong without even touching the DS model.

 

[phosphor112]

i blame internet for double post -_-

 

[HTupolev]

ZombiU uses spherical harmonics to render it's lighting, which is a solid proof that it's not a totally pre-backed radiosity effect

Using spherical harmonics to calculate static light maps has been done all the time, like in Halo 3.

Spherical harmonic transfer functions are really good at representing the effects of infinite-distance environment lights. I've heard of people BSing spherical harmonic shadowing by representing a point light as a web of sample points with different spherical harmonic representations. If they're doing spherical harmonic realtime GI from non-infinite lights as you claim, this would get really sketchy, as you break your assumptions about the angles and magnitudes at which light is striking other objects in the scene. If anything, this seems like a very expensive way to get very flawed GI, and I'd be surprised if it's what they're doing.

 

[Easy_D]

Yes, it does. Sorry. This is one thing I will definitely have to say you are wrong on.

That was actually a bad choice to post to me to being with because I was on the scene when people analyzed that when it first released. I already knew that all of the DS Mario 64 models had more polygons by heart. Its even on the wikipedia page and a lot of other places around the web if you want to verify.


http://upload.wikimedia.org/wikipedia/en/d/d1/Super_Mario_64_DS-Graphics_comparison.jpg


Can't remember the exact number(they actually had the count), but Mario's modal has about 20% more polygons compared to his n64 modal. Analyzing levels of graphical details is one of my specialties. I can quicly analyze the level of animation, "extent" of graphical effects in use and the amount of detail that is drawn on a texture vs real geometry.

I've seen evidence posted on GAF that says the DS model uses fewer polygons. All the sources on the internet that say DS has higher polygon counts don't provide any technical sources (Just like you aren't ). When someone brings up the facts, prepare to have your mind blown

 

[lostinblue]

I've seen evidence posted on GAF that says the DS model uses fewer polygons. All the sources on the internet that say DS has higher polygon counts don't provide any technical sources (Just like you aren't ). When someone brings up the facts, prepare to have your mind blown

Relax, I have his back. It's gonna blow alright

 

[wsippel]

I don't see why you guys always quote Shin'en. Everything that isn't the Wii U is limited by bandwidth, NOT latency. Latency is still important, yes. That's why Durango has eSRAM and why PS4 has heavy cache and bus customizations to avoid GDDR5 problems, but when it comes to games. Bandwidth is the biggest issue because most of the processing is graphics.

But how much bandwidth does the Wii U actually have? It has four "main" memory pools and several dedicated local pools/ caches (Latte blocks D and X for example) - we only know the bandwidth of one of those.

 

[krizzx]

Wait a while, I'll have to go through old backed up files on the external disc.

Also, since we're pulling out being good at pointing something polycounts, I should say I'm a 3D modeler (not strictly, but I do 3D modeling and was trained to do so on an academic level on various software suites, I think I know a thing or two most don't).

EDIT: I'll prove you're wrong without even touching the DS model.

http://en.wikipedia.org/wiki/Super_Mario_64_DS
http://zonenintendo.weebly.com/super-mario-64-ds.html
http://inintendo.net/Articles/FarewellDSp1.php
http://www.gameyum.com/other-mario-games/63197-super-mario-64-ds-nintendo-ds-review/
http://www.nintendocosmos.com/super_mario_64_ds-info.htm

I've seen evidence posted on GAF that says the DS model uses fewer polygons. All the sources on the internet that say DS has higher polygon counts don't provide any technical sources (Just like you aren't ). When someone brings up the facts, prepare to have your mind blown

I just posted a clear shot of how much more detailed Bowerser was. I doubt Nitnendo gave anyone actual documents with the words, "The DS version has more polygon's" written on if you are looking for official technical documentation. I figured it would be a plain as day from just the photos and the mulitple proffesional analysis.

On top of that, I actually posted things to back up what i stat. You are outright dismissing what I provided while providing no proof to the contrary at all.

 

[Red UFO]

http://en.wikipedia.org/wiki/Super_Mario_64_DS
http://zonenintendo.weebly.com/super-mario-64-ds.html
http://inintendo.net/Articles/FarewellDSp1.php
http://www.gameyum.com/other-mario-games/63197-super-mario-64-ds-nintendo-ds-review/
http://www.nintendocosmos.com/super_mario_64_ds-info.htm

Please post technical evidence/polycounts, not just reviews from website we've never heard of that happen to agree with you.

 

[Easy_D]

http://en.wikipedia.org/wiki/Super_Mario_64_DS
http://zonenintendo.weebly.com/super-mario-64-ds.html
http://inintendo.net/Articles/FarewellDSp1.php
http://www.gameyum.com/other-mario-games/63197-super-mario-64-ds-nintendo-ds-review/
http://www.nintendocosmos.com/super_mario_64_ds-info.htm

None of those sites cite a source. They're only going by what you are, what they can see with the naked eye. I doubt any of those sites ripped the respective models and counted the polygons in a wireframe view. I'm not pointing fingers here, but believe me when I say I know I'm right. I wouldn't post if I weren't. I looked for a source but couldn't find one :|. Luckily lostinblue is on the case.

Edit: I'm not trying to come off as a dick by the way, so sorry if my posts sound mean-spirited to you. I really wanted to provide actual facts, because as you say, I just blurted it out like it was common knowledge . Couldn't find shit though, because apparently it isn't a subject that's talked about a lot on the internet

 

[krizzx]

None of those sites cite a source. They're only going by what you are, what they can see with the naked eye. I doubt any of those sites ripped the respective models and counted the polygons in a wireframe view.

Please post technical evidence/polycounts, not just reviews from website we've never heard of that happen to agree with you.

I will ask you to do the same. I've already provided more evidence towards my point than you have so far. Regardless of whether not you consider it "enough, or choose to believe, you have provided none to contrary.

You ask for that which does not exist to my knowledge. I have given you what is available and it is all in agreement.

I will take the opinion of the dozens of people and professionals who analyzed it and wrote their conclusions. I'm not going to call a dozen different individual analysis which are all in agreement wrong without anything to refute their claim.

Every analysis of this I have ever seen has always said the exact same thing. You are the first people I have ever encountered to say otherwise, much more deny that Mario 64 DS has a higher polygon count than the original N64 version. I thought the DS having the ability to push more polygons was well known common knowledge by this date.



http://img1.livegen.fr/00/00/00/67/0000006762.jpg(You can actually see the geometry in his hat and hair in this one and count the polygons).
http://purenintendo.com/wp-content/uploads/2008/12/mario64-300x166.jpg
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094036738.jpg(The bridge to peaches castle clearly has more geometry as well as the archway above it)
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094038097.jpg(The rocks have a few more side but its hard to notice)
http://www.abload.de/img/super-mario-64-face-o4zobg.jpg(The goombas are much more detailed)
http://upload.wikimedia.org/wikipedia/en/d/d1/Super_Mario_64_DS-Graphics_comparison.jpg(do I even need to point it out?)


Are you going to tell me that Link in the Ocarina of Time 3D has less polygons than the original next?

 

[LeleSocho]

The one on the DS. It clearly has more Geometry.

You are the first people I have ever encountered to say otherwise, much more deny that Mario 64 DS has a higher polygon count than the original N64 version.

Back in the day people were bad at 3d modeling

Seriously i always try when people come up with the stupid as hell "Zelda tech demo look worse than the actual game" argument to make them see that the actual tech demos look technically better than the released game but people really don't understand.

 

[krizzx]

Back in the day people were bad at 3d modeling

Seriously i always try when people come up with the stupid as hell "Zelda tech demo look worse than the actual game" argument to make them see that the actual tech demos look technically better than the released game but people really don't understand.

We're not disucssing what "looks better". That is entirely a matter of opinion. We are discussing which is more detailed ie. has more geometry/polygons. That is a matter of fact.

The number of polygons used to create a 3D modal are static and not subject to your preference. I think links sprite modal from Zelda 3 "looks better" than his polygon modal from Ocarina of Time and that the original brown haired link looks better than the blond one. I also think Link from Wind Waker looks better than Link from Twilight Princess. That is not a matter of technical capability. That does make Wind Waker's Link more detailed. Ocarina of Time is still clearly technically superior to Zelda 3 regardless of how I think it looks.

 

[lwilliams3]

I will ask you to do the same. I've already provided more evidence towards my point than you have so far. Regardless of whether not you consider it "enough, or choose to believe, you have provided none to contrary.

You ask for that which does not exist to my knowledge. I have given you what is available and it is all in agreement.

I will take the opinion of the dozens of people and professionals who analyzed it and wrote their conclusions. I'm not going to call a dozen different individual analysis which are all in agreement wrong without anything to refute their claim.

Every analysis of this I have ever seen has always said the exact same thing. You are the first people I have ever encountered to say otherwise, much more deny that Mario 64 DS has a higher polygon count than the original N64 version. I thought the DS having the ability to push more polygons was well known common knowledge by this date.



http://img1.livegen.fr/00/00/00/67/0000006762.jpg(Seriously, you can see the geomoetry in his hat and hair in this one and actually count the polygons).
http://purenintendo.com/wp-content/uploads/2008/12/mario64-300x166.jpg
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094036738.jpg
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094038097.jpg
http://www.abload.de/img/super-mario-64-face-o4zobg.jpg
http://upload.wikimedia.org/wikipedi...comparison.jpg
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094036738.jpg

Are you going to tell me that Link in the Ocarina of Time 3D has less polygons than the original next?

I don't have any info about the polygon counts, but in the case of Mario 64, the original version came out when developers were still getting use to making 3d character models. It is very possible for a better looking model to use up less polygons than another, especially when you add better textures.

Anyway, I'm interested in learning the actual polygon count of the DS model. The original N64 one was around 750 polygons.

 

[krizzx]

I don't have any info about the polygon counts, but in the case of Mario 64, the original version came out when developers were still getting use to making 3d character models. It is very possible for a better looking model to use up less polygons than another, especially when you add better textures.

Anyway, I'm interested in learning the actual polygon count of the DS model. The original N64 one was around 750 polygons.

I editied the post. I pointed out the differences in detail specifically for those who have trouble spotting them.

The N64 version is the better looking modal specifically because it uses texture filtering. Polygon detail goes straight to the DS version, however.

 

[lostinblue]

http://en.wikipedia.org/wiki/Super_Mario_64_DS
http://zonenintendo.weebly.com/super-mario-64-ds.html
http://inintendo.net/Articles/FarewellDSp1.php
http://www.gameyum.com/other-mario-games/63197-super-mario-64-ds-nintendo-ds-review/
http://www.nintendocosmos.com/super_mario_64_ds-info.htm

Like I said, you shouldn't believe everything you read out and about.

On the Mario 64 issue, I always knew it had less polygons from the moment I've seen it running. I use it as an example all the time too, as to how less polygons can look better.

Better than 1996 modeling, that is.


So here we go, this is the ripped Mario 64 model:

752 polygons. Whoa. That's more than child link from OoT (standing at 662 polygons, and notice how it looks so much better).

This was before polygon transformation/animation was common, and it was hardware intensive to do (Banjo Kazooie did so though, so banjo has 405 polygons) so instead developers opted to keep it to a minimum and sort it out by assembling models via joining "floating parts" as you can notice to be the case on Mario 64's case above.

So this is the first part of your refutal, next we have Mario 128, why? You'll see:

Video here.

How many polygons do you think those have? *drumroll*

The demo basically opened up a flat 3D plane at which time Miyamoto began pouring dozens upon dozens of 700-poly Mario models, ultimately totaling 128 moving plumbers on-screen at once.

Source: http://cube.ign.com/articles/084/084030p1.html

They have less polygons than the original model for Mario 64 did!

See were I'm going?

Do you still think this:

Has more polygons than those GC models?

It's simply using the same more modern techniques, legs are paralelipiped blocks that animate on the joint for instance. It's 300/400 polygons at most, perhaps less. Also, there's no reliable DS model ripper (to my knowledge), but most DS 3D games won't ever think of using 700 polygons for a model, if anything because DS hardware is capped to rendering 2048 polygons per frame!

Note: I found the Mario 64 model rip on my external HDD in a matter of minutes, but took this long because 3 people called me (and talked for a long time).



EDIT: As for Link from Spaceworld demo versus Link from Zelda TP... Link's model from TP has 6900 polygons which isn't a lot; The Spaceworld link seems way more detailed than that:

We're talking as detailed as the model for a 128-bit high-end fighting game (with two character onscreen being the focus), it's just the texture un-skinned look or flat texturing going on that makes it appear more dated (ie: first generation modeling).


It's clearly way more detailed all around, including the face going by teeth/jaw outline detail alone... than this:

Remember, less polygons doesn't necessarily mean worse, it just means less polygonal detail.

 

[phosphor112]

But how much bandwidth does the Wii U actually have? It has four "main" memory pools and several dedicated local pools/ caches (Latte blocks D and X for example) - we only know the bandwidth of one of those.

Does anyone know or even have a guess? And it has such an odd amount... like 38 or 40 or something. 3 pools at different speeds. I don't understand it.

 

[Schnozberry]

Does anyone know or even have a guess? And it has such an odd amount... like 38 or 40 or something. 3 pools at different speeds. I don't understand it.

I don't think anyone could do anything but guess at it. We'd need someone with extensive experience on the platform to leak it, because Nintendo will remain mum forever. I think the argument will die anyway at E3 when we see games built from the ground up on the console with decent budgets. That will be the answer, one way or another, on the console's power relative to current and future consoles.

 

[freezamite]

Using spherical harmonics to calculate static light maps has been done all the time, like in Halo 3.

Spherical harmonic transfer functions are really good at representing the effects of infinite-distance environment lights. I've heard of people BSing spherical harmonic shadowing by representing a point light as a web of sample points with different spherical harmonic representations. If they're doing spherical harmonic realtime GI from non-infinite lights as you claim, this would get really sketchy, as you break your assumptions about the angles and magnitudes at which light is striking other objects in the scene. If anything, this seems like a very expensive way to get very flawed GI, and I'd be surprised if it's what they're doing.

Well, my point wasn't that ZombiU is the first game to use spherical harmonics, but that the fact that it uses this mathematical aproximation is a proof that the Radiosity effect is not pre-backed like on Mirror's Edge.

In fact, spherical harmonics is what was used in Battlefield 3 to achieve this type of lighting.
About what is used in ZombiU, we have evidences of this thanks to the profile of a programmer that worked on the game.
He specifically states that worked on the spherical harmonics approximation and also radiosity normal maps, so this is what is used in ZombiU AT LEAST.

 

[LeleSocho]

We're not disucssing what "looks better". That is entirely a matter of opinion. We are discussing which is more detailed ie. has more geometry/polygons. That is a matter of fact.

The number of polygons used to create a 3D modal are static and not subject to your preference. I think links sprite modal from Zelda 3 "looks better" than his polygon modal from Ocarina of Time and that the original brown haired link looks better than the blond one. I also think Link from Wind Waker looks better than Link from Twilight Princess. That is not a matter of technical capability. That does make Wind Waker's Link more detailed. Ocarina of Time is still clearly technically superior to Zelda 3 regardless of how I think it looks.

Read again mine and your post because i think that you understood the exact opposite of what i said
back then people were really bad at 3d modeling (it was the '90s in fact) so even with tons and tons of polygons the final result was really bad

The N64 uses way more polygons than the ds version just like the zelda demos are technically way better than the released game, do they look worse because of the art style? of course! Are they technically worse? ahahahahah no.
You are victim of what you said in your post, it looks more visually appealing? well then it should be more advanced technically.

It shouldn't even be that hard with the SM64 example, i think the emulators for DS and N64 have the "show wireframe" options... active it, count the poly yourself then return here with the results.

 

[Captainsweatpants]

Does anyone know or even have a guess? And it has such an odd amount... like 38 or 40 or something. 3 pools at different speeds. I don't understand it.

Didn't somebody on Gaf have a Devkit or something? I'm pretty sure that some competent Gaffers could code a Benchmark and test it if they put their heads together.

 

[HTupolev]

Well, my point wasn't that ZombiU is the first game to use spherical harmonics, but that the fact that it uses this mathematical aproximation is a proof that the Radiosity effect is not pre-backed like on Mirror's Edge.

What proof? You literally just said "it uses this realtime GI approximation that HTupolev just described therefore it has realtime GI." That's completely circular.

About what is used in ZombiU, we have evidences of this thanks to the profile of a programmer that worked on the game.
He specifically states that worked on the spherical harmonics approximation and also radiosity normal maps, so this is what is used in ZombiU AT LEAST.

Link? Based on what you just said, I don't know how you know that he wasn't talking about baking.

 

[Fourth Storm]

Does anyone know or even have a guess? And it has such an odd amount... like 38 or 40 or something. 3 pools at different speeds. I don't understand it.

My best guess at the 32 MB eDRAM bandwidth is 70.4 GB/s. That's at 550 Mhz on a 1024-bit bus. Actually, I'd call it more than a guess, because looking at the arrangement of the interface and counting the connectors seems to confirm.

The 2 MB eDRAM and 1 MB SRAM should be in line w/ their Wii counterparts. Marcan claims these are currently off limits to devs though (and I have no reason to not believe him), so it's kind of a moot point. It sure would be interesting if they could find a way to take advantage of those pools for Wii U mode in the future.

The topic of the caches is quite interesting, especially given the comments on how memory was a major focus in this machine. I find it disconcerting that we cannot positively identify L1 and L2 texture caches yet, as they could go a long way in mitigating the modest speed of the DDR3 when it comes to feeding the TMUs. Block D is a good guess for the L2 cache, but then where is the L1? Did Nintendo streamline the pipeline so that it's not really necessary? Seems unlikely. I have an alternative theory that Block D actually houses the Instruction Cache and Constant Cache. These all need to be accounted for somehow.

 

[krizzx]

Like I said, you shouldn't believe everything you read out and about.

On the Mario 64 issue, I always knew it had less polygons from the moment I've seen it running. I use it as an example all the time too, as to how less polygons can look better.

Better than 1996 modeling, that is.


So here we go, this is the ripped Mario 64 model:

http://i.imgur.com/tCuFbT6.jpg

752 polygons. Whoa. That's more than child link from OoT (standing at 662 polygons, and notice how it looks so much better).

This was before polygon transformation/animation was common, and it was hardware intensive to do (Banjo Kazooie did so though, so banjo has 405 polygons) so instead developers opted to keep it to a minimum and sort it out by assembling models via joining "floating parts" as you can notice to be the case on Mario 64's case above.

So this is the first part of your refutal, next we have Mario 128, why? You'll see:

http://i.imgur.com/3eIrNFl.jpg

http://i.imgur.com/owPl9Qz.jpg

Video here.

How many polygons do you think those have? *drumroll*

Source: http://cube.ign.com/articles/084/084030p1.html

They have less polygons than the original model for Mario 64 did!

See were I'm going?

Do you still think this:

http://i.imgur.com/Xp82wGy.jpg


Has more polygons than those GC models?

http://i.imgur.com/hPnBBnE.jpg

It's simply using the same more modern techniques, legs are paralelipiped blocks that animate on the joint for instance. It's 300/400 polygons at most, perhaps less. Also, there's no reliable DS model ripper (to my knowledge), but most DS 3D games won't ever think of using 700 polygons for a model, if anything because DS hardware is capped to rendering 2048 polygons per frame!

Note: I found the Mario 64 model rip on my external HDD in a matter of minutes, but took this long because 3 people called me (and talked for a long time).



EDIT: As for Link from Spaceworld demo versus Link from Zelda TP... Link's model from TP has 6900 polygons which isn't a lot; The Spaceworld link seems way more detailed than that:

http://www.1amgeek.com/blog/wp-content/uploads/2011/06/FileZelda128.jpeg

We're talking as detailed as the model for a 128-bit high-end fighting game (with two character onscreen being the focus), it's just the texture un-skinned look or flat texturing going on that makes it appear more dated (ie: first generation modeling).


It's clearly way more detailed all around, including the face going by teeth/jaw outline detail alone... than this:

http://assets2.ignimgs.com/2005/05/...t-princess-20050517032706999-1123106_640w.jpg
Remember, less polygons doesn't necessarily mean worse, it just means less polygonal detail.

That's a very nice analysis but I never questioned Mario 128's Mario polygon count. Even I can tell that they not more detailed. I always thought they were copied and pasted N64 modals with a few tweeks. Now I know they are actually less than that.

Your Zelda analysis is completely fallacious. You didn't actually state the amount of polygons in the tech demo's link. You just stated the amount in Twlight Princesses Link vs "looks more detailed". You used the same fallacy with your Mario comparison. You didn't in any way prove that Mario 64's Mario has a higher polygon count. "Do you still think this: Has more polygons than those GC models?" You shifted from comparing the model to the N64 to comparing it to the better textured GC model. You tried to run the same point you were attempting to contradict earlier.

In case you missed it, I pointed out the increased polygons.

http://img1.livegen.fr/00/00/00/67/0000006762.jpg(You can actually see the geometry in his hat and hair in this one and count the polygons).
http://purenintendo.com/wp-content/uploads/2008/12/mario64-300x166.jpg
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094036738.jpg(The bridge to peaches castle clearly has more geometry as well as the archway above it. You can see the vertices clearly and count them by hand. There is a peak at the archway now. There is exactly 1 more side to the archway and the bridge rail which was flat before, now arcs up and comes back down.)
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094038097.jpg(The rocks have a few more side but its hard to notice)
http://www.abload.de/img/super-mario-64-face-o4zobg.jpg(The goombas are much more rounded and less edgy)
http://upload.wikimedia.org/wikipedia/en/d/d1/Super_Mario_64_DS-Graphics_comparison.jpg(do I even need to point it out?)

The textures are the only thing better in the N64 version and all of sources I posted were in agreement on that as well. The 64 version lost the texture filtering but had an increase in polygon detail. So I will stick to what all of the old analyst confirmed and what I see with my own eyes on this matter.

 

[lostinblue]

Read again mine and your post because i think that you understood the exact opposite of what i said
back then people were really bad at 3d modeling (it was the '90s in fact) so even with tons and tons of polygons the final result was really bad

That's a bold statement. Modeling tools sucked ass.

And manipulating polygon transformations for animations took a lot CPU power, it was simpler for the hardware to use and manipulate different over-imposed-each-other geometry parts. It often was cheaper to do things that way, later games like Zelda OoT used a mix, arms, legs and neck would transform polygons whilst animating joints (and be un-textured) while some other parts of the model were composed of the regular "separate parts" deal.

It shouldn't even be that hard with the SM64 example, i think the emulators for DS and N64 have the "show wireframe" options... active it, count the poly yourself then return here with the results.

It's not easy to count polygons/faces that way. That said I dunno never used a DS emulator.

I know 3D model ripping tools seem to be behind schedule (seeing it's now an almost defunct platform and still no dice).

That's a very nice analysis but I never questioned Mario 128's Mario polygon count. Even I can tell that they not more detailed. I always thought they were copied and pasted N64 modals with a few tweeks. Now I know they are actually less than that.

In case you missed it, I pointed out the increased polygons.

http://img1.livegen.fr/00/00/00/67/0000006762.jpg(You can actually see the geometry in his hat and hair in this one and count the polygons).
http://purenintendo.com/wp-content/uploads/2008/12/mario64-300x166.jpg
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094036738.jpg(The bridge to peaches castle clearly has more geometry as well as the archway above it)
http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094038097.jpg(The rocks have a few more side but its hard to notice)
http://www.abload.de/img/super-mario-64-face-o4zobg.jpg(The goombas are much more detailed)
http://upload.wikimedia.org/wikipedia/en/d/d1/Super_Mario_64_DS-Graphics_comparison.jpg(do I even need to point it out?)

Your Zelda analysis is completely fallacious. You didn't actually state the amount of polygons in the tech demo's link. You just stated the amount in Twlight Princesses link vs "looks more detailed". You used the same fallacy with your Mario comparison. You didn't in any way prove that Mario 64's Mario has a higher polygon count. You just stated.

Huh?

Mario 64 DS Mario model has less polygons going for it than Mario 64 on N64, give it up it's more detailed YES, it's not 700 polygons, NO; which was your original point and I see you're making it broader topic now (while still insisting on the Mario model, which is what amazes me!), but I won't bite. You saved 300/400 polygons on the character model and wasted a few dozen on an archway and bridge... so what. It would be more clever of you to suggest that the multiplayer mode with 4 players might surpass the total polygons on-screen that Mario 64 had (perhaps), and the coins being tridimentional and animated this time around brings extra stress to the hardware (for sure).

And Zelda Spaceworld Link is more polygon detailed than Zelda TP link's model.

On Mario 64 DS case I really outlined it and proven it to you that it had less polygon detail going on the character which you didn't know, and on Zelda Spaceworld case I looked for a good image of the character model and asked you to look and see, since you said "Analyzing geometry details one of my specialties".™

Apparently not (on both accounts).

And do you really expect me to know the exact polycount for a game I can't rip the models out of (Mario 64 DS) and pull the same shenanigans for an unreleased tech demo? (Zelda Spaceworld)
Newsflash: you didn't have polycount numbers for anything, at least I am constructing a line of thought here.


Also there's nothing wrong with being wrong; but please learn to take it as a man. Jesus I'm out of here.

 

[krizzx]

That's a bold statement. Modeling tools sucked ass.

And manipulating polygon transformations for animations took a lot CPU power, it was simpler for the hardware to use and manipulate different over-imposed-each-other geometry parts. It often was cheaper to do things that way, later games like Zelda OoT used a mix, arms, legs and neck would transform polygons whilst animating joints (and be un-textured) while some other parts of the model were composed of the regular "separate parts" deal.It's not easy to count polygons/faces that way. That said I dunno never used a DS emulator.

I know 3D model ripping tools seem to be behind schedule (seeing it's now an almost defunct platform and still no dice).Huh?

Mario 64 DS Mario model has less polygons going for it than Mario 64 on N64, give it up it's more detailed YES, it's not 700 polygons, NO; and Zelda Spaceworld Link is more polygon detailed than Zelda TP link's model. On the first case I really outlined it and proven it to you, on the second I asked you to look and see, since you said "Analyzing geometry details one of my specialties".

Apparently not.


And do you really expect me to know the exact polycount for a game I can't rip the models out of (Mario 64) and pull the same shenanigans for an unreleased tech demo? (Zelda Spaceworld) Newsflash: you didn't have polycount numbers for anything, at least I am constructing a line of thought here.

Also there's nothing wrong with being wrong; but please learn to take it as a man. Jesus I'm out of here.

That was a really dismissive respone. I never said anything more than that Link's "face" was more detailed in Twilight Princess. I said I couldn't gauge the rest of the body and I still say the face is more detailed.

Also, once again, I pointed out the actual "geometrical" increases in detail. Not textures. Every source confirms that Mario 64 has more polygon detail if nothing else.

This is a clear increase in "POLYGON" detail.
http://upload.wikimedia.org/wikipedia/en/d/d1/Super_Mario_64_DS-Graphics_comparison.jpg
Bowser actually has fingers in the DS version. His horns curve 2 times as opposed to one. His shell is actually separate from his back and has multiple curves on it where it is just a flat texture in the N64. He has many more spikes on it(are you going to tell me the spikes aren't polygons?). His arms, and legs are much larger, rounder.His face and his hair have many more points is overall more round.

http://img1.livegen.fr/00/00/00/67/0000006762.jpg
His hair has 4 tails as opposed to 2. His hat has 10 sides as opposed to the 6 in the N64 version. His gloves tail sticks outward now where it was just drawn on before. You can also see that the top of the archway is angular in the DS version where it is completely flat and straight in the N64 version like I pointed out in this picture.

http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094036738.jpg
You can also see that on the N64 version, the bridge side is "straight" flat. It arcs up, over and back down in the DS version. If that is not the result of more polygons then I don't know what is unless you're going to tell me the DS was using steep parallax mapping or tessellation to create pronounced detail.

As I said, you can manually count the polygon detail increases. Its plainly visible to the eye. It is no miniscule increase. You are refusing to even look at it or acknowledge it, and then you are calling yourself right and me wrong despite all evidence to the contrary.

 

[freezamite]

What proof? You literally just said "it uses this realtime GI approximation that HTupolev just described therefore it has realtime GI." That's completely circular.


Link? Based on what you just said, I don't know how you know that he wasn't talking about baking.

Search for it on google, first page. And the whole thing we were discussing, was if that radiosity effect was totally pre-backed like on Mirror's edge or something rendered in real time.

For example, the nursery is full of lights with varying values which couldn't be possible if the lighting was fully pre-backed.
You only have to watch at the nursery playthrough to confirm it.

 

[lostinblue]

That was a really sore repsonse.

Sore for being right? no, it's just that I was expecting a different reply than that nonsense argumentation and I have more things to do.

I never said anything more than that Link's "face" was more detailed in the Twlilight Princess. I said I couldn't guage the rest of the body and I still say the face is more detailed.

Both face and body have more polygons going in there.

They don't look better/more polished though, but it appears rounder and more polygon crevassed all around. Like several people tried to point out before too.

Also, once again, I pointed out the actual "geometrical" increases in detail. Not textures. Every source confirms that Mario 64 has more polygon detail if nothing else.

Not Wikipedia links again.

And I chimed in when we were comparing Mario 64 3D models and you were saying it had more polygons, with the same sources you're using now.

I did exactly what I set out to prove.

This is a clear increase in "POLYGON" detail.
http://upload.wikimedia.org/wikipedia/en/d/d1/Super_Mario_64_DS-Graphics_comparison.jpg
Bowser actually has fingers in the DS version. His horns curve 2 times as opposed to one. His shell is actually separate from his back and has multiple curves on it where it is just a flat texture in the N64. He has many more spikes on it(are you going to tell me the spikes aren't polygons?). His arms, and legs are much larger, rounder.His face and his hair have many more points is overall more round.

http://img1.livegen.fr/00/00/00/67/0000006762.jpg
His hair has 4 tails as opposed to 2. His hat has 10 sides as opposed to the 6 in the N64 version. His gloves tail sticks outward now where it was just drawn on before. You can also see that the top of the archway is angular in the DS version where it is completely flat and straight in the N64 version like I pointed out in this picture.

http://insidermedia.ign.com/insider...super-mario-64-face-off-20041124094036738.jpg
You can also see that on the N64 version, the bridge side is "straight" flat. It arcs up, over and back down in the DS version. If that is not the result of more polygons then I don't know what is unless you'r egoing to tell me the DS was using parallax mapping to pronounced detail.

I'm not going to tell you anything stupid (parts underlined).

You're scrapping for change here, it looks better, and perhaps certain models that suffered before can now be a little more detailed; me and others pointed out how inefficient modeling for this era was (and showed it, I insist), you must have also realized that by now. Bowser was no different (separated parts instead of animating joints), so it's not even certain to be using more polygons, but I pointed out a 300/400 polygon saving by using less polygons; that's huge by N64/DS standards as DS couldn't render the original Mario 64 model alongside 3 times without hitting it's 2048 polygon threshold. That means whatever you're looking at and is a character model was changed and optimized, and yes the savings could be used elsewhere, but total geometry though probably doesn't extend past the N64 level of polygonal detail; even because N64 unlike DS, wasn't agressively capped to a limit of 2048 polygons per second.

Being capped to 2048 polygons per frame though, and taking into account it ran at 30 frames per second on the N64 that would also make it a 61.440 polygon game. It's too low for N64 standards; most game didn't surpass the 100.000 mark, let alone reach the 150.000 advertised polygons, but 60.000 is too low.

Like I said you'd be better pointing out 4 character multiplayer and the 3D coins, than freaking low polygon archways having a spanking 4 polygon increase going for them.

As I said, you can manually count the polygon detail increases. Its plainly visible to the eye. It is no miniscule increase. You are refusing to even look at it or acknowledge it, and then you are calling yourself right and me wrong despite all evidence to the contrary.

Then count them yourself, you can manually count them yourself.

EDIT: I thought you were telling me to go count wireframes on this last quote so I'm backtracking on the tone I used in this last sentence. I still feel I addressed it in the rest of the post though.

 

[GhostTrick]

As Lostinblue said, more detailled doesn't means more polygons.

Would be cool to put an end to this with a 3D model ripper. Can't someone use it with DS emulator ?

 

[Popstar]

The Nintendo DS has a higher polygon throughput than than N64.

They differ in that the DS does not have a z-buffer, and has a hard limit in the number of vertices you're allowed per-scene since they all have to fit in the queue for the span buffer renderer. The DS can render quads and triangle strips, so it does not necessarily require 3 vertices to make up each triangle.

You can't halve your framerate to double your polygon count on the DS, whereas on the N64 you could keep drawing as many polygons as you want into a scene as long as you didn't care about your framerate dropping. This could allow you to make more detailed scenes if you were willing to take the hit.

But going by polygons/second the Nintendo DS is the more powerful device.

 

[phosphor112]

Search for it on google, first page. And the whole thing we were discussing, was if that radiosity effect was totally pre-backed like on Mirror's edge or something rendered in real time.

For example, the nursery is full of lights with varying values which couldn't be possible if the lighting was fully pre-backed.
You only have to watch at the nursery playthrough to confirm it.

Are you sure you know what radiosity is? I'm assuming you're talking about the flickering lights that don't just "on and off" but instead... have "varying values" between on and off. Like the bathroom? Pretty sure that's just a dynamic light that changes intensity. There is ZERO indication of real time radiosity.

 

[freezamite]

Are you sure you know what radiosity is? I'm assuming you're talking about the flickering lights that don't just "on and off" but instead... have "varying values" between on and off. Pretty sure that's just a dynamic light that changes intensity.

Of course it is a dynamic light changing intensity. The fact is that the illumination casted by that light has a radiosity effect.
It's like the one I posted on the picture before, but with a dynamic value that changes constantly.

Look closely at how the varying intensity of the light affects at the whole color of the scene, instead of just making the lighting halo smaller.

 

[lostinblue]

Would be cool to put an end to this with a 3D model ripper. Can't someone use it with DS emulator ?

No dice, there are no 3D model ripping tools for DS that I know of. This is the closer it gets for Mario 64 DS.

The Nintendo DS has a higher polygon throughput than than N64.

It does in realworld (ie more polygons per second) but with a few conditions.

For starters, every N64 game minus F-Zero X ran at 30 frames per second; that made it so that it's not 100.000 polygons divided by 60 frames but 100.000 at a 30 frames per second rate which, at translates to 3300 polygons per frame being sustainable, that's more than DS can render (2048 polygon cap). Most N64 games dared to go as low as 20 frames per second, so we're talking about surpassing the 3300 polygon per frame number (with good microcode at least). Mario 64 was a launch title though, but it being a 61.440 polygon game (2048 polygons times 30 frames) seems amiss, I mean certainly not consistently so.

Most DS games run at 60 frames for that reason, it's not that they couldn't do more per frame it's just that they can't (but you know this, I'm recapping it). But by doing that at 60 frames per second they're doing more than the N64 ever could... at that rate; in regards to well engineered N64 games though, it's rarely pushing more polygons per frame than N64, making late ports difficult (seeing 3D models started being more optimized towards the end so they couldn't make them look better and still spare some polygons).

They differ in that the DS does not have a z-buffer, and has a hard limit in the number of vertices you're allowed per-scene since they all have to fit in the queue for the span buffer renderer. The DS can render quads and triangle strips, so it does not necessarily require 3 vertices to make up each triangle.

I reckon no one uses quads, not on Saturn days (well, they had to on Saturn) and not since.

Good to have on a DS though, not disagreeing on that, but I doubt it got used much if at all.

 

[Popstar]

(2048 polygon cap).

This isn't true and mentioned in the post you're quoting.

I reckon no one uses quads, not on Saturn days (well, they had to on Saturn) and not since.

You're wrong. Lot's of 3D modelling packages (Maya for instance) produce quads as a matter of course and it's easy enough to export them. People used quads on the NDS.

 

[Duckhuntdog]

To those discussing the lighting in ZombieU, if this is of any help, the lighting system used in the game isn't something made by UBI, instead it uses a middleware app called Beast, from autodesk.

Since this is part of Gameware, Beast is free to developers as part of the agreement between Autodesk and Nintendo.

That said, Beast is an very amazing piece of software.

 

[Shin Johnpv]

Quads are the norm in high end software rendered stuff.

 

[joesiv]

Which one of these two Mario models do you honestly think has more polygons?

Well if the hands are any indication of the rest of the models polygon density, I would say the N64 version has a clear advantage. Generally you can judge the density by how sharp the angles are when trying to approximate rounded surfaces. The fist on the DS version appears much more angular, probably more of like 4 or 5 quads (8-10 triangles), wheras the N64 one seems to have more. This can also be seen in his wrist, where the N64 version appears to have 5 or more quads to do the ring around the wrist, wheras the DS one looks like it is doing it with maybe 4 quads. The nose also looks less sharp on the N64. hard to tell given the difference in poses, I'm sure we can find better examples no?

Phong shading sure does a good job hiding poly counts on everything outside of edges eh? lol...

If the N64 version if ~750, I would guestimate the DS version to be around 4-500...

 

[joesiv]

hard to tell given the difference in poses, I'm sure we can find better examples no?

how about these ones?

This one is interesting, which flips the whole matter on it's head IMO:

Seems like the DS has more poly's for marios head (hair, maybe hat), but the world geo seems reduced (castle).

 

[lostinblue]

This isn't true and I just told you that.

Oh it is.

I'll put the MechAssault dudes on:

What happens when you go over the 2000 poly/frame limit?

BBE: Basically, when you go over the limit, polys stop drawing and you won't always know which ones those are. The real kicker is that if 1 poly is over limit, it takes out any polys that share vertices with it too like a bad chain reaction. There's always the possibility of introducing bugs when you're straining the hardware too much so it's just best to keep an eye on the poly count as you go through levels to ensure you're always under your limits.

Source: http://www.insertcoin-pt.net/index.php?showtopic=5471#entry126868

I rest my case.

You're wrong. Lot's of 3D modelling packages (Maya for instance) produce quads as a matter of course and it's easy enough to export them. People used quads on the NDS.

I know they can use and export in quads.

I just don't know if that's a feature that actually gets used. Say, like NURBS.

Seems like the DS has more poly's for marios head (hair, maybe hat), but the world geo seems reduced (castle).

Overall and not going through details, Mario 64 DS model has less polygons than the Mario 64 one on N64, not what you're saying I think; but it has less polygons all around, no way it's even going close to the 752 of the N64 version, meaning it's never a "model is more polygon detailed, it's the game world that isn't" or something like that.

As for that castle area, it might have less polygons, but if it does it's for a reason. I reckon the multiplayer only takes place in certain areas, castle area was one of them; and being able to take 2 other characters as detailed as the main one in that space is bound to take a toll on the DS easily reaching half the console's throughput; simplifying world geometry would be the way to do it, and they had to do it for the 2048 polygon cap.

They didn't mess with most of the game's geometry (would defeat the purpose), castle area is the one with most changes if I recall correctly; probably due to that.

 

[Popstar]

Oh it is.

I'll put the MechAssault dudes on:

Source: http://www.insertcoin-pt.net/index.php?showtopic=5471#entry126868

I rest my case.I know they can use and export in quads.

I just don't know if that's a feature that actually gets used. Say, like NURBS.

I'm not speculating with you. I'm telling you.

I've done Nintendo DS development.

 

[lord pie]

nursery is full of lights with varying values which couldn't be possible if the lighting was fully pre-backed.
You only have to watch at the nursery playthrough to confirm it.

To my eye they are using dynamic lights added on top of baked lighting. Pretty standard stuff.

 

[HTupolev]

Search for it on google, first page.

That's not how technical discussions work. Give me a source, then we can talk.

And the whole thing we were discussing, was if that radiosity effect was totally pre-backed like on Mirror's edge or something rendered in real time.

For example, the nursery is full of lights with varying values which couldn't be possible if the lighting was fully pre-backed.
You only have to watch at the nursery playthrough to confirm it.

I've just watched a playthrough of the nursery, and I'm not convinced that that's true. There are dynamic lights, but I'm not seeing any obvious light bounces. Even if they were experiencing radiosity, lighting is linear, so it's possible to bake bounce patterns for pre-computed lights separate from your main light map. For instance, Halo 3 flashes its GI-computed lights at the end of the level Crow's Nest. It does it with pretty much the entire light map, so almost everything on-screen visibly dims, but you could easily engineer a system that stores multiple light maps and flashes them independantly when desired based on the behaviour of their associated "lights". Actually, the portal-based engines used in some mid-90's games also allow for basic hand-made animated "global illumination" using built-in flicker patterns and the ability to modify light levels on particular surfaces and simple volumes (not that this is useful for modern engine development, but it's a cool historical footnote in computer graphics).

What's really interesting and a hallmark of what would be considered real-time GI is when lights whose location can change in a non-preplanned way experience radiosity... and honestly, some lights placed by the player in ZombiU are blatantly obviously not experiencing any GI at all. For instance, look at when the pad is pulled out at around 22:20. That's an extremely vibrant blue dynamic light, and it seems very clear that the only illumination from it is direct. It's casting shadows, but a light that bright *should* absolutely be casting bounced light all over the room, especially since it's a player-controlled light and therefore worth devoting real-time GI resources toward.

 

[Duckhuntdog]

That's not how technical discussions work. Give me a source, then we can talk.


I've just watched a playthrough of the nursery, and I'm not convinced that that's true. There are dynamic lights, but I'm not seeing any obvious light bounces. Even if they were experiencing radiosity, lighting is linear, so it's possible to bake bounce patterns for pre-computed lights separate from your main light map. For instance, Halo 3 flashes its GI-computed lights at the end of the level Crow's Nest. It does it with pretty much the entire light map, so almost everything on-screen visibly dims, but you could easily engineer a system that stores multiple light maps and flashes them independantly when desired based on the behaviour of their associated "lights". Actually, the portal-based engines used in some mid-90's games also allow for basic hand-made animated "global illumination" using built-in flicker patterns and the ability to modify light levels on particular surfaces and simple volumes (not that this is useful for modern engine development, but it's a cool historical footnote in computer graphics).

What's really interesting and a hallmark of what would be considered real-time GI is when lights whose location can change in a non-preplanned way experience radiosity... and honestly, some lights placed by the player in ZombiU are blatantly obviously not experiencing any GI at all. For instance, look at when the pad is pulled out at around 22:20. That's an extremely vibrant blue dynamic light, and it seems very clear that the only illumination from it is direct. It's casting shadows, but a light that bright *should* absolutely be casting bounced light all over the room, especially since it's a player-controlled light and therefore worth devoting real-time GI resources toward.

The lighting in ZombieU was done with Beast.

 

[UltraMagnanimous]

#TeamLostinblue

 

[Fourth Storm]

So you tech experts look at this. On this site their resident tech gurus seem to have solve the Ram bandwidth issue. Go check out the last 5 pages or so and see what you think... Are they crazy?

http://thewiiu.com/topic/6973-wii-us-ram-is-slower-than-ps3xbox-360/page-9#entry189991

They're so off base it's not even funny.

 

[phosphor112]

Of course it is a dynamic light changing intensity. The fact is that the illumination casted by that light has a radiosity effect.
It's like the one I posted on the picture before, but with a dynamic value that changes constantly.

Look closely at how the varying intensity of the light affects at the whole color of the scene, instead of just making the lighting halo smaller.

I don't see that at all. All I see is it changing how much in the room is being lit.

Easiest way to tell if there is real time radiosity is to use the UV scanner. It doesn't do anything radiosity related at all.

 

[lostinblue]

I'm not speculating with you. I'm telling you.

I've done Nintendo DS development.

What happens if you exceed the 2048 polygon cap then? Nothing?

Because there was a cap (I reckon you've called it "hard limit"), and fact is we didn't see people running DS into the ground framerate-wise as we've seen on the N64 for a reason. I understood you said as much in your first post (so I didn't get what you meant when you said you said it wasn't the case later on).

Also this:

(...) The DS can render at most about 6144 vertexes per frame (which translates to roughly 2048 triangles per scene or 1536 quads), at 60 frames per second. Rendering can occur only to a single screen at a time, so having 3D on both screens decreases performance significantly. The DS is more limited by its polygon budget than by its pixel fill rate.

Source: http://osdl.sourceforge.net/main/documentation/misc/nintendo-DS/homebrew-guide/HomebrewForDS.html


I don't realize what we're discussing anymore. Or is it the quad thing?

If it's the quad thing consider me enlightened, I didn't think most devs would bother.

 

[freezamite]

To my eye they are using dynamic lights added on top of baked lighting. Pretty standard stuff.

This could be true for the "rave" segment, where the "radiosity effect light" is fixed at a certain point, and then there are projections which don't have any radiosity on them (and I don't think that such a small light sources could have any impact on the surrounding areas at that distance anyway, so it would be a waste to use radiosity on them).
But this is not the case with the lights on the nursery. They are lights like the one I posted on the picture, with a varying intensity that obviously affects at how the whole scenario is lighted.
If the lighting was truly backed, you would have a... well, you wouldn't have a light with a varying intensity, or it would look like the projections on the rave, with the lighting behind being constant an then the projections being there, moving, on or off, but not the effect that you actually have.

I don't see that at all. All I see is it changing how much in the room is being lit.

Easiest way to tell if there is real time radiosity is to use the UV scanner. It doesn't do anything radiosity related at all.

The Scanner is a strong-direct light, it of course doesn't have a radiosity effect.
In fact, now that I've looked closely at it, the lantern also is a dynamic radiosity source.
http://www.youtube.com/watch?v=Ds4qmuCpsrQ

Look at that video. Look at when he is using the lockpick on the door, and how the light of the lantern rebounds in real time from the door to the wood of the door frame, giving it a greener at the parts the light is closer. This is what radiosity is, and is pretty obvious that ZombiU uses it dynamically. Now, I've only played the PC version of Battlefield 3, so I don't know if the PS3/360 also have dynamic radiosity lighting.
It uses radiosity, that's for sure, but I don't know if there are mobile or intensity-varying lights that affect the scenario in real time. If not, then ZombiU is the first game with an official next-gen lighting.