Digital Foundry Article Technical Discussion Archive [2011]

[kagemaru]

The last Red faction chapter is worse than the previous, talking only of a gameplay perspective but tech wise, is really remarkable the amount of physics, interactivity, buffer, both mixed on the screen... maybe the 60 fps target aren't so steady, but the tecnology behind this game imho deserve more credits even if is it 'only' 960x540p. Talking of resistance 3, honestly, really ps3 can hit only that at 960x704P ? I'm a bit surprise & counfonded. I see monochromatic colours, mediocre texture, fire effect really opinable (consider the plot fantascientic of the game); the mixed bug not appear surely the worse thing ever, probably the use of the overdraws is remarkable, but the cost for something so subtle it's too much,imho, which could be realized with more cheaper, artigianal 'alternatives' ways & to dedicate more sources in others areas more evident .

But Red Faction doesn't target 60fps, in fact it hardly ever goes above 30fps. IMO, outside of the physics, all of your complaints about R3 can also be directed to Red Faction.

Judging by your comment, your issue is that a ps3 exclusive is at this resolution with what you consider average effects, correct? It's not like the system has finite amount of resources, and I'm sure Insomniac didn't come to the resolution decision lightly.

IIRC colors don't make a difference in performance, and from what I saw, the texture quality hasn't suffered compared to R1 or R2 textures. Fire looks fine to me as well, at least compared to R2. Basically, if you're a fan of the Resistance series, R3 is very comparable to R2 in many of the aspects that you're complaining about. Also the effects that can create overdraw are hardly subtle IMO. It can mean the difference between a game looking sterile and a game looking like a battlefield. Smoke, fire, explosions, particles, etc. all create overdraw.

Besides, that's only one [logical] reason for the resolution drop. We also know they improved their lighting, so maybe that could have played a part in the resolution drop as well.

In the end we just don't know, but I think you're being just a bit too down on the game. :smile:

The quality of visuals always represent the set of tradeoffs you choose to make, coupled with artistic decisions.
I wouldn't judge the technical merits of a product based off its visuals at this point.
It's not like PS1/N64 where getting 50% more polygons allowed artists to make things look a lot better.
For example trading of polygon count for more complex shaders or better shadows or whatever, could result in better or worse looking product depending on how the art department utilizes them.

Good looking has as much or more to do with art direction than technology at this point.

I understand and agree with this, have made similar comments myself in the past.

I concur in part, but I think it's more a mixed of both than the prevalence of only one of this, at least judging the console department imho. Frankly don't like the choice on both in Resistance 3, but I just tries to motivate the reason more from the tech perspective to not go too much OT.

What I think he's trying to say is that many of these games share similar technologies today. So better art will usually contribute to a better looking game, more so than better tech. You can have the most advanced engine in the world, but your game still look poor with bad art. While there are great looking games that may not be so impressive on the tech side, but exhibit great art.

For example, while many people like to look at Uncharted 2 as a tech benchmark, I have always thought the great art is what makes that game a visual feast. Don't get me wrong, the tech is impressive as well, but again I think the art is what carried it to that top-tier status.

 

[assurdum]

But Red Faction doesn't target 60fps, in fact it hardly ever goes above 30fps. IMO, outside of the physics, all of your complaints about R3 can also be directed to Red Faction.

Judging by your comment, your issue is that a ps3 exclusive is at this resolution with what you consider average effects, correct? It's not like the system has finite amount of resources, and I'm sure Insomniac didn't come to the resolution decision lightly.

IIRC colors don't make a difference in performance, and from what I saw, the texture quality hasn't suffered compared to R1 or R2 textures. Fire looks fine to me as well, at least compared to R2. Basically, if you're a fan of the Resistance series, R3 is very comparable to R2 in many of the aspects that you're complaining about. Also the effects that can create overdraw are hardly subtle IMO. It can mean the difference between a game looking sterile and a game looking like a battlefield. Smoke, fire, explosions, particles, etc. all create overdraw.

Besides, that's only one [logical] reason for the resolution drop. We also know they improved their lighting, so maybe that could have played a part in the resolution drop as well.

In the end we just don't know, but I think you're being just a bit too down on the game. :smile:



I understand and agree with this, have made similar comments myself in the past.



What I think he's trying to say is that many of these games share similar technologies today. So better art will usually contribute to a better looking game, more so than better tech. You can have the most advanced engine in the world, but your game still look poor with bad art. While there are great looking games that may not be so impressive on the tech side, but exhibit great art.

For example, while many people like to look at Uncharted 2 as a tech benchmark, I have always thought the great art is what makes that game a visual feast. Don't get me wrong, the tech is impressive as well, but again I think the art is what carried it to that top-tier status.

I understood your thought, I have just give my opinion about the tech behind the insomniac engine. Pretty curios to see the next multi project to relieve the difference on both platform; honestly I remain to the same opinion about the graphic. Surely not something to stay away or to blame in absolut, but not reach the same level seen in the others sony's first parties imho.

 

[kagemaru]

I understood your thought, I have just give my opinion about the tech behind the insomniac engine. Pretty curios to see the next multi project to relieve the difference on both platform; honestly I remain to the same opinion about the graphic. Surely not something to stay away or to blame in absolut, but not reach the same level seen in the others sony's first parties imho.

You really shouldn't be comparing R3 to other first party games IMO. We aren't informed enough about the design decisions behind each game to really make any accurate comparisons.

Also what do you mean by "curious to see the next multi project to relieve the difference on both platform"? Would you be so bothered by the resolution if it were a multi-platform title or 360 exclusive?

 

[assurdum]

You really shouldn't be comparing R3 to other first party games IMO. We aren't informed enough about the design decisions behind each game to really make any accurate comparisons.

Also what do you mean by "curious to see the next multi project to relieve the difference on both platform"? Would you be so bothered by the resolution if it were a multi-platform title or 360 exclusive?

Why the discussion to me seem suddenly become so personal? I think my opinion is clear, I explain what appears to me, you not concur, ok not need to take this way. I don't think there is reason to continue at this point, will see when coming out whether we have furthers specs.

 

[kagemaru]

Why the discussion to me seem suddenly become so personal? I think my opinion is clear, I explain what appears to me, you not concur, ok not need to take this way. I don't think there is reason to continue at this point, will see when coming out whether we have furthers specs.

No no, not personal, I was just asking because I didn't want to misunderstand. Sorry about that. I hear what you're saying about the game though and it would be interesting to find out what the exact reasons were.:smile:

 

[assurdum]

No no, not personal, I was just asking because I didn't want to misunderstand. Sorry about that. I hear what you're saying about the game though and it would be interesting to find out what the exact reasons were.:smile:

Ah ok. Probably would sound like too much 'inconsistent' or insufficient how reasons, but I just evaluate the engine from what I have seen until now & even the overdraws don't justify so clearly to me the framebuffer decision because there would be a lot of alternative than a muddy IQ; I think the definition is more touchable than artistical, artigianal substitute of the overdraws, at least to the common gamers perspective.

 

[Devious]

Digital Foundry vs. Unlimited Detail

 

[(((interference)))]

From their tone, they seem to be knowingly misleading the general public - does anyone else think that their tech won't come with huge caveats?

The two dealbreakers seem to be the memory needed for the level of detail and the problems with animation - both things which have no conceivable solutions at the current moment, correct?

 

[Laa-Yosh]

In case you haven't seen it yet, this is what they're probably using to 'scan' real world content:

http://www.youtube.com/watch?v=CGZXkSke_vI

Basically a very fast (calculations are done remotely on Autodesk's farm) and advanced photo modeling system that generates a high poly mesh with textures; but of course all lighting, shading, reflections etc. are baked into that texture.

 

[Laa-Yosh]

The two dealbreakers seem to be the memory needed for the level of detail and the problems with animation - both things which have no conceivable solutions at the current moment, correct?

Exactly right.
Although by memory I'd rather say background storage and streaming tech - you don't need to load the entire level into memory in order to display it, but the streaming tech would face challenges very similar to id's Megatexture stuff.
At this time there's no commercial media format in sight that a game using this tech could be released on. Nvidia's cathedral demo uses 2.7 GBs of data for a single interior at a fraction of the detail level that's promised with 'Unlimited detail' tech.
http://www.youtube.com/watch?v=lpfaFrazOn4

According to Carmack it took id 5 years to go from a working virtual texturing techdemo to a completed game, with the ~20 years of id's game development experience and their nearly unlimited resources.
Euclidean is two guys on programming, as far as I know, and this Bruce Dell did not know in 2008 what CPU caches are (posted here on B3D). Draw your own conclusions...

 

[(((interference)))]

Exactly right.
Although by memory I'd rather say background storage and streaming tech - you don't need to load the entire level into memory in order to display it, but the streaming tech would face challenges very similar to id's Megatexture stuff.
At this time there's no commercial media format in sight that a game using this tech could be released on. Nvidia's cathedral demo uses 2.7 GBs of data for a single interior at a fraction of the detail level that's promised with 'Unlimited detail' tech.
http://www.youtube.com/watch?v=lpfaFrazOn4

According to Carmack it took id 5 years to go from a working virtual texturing techdemo to a completed game, with the ~20 years of id's game development experience and their nearly unlimited resources.
Euclidean is two guys on programming, as far as I know, and this Bruce Dell did not know in 2008 what CPU caches are (posted here on B3D). Draw your own conclusions...

Yeah, I meant storage memory rather than runtime memory (We need a different name for the two kinds of 'memory').

And yes, it'd be impossible to do any kind of game with varied environments and fit it on even a triple layer Blu-Ray, then you'd have to stream all that data in from disk. So unless they've also developed a brand new geometry compression system I don't see them getting around that.

I also don't trust them based on their rather supercilious tone you see in the video commentary and the DF interview.

But perhaps their tech may be useful in games with limited environments or with no variation in art assets from level to level like some XBLA/PSN titles (eg tower defense and other maze type games) or with procedurally generated environments (like From Dust/Minecraft) or even say for fighting games - would you be able to use their voxel system to render the static environments (which don't need to be animated) and use polygons for the characters?

 

[Laa-Yosh]

It's perfectly possible to combine voxels and polygons, in fact Sebbi made a note that with deferred rendering you could even make sure that the shading is completely consistent, since you're working in 2D, basically, and the algorithms don't care how you've rendered the contents of the g-buffer.

A game with a more limited scope could adapt this kind of tech more easily, indeed. But I wonder if an XBL/PSN game would have the budget to develop the proper tools and if it was worth the effort as they're not interested in AAA quality graphics anyway. Why spend your limited resources on this tech if polygons can get you along the way just as well?

 

[TheWretched]

Well... Outcast already did it over ten years ago... so yeah it's possible^^

 

[kagemaru]

would you be able to use their voxel system to render the static environments (which don't need to be animated) and use polygons for the characters?

IIRC Reach used voxels for their imposter system. Powerpoint slide here I think:

http://www.gdcvault.com/play/1014344/Automated-Level-of-Detail-Generation

 

[MJP]

Reach didn't render voxels or use them as any sort of runtime data format. Their offline pipeline for generating imposters would use a voxelized representation of the mesh to generate the non-textured low-poly imposter.

 

[kagemaru]

Reach didn't render voxels or use them as any sort of runtime data format. Their offline pipeline for generating imposters would use a voxelized representation of the mesh to generate the non-textured low-poly imposter.

So I didn't remember correctly after all lol

Thanks for the correction, I couldn't check the slides to make sure while I was at work.

 

[sebbbi]

Well... Outcast already did it over ten years ago... so yeah it's possible^^

Outcast had just a voxel height field for the terrain, just like Comanche and other games before it. The buildings and the characters were rendered by a software polygon rasterizer. Actually the first commercial game I worked on was Pathway to Glory for N-Gage. I programmed the voxel terrain rendering system for it

 

[TheWretched]

There's no rule that says voxels HAVE to be volumetric data or that a software rasterizer isn't allowed.

The original post was

It's perfectly possible to combine voxels and polygons

to which I just answered with Outcast...

 

[Cyan]

Outcast had just a voxel height field for the terrain, just like Comanche and other games before it. The buildings and the characters were rendered by a software polygon rasterizer. Actually the first commercial game I worked on was Pathway to Glory for N-Gage. I programmed the voxel terrain rendering system for it

OMG, I'm super impressed by your skills. Some of the games you worked on are techically great. I don't know if you will use voxels in the future... but I think it's a good start for someone who might enjoy that technology .Anyway, that's beside the point...

Talking about the N-Gage, which was meant to be my first portable console back at the time, injects an alternative perspective into this discussion. I've never ever been keen on Apple, and knowing how the gaming world is shaping these days, the whole N-Gage idea was certainly good. It was designed with a philosophy that was well ahead of its time. Too bad Nintento didn't partner with Nokia to create a similar console to 3DS but with some extra functionalities. It would be a fine gaming handheld, and even a phone, using two touch screens.

That would bring fundamental changes to the industry while also keeping the classic portable design. Those events should happen in a timely manner though, but sadly Nokia abandoned the project and Nintendo took a different approach. The chance for misapplication could occur, but I think a Nintendo + Nokia handheld console would be the best ever made.

 

[Kameradschaft]

I can't wait to see a face-off for El Shaddai from the guys at DF, I can't wait because I played the PAL demo recently on the 360 and it looked like the frame-rate was higher than the Japanese demo...maybe it was a placebo effect but I'm really curious to see if there really is any improvement on the PAL versions of the game.

I'll probably try to download the Japanese demo again to compare it and see if I'm saying BS or not.