Digital Foundry tech analysis channel at Eurogamer

[Laa-Yosh]

Ah... interesting.. Any idea why they wouldn't/couldn't have used FaceFX

I don't believe in any of these procedural human generator stuff. They're all totally lame and should be avoided for all but the most basic characters (ie. architectural visualisation and such).

By the way, that FaceFX is for animation, whereas Facegen is used to "create" models and textures.

 

[Laa-Yosh]

Considering that they've received a lot of bashing for the sub-720p resolution (even if a lot of people noted how good the lighting is) if they want to improve their engine, the number one feature would be fixing the framebuffer resolution. And now they have actual measured data from two games on the engine to design the new one, so they can implement efficient tiling as one possible solution.

Or they could redesign the entire HDR pipeline too but it is unlikely.

 

[AlNom]

Considering that they've received a lot of bashing for the sub-720p resolution (even if a lot of people noted how good the lighting is) if they want to improve their engine, the number one feature would be fixing the framebuffer resolution. And now they have actual measured data from two games on the engine to design the new one, so they can implement efficient tiling as one possible solution.

I'd argue 2xMSAA @ 1152x640 would be more beneficial than a <2x increase in pixels, but maybe that's just me.

Or they could redesign the entire HDR pipeline too but it is unlikely.

I'm be curious as to their evaluations of other studio's works, particularly light pre-pass and deferred shadows.

 

[obonicus]

Considering that they've received a lot of bashing for the sub-720p resolution (even if a lot of people noted how good the lighting is) if they want to improve their engine, the number one feature would be fixing the framebuffer resolution. And now they have actual measured data from two games on the engine to design the new one, so they can implement efficient tiling as one possible solution.

Well, I'm not saying they won't, but rather that if they to make sacrifices, I'd bet that 720p resolution will hit the chopping block again (but there's no reason to assume that they'll need to make sacrifices). Despite complaints, their game was an enormous success and was selling well a year and a half after release. The only other shooter in the same sales league also sacrificed HD resolutions.

 

[Laa-Yosh]

If you mean COD4, it has 2x MSAA which also doubles framebuffer size and fills the EDRAM completely.

Bungie's options are like this:
- drop the double framebuffer for HDR and use a single buffer at 720p
- drop the double framebuffer for HDR and use a single buffer at 2x MSAA sub 720p
- use tiling with 720p
- use tiling with 2x MSAA and a sub HD resolution
- don't change anything with the framebuffer (wrong idea IMHO)

Which one's best would depend on a lot of factors...

 

[obonicus]

If you mean COD4, it has 2x MSAA which also doubles framebuffer size and fills the EDRAM completely.

That's not the issue. You supposed that the complaints about sub-HD resolution might convince Bungie to go in a different direction. I'm saying sub-HD resolutions have been more than vindicated, Bungie has no reason to listen to complaints. Maybe they'll look at CoD4 and decide that trying for 60fps is more important.

 

[Laa-Yosh]

If they add MSAA, their double-HDR framebuffer won't fit into the EDRAM either.

I also don't think that Halo would need 60fps, it's a relatively slower paced game.

 

[Acert93]

Considering that they've received a lot of bashing for the sub-720p resolution (even if a lot of people noted how good the lighting is) if they want to improve their engine, the number one feature would be fixing the framebuffer resolution. And now they have actual measured data from two games on the engine to design the new one, so they can implement efficient tiling as one possible solution.

Or they could redesign the entire HDR pipeline too but it is unlikely.

If you mean COD4, it has 2x MSAA which also doubles framebuffer size and fills the EDRAM completely.

Bungie's options are like this:
- drop the double framebuffer for HDR and use a single buffer at 720p
- drop the double framebuffer for HDR and use a single buffer at 2x MSAA sub 720p
- use tiling with 720p
- use tiling with 2x MSAA and a sub HD resolution
- don't change anything with the framebuffer (wrong idea IMHO)

Which one's best would depend on a lot of factors...

They should rip HDR off Fable 2.

As for as things to fix in Halo 3's engine I think the big ones are:

1. Texture Filtering
2. Edge Aliasing
3. Shadow Quality/Filtering

From there I would argue that the game, based on what they have done, would look amazing with Ambient Occlusion to go along with their IL hack and HDR. More particles, more dynamic shadows, etc would also be high on my list. Artistically--FIX THE HUMANS. Next, address animation.

If they are willing to address aliasing issues I don't think getting to 720p should be on the radar.

They should skip tiling, keep the 10MB frambuffer sub-HD, and go NUTS visually. Push comes to shove, do HUGE battlefields with a lot of smoke. 720p should be the last thing on the check list. IMNSHO.

 

[Rangers]

Bungie has stated Reach is a new engine. I dont have the link but I saw a video interview, somebody asked to compare ODST team size/dev time to Reach, and the Bungie guy noted a lot of the effort/time on Reach was being expended creating a new engine (and he said they didn't want to use a 2-3 year old engine for Reach, as close to an indictment of Halo 3's looks as you'd ever get from Bungie), whereas that wasn't an issue on ODST.

I guess it could have elements of Halo 3 engine in it, but it certainly doesn't lean that way.

Playing ODST now though, I'm reminded of a lot of the good fun stuff about Halo 3 (in the flashback scenes of course), which is they put you in these huge open battlefields, with massive draw distances and scope, dozens of vehicles littering the battlefield you can hop on at any time, etc. It really is in some ways impressive and different than any other FPS I know of out there. I rather hope Bungie doesn't look at COD's success and do away with the open nature of Halo. If anything, openness in FPS seems to be going the way of the dodo in most FPS, in favor of linearity and scripting with perhaps better looks.

 

[Laa-Yosh]

They do need a better character artist, but I have hopes that Kenneth Scott of id Software fame will have something to do with the next Halo game. Even though he's been hired for the Microsoft team, they don't seem to have any projects going on with the Peter Jackson stuff canceled, so there's hope that he's gonna help out Bungie. Then again, he did not help with ODST...

Animation is a worry. Bungie has some hardcore animators - hardcore in terms of completely rejecting mocap. Unfortunately, they don't seem to be good enough, and even H3 has a lot of really bad animation - particularly when Master Chief has to run in cutscenes. But they are improving... and you can't really mocap the Covenant, whom you see the most through the game anyway.

As for your other points...

1. This is a performance issue and actually it seems to have been looked into with ODST to a small level already. But I don't have high hopes, it's clearly a secondary or tertiary issue.

2. Just going full 720p res should help a lot here already, but I'd really expect a second generation X360 release to go with 720 and 2xMSAA. A lot of titles have done this before, even if admittedly lacking on the HDR front... Halo3 is beautiful in this regard, I've re-watched its cutscenes and some of the stuff is really amazing.
Just 2xAA wouldn't help that much IMHO. It is not enough to balance the lack of detail... even 720p with no AA would look better. Large battlefields are the very reason why Halo needs more pixels instead of a little bit of multisample AA. Stuff in the distance just looks horrible in Halo3...

3. They are on it, based on the Siggraph papers and other presentations. It's actually a bit surprising to me that shadows seem to become one of the most complicated issues in this generation. But it makes sense after I reconsider all the pain we went through with PRMan's shadow maps before we've moved to Mental Ray and raytraced shadows (basically trading tech artist time for rendering time). So this is the one I have the least worries about.

 

[Laa-Yosh]

I rather hope Bungie doesn't look at COD's success and do away with the open nature of Halo. If anything, openness in FPS seems to be going the way of the dodo in most FPS, in favor of linearity and scripting with perhaps better looks.

Fortunately it seems that they plan to go even more open and sandbox with Reach. Although I do expect them to introduce classes for Spartans (based on the Reach logo) so the single player game could end up very interesting - different experiences with a sniper, a heavy weapon guy and so on, not to mention four-player coop with such distinct characters. Basically a large, open battlefield with objectives that have multiple solutions and approaches.

Reactive AI and emergent gameplay are their trademarks and they have ~10 years of research and experience here. It's a no-brainer to capitalize on this, instead of trying to mimic the COD games... IMHO.

 

[Acert93]

Hmmm Maybe Bungie with 4 or 5 classes can essentially created 4-5 "campaigns" around a single huge battle. That would be one way to really create a large game and re-purpose assets. 5 stories, 4 hours each, 20 hours. Could be interesting. Especially if they are a loose squad to allow 4 player online cooperative. e.g. The sniper a half mile up trying to get to a way point (through a thicket of grunts) while you are in a Warthog in the valley. Your ease of success depends on your friend getting to the hill and giving cover at the check point.

Basically the same action from different angles. It would be neat to see someone do that.

 

[Laa-Yosh]

Well, odds are that we'll have to wait until the next E3 to find out more about Reach. Or maybe GDC, that's in march I guess, and I think Bungie has already made some presentations on unreleased titles before so there's some hope.

 

[rekator]

Well, odds are that we'll have to wait until the next E3 to find out more about Reach. Or maybe GDC, that's in march I guess, and I think Bungie has already made some presentations on unreleased titles before so there's some hope.

Or wait for the beta, it's probably in the same time frame than Halo 3 beta (may) if Bungie want to have time to use data from it.

 

[Laa-Yosh]

Beta's for the multiplayer, although you're right that we'll probably be able to draw some conclusions from it

 

[grandmaster]

Today's feature is mostly from MazingerDUDE and concentrates on Ninja Gaiden vs Sigma 2. A tip of the hat to joker454 for raising interesting questions in the game technology thread on how to port a game seemingly designed exclusively with the Xbox 360 in mind...

 

[AlNom]

Great work on the screen captures.

 

[Billy Idol]

the perception of the werewolves scene versions was astoundingly different

 

[AlNom]

Now if only we could have both massive amounts of enemies and great lighting/textures...

 

[egoless]

Better have your sarcasm detectors engaged

On page 1 you say NG is running @ 1280x718 on PS3. On page 2 you callout 360 with the sub-HD decry. Pixel counters and the like, pride themselves on technicalities. It doesn't matter how far above the 480 hardline of the previous generation we are now, nor does it matter what eye candy is on screen at said higher-than-480-resolution. There is a hardline 720p or 1080p or bust mentality.

Therefore, NG is also (insert zomg) sub-HD and entitled to all derision and soapboxing that entails from the usual ilk.

Unless....