Xbox boss questions “consumer demand” for 3D

I’m watching these arguments, and they seem very familiar. Like with HDTVs and Blu-Ray, I think.

Sony was a major supporter of Full 1080p, going so far as to include 2 HDMI ports in the PS3. People said it was too expensive. Those people were wrong, correct?

Blu-Ray. You want 1080p movies? How about a minimum of 25Gb capacity? Not in the beginning; It was too expensive, it had a war with HD-DVD. How about now? Do we still have to debate its successes?

Point is, if it was just Sony, fine, but major TV companies other then Sony are supporting this, within 2-3 years every TV you buy will have this tech; Sony has that inside scoop in this area, seeing as they’re in the TV business. So they’ll continue to implement this tech prominently, when it does hit that ‘acceptable’ penetration level, Sony will have an evolved and matured 3D implementation process shared among their studios working on their console, MS will half-heartedly follow suit, and everyone will forget this ever happened.

At least, historically.

Seems like a good idea to get a head start imo. Yeah its not in demand now, but at least they will have alot of the kinks worked out when it does become high demand.

Wow u guys musta fixed the sending messages to fast bug

what are you guys griping about?
You all always complain that the games arent in 60fps….well if 3D is the next step in tech, you are guaranteed that 60fps you wanted….
I also here gameplay is most important right? how does 3D ruin gameplay. In reality, it might add depth to a regular 2d experiences.

oh Gadzooks! I’m sure it would make your day to see me calling you names furiously (like you have done to me all this past week) trying to anger me all I can say is…..

PEW! PEW! PEW! keep trying kid, maybe one day you’ll be as great as me.

@53, no they didnt. I still get that error. Like right now.

lol, oh i’ve been posting away freely

@55

It would make my day to see you post something that wasnt an inflammatory string of faeces.

Try it once, you might like it.

@41

“You don’t need to overcome this though! If you move the camera and render two frames you don’t *need* the z-buffer info for the 3D effect. That’s the crux of why it’s a problem for TriOviz and not for a proper 3D implementation.”

Kinda lost track of this discussion, so back on topic:

Right, if the current cheapest method of rendering transparencies is to overlay them in a final pass then that means there is no depth information for them. Fine for a single viewpoint.

If you render two camera views you render two views with transparencies overlaid right at the front of the scene so it will not solve the problem. Transaparencies will still look like they are hovering in front of the 3D scene.

The one and only way to overcome this is to render transparencies along with the other 3D geometry and that means large changes to rendering engines and greatly increased workloads for engines that dont already do this.

@45. Interesting how you see the world: someone who is not interested in 3 d yet must be poor. A guy who needs glasses shouldnt drive a car because he would cause a massmurderesque crash…if i try to look at the world through your eyes i cant help but imagine you in front of your 40″ tv in an ss uniform wearing a tiny beard upon your upper lip.

@Gadzooks
Nah think about it some more the transparencies will be fine, the depth is generated by the difference of where it is placed between the two views. It doesn’t matter. Hard to explain without pictures but it really matters not a jot.

@61

Okay we agree to disagree then. I know for a fact that what I am describing is a problem for stereoscopic 3D. I might not be explaining it clearly enough but I can’t be bothered saying the same thing over and over.

It is a problem for stereoscopic 3d generated in a particular way (using the z buffer to generate the second image rather than rendering a second view properly). It’s a factual point, not something you can agree to disagree on. I can post a link to a technical doc later which will clarify.

@63

As I said, we agree to disagree because you are convinced you are right and so am I.

We will leave it there.

Well here are some facts which back up my view – can you supply the same?

http://www.technology.scee.net/files/presentations/Stereoscopic_3D/PS3_Making_Stereoscopic_3D_Games.pdf

Slide 35 – Covers *Alternative* ways to generate Stereoscopic 3D – Cons – Transparencies and reflections won’t work. Meaning that these are perfectly fine with the main described method. This problem you are talking about is very much specific to using a single 2D view + depth buffer.

Nice document. It actually highlights more restrictions in using stereoscopic 3D than I had known about.

Things like ‘avoid using small fast object’. So particle showers (explosions, etc) will need to be simplified or toned down.

The document also mentions to avoid shimmering by increasing texture filtering, so that’s going to be a huge burden on rendering times.

Also you keep bringing up 2D+depth map for some reason. What I’m talking about is nothing to do with 2D+depth at all.

I’m talking about multipass scene rendering to accomodate transparencies, things like deferred renderers/lighting which blend transparencies in as separate objects after basic geometry has been rendered.

We were led to believe that basic geometry and textures could be re-used for both eye frames, but multipass renderers can introduce object obscuring and texture blending by transparencies at the final lighting stage.

Sure, you can render two full scenes from different cameras, but to stop the game crawling to a juddering halt you would need to have at least one pass at transparencies while processing geometry.

This makes dual view rendering faster but slows regular single view rendering.

We’ve not touched on stippling yet either. Stippled shadows and transparencies are used in a huge number of games and will cause huge problems for dual view scenes, much like unfiltered textures. The problem is synchronising the stippling in both views. If you dont get them aligned then your 3D illusion is shattered.

So there are a great number of problems with dual view rendering that are not problems for single views. For these problems to be addressed renderers need to use techniques that in a lot of cases will catagorically slow down single view scenes. All the quick and dirty techniques developed over the last 5 years or so will need to be evaluated.

So please, do not tell me that 3D does not come at a cost to everyone, as it clearly does.

@66 It does come at a cost but the gaming industry is very adaptive. I’m sure the geniuses that are working with 3D at the moment will find alternative ways to do things, you can bet that any company going into 3D has already found ways to deal with most of the thing’s you have listed.

@67

Aye, naturally devs will get it all working. How efficiantly though? Nobody knows yet as we’ve only seen retrofitted 3D so far.

My original point was that as a non-3D gamer my gaming experience will be worsened, to a greater or lesser degree depending on how much additional resource is thrown at game engine development, assets, testing, etc.

This will naturally steer me away from titles developed as being 3D-compatible.

@Gadzooks
“Also you keep bringing up 2D+depth map for some reason. What I’m talking about is nothing to do with 2D+depth at all.”

See post 40 where you bought up transparencies and Batman:AA – that is all to do with 2D+Depth. You went on to talk about needing to include these sort of effects in the Z buffer as a work around.

There are effects which don’t work so well in 3D. You are missing the more obvious point though, in the transistion to 3d you need to save clock cycles so alot of the time these will be prime candidates for taking that particular effect out. In some cases these are actually catered for in the 3D effect anyway.

What is the dev trying to do? They need to save and optimise. To suggest they would not only take a more costly approach but also retrofit this more costly approach into and to the detriment of the 2D version is ludicrous.

@69

Read post 40 again. It was relevant in context to post 66. If there was z buffer info (i.e transparencies were in the geometry pass) then there would be no problem rendering 2 views from them. 1 pass for geometry, 2 passes for lighting = quicker than 2 passes for geometry+lighting.

“in the transistion to 3d you need to save clock cycles so alot of the time these will be prime candidates for taking that particular effect out.”

Well that makes me feel a whole lot better about my games not being compromised!

I get it okay, you like your gadgets and new tech! I’m not trying to downplay 3D, you’ll get it and probably love it and I’ll be pleased for you, but there is an impact on us ‘dont haves’, or more accurately ‘dont wants’.

I personally am not going to part with my hard earned so that a tiny minority of people can enjoy new toys while I get a reduced service.

As always I will vote with my wallet, and my wallet says ‘steer clear of games developed with 3D in mind’. That is the end of it.

Can we now drop this subject?

@70
When I said take the effect out I was talking specifically in the 3D implementation – it would stay in the 2D.

We can drop it, but I have finally worked out what you are talking about in 66 so if you don’t mind I’ll continue a little you can stop reading if you want . It’s a useful idea in the 2D + depth type implementation but it’s not an option considered in the render 2 full views approach in the doc I linked. The only thing moved out of the individual views are the shadow maps which are scene specific. I still can’t see them retro fitting something more costly into the 2D renderer. The 3D render has to be different anyway they will continue to do things differently. Use the best approach on a per render basis.

On the general point if they used the resource they put into the 3D on the 2D would it make it better? Yeah of course. Will they spend that resource on 2D if 3D wasn’t an option? Possibly not. They are willing to put the resource on 3D because they see a benefit in terms of potential higher sales and experience in an emerging tech. Obviously real world some games will see people pulled off 2D to make 3D a feature and 2D will suffer because of it. If resources are tight though I can see 3D being one of the first features dropped.

The vast majority of 3D stuff this gen is going to come from Sony internal and they are evidentially sharing a lot of tech and experience.

Realistically I also see a lot of 3D implementations taking the scaling route to quick and dirty optimisation. Not ideal but something which is quick and also has less impact in a 3D implementation. 640×720 in 3D looks better than 640×720 in 2D.

As mentioned previously games tend to have a fuckton of features these days many of which I don’t use or want. I’ll continue to base my decisions on the features they have which I want rather than the features they have which I don’t though.

On a general note for the document I linked – whilst it highlights a lot of issues you might see with 3D, I thought it was quite interesting in the aspects they consider – in particular in relation to ensuring the end image is comfortable to the viewer.

“you can stop reading if you want ”

No, I think I’ll continue this ‘last word’ competition if you dont mind.

You are talking about games having completely different 2D and 3D renderers, correct?

It’s a nice idea and devs with infinite time and money budgets might adopt it. Lets please just be realistic though, shall we? It just isnt going to happen in most cases.

I’d be perfectly happy if a guarantee was made by devs that games are being optimised for traditional viewing but we both know that showcasing 3D will be the overriding priority right now for most devs developing a 3D-compatible game.

@72
No not two completely different renderers – it’s easy enough to turn certain features on/off though isn’t it? Demos of render tech do it all the time.

It’s similar to supporting split screen play – which incidentally supporting 3D is a benefit for (and vice versa). There are some differences sure, but if you have split screen already then it is easier to support 3D and if you are going to support 3D then it’s also easier to put in split screen where you may not have considered it before.

Take up of 3D tech is going to ensure for some time yet that the viewing is going to be optimised for 2D though. Devs would be crazy not to.

Point taken on the splitscreen issue.

3D showcasing will take priority though, mark my words.

Split screen has taken too much of a back seat in these online times – it’s due a bit of attention.

@75

I hope so. This point needs to be hammered into the brain of every single Criterion employee.

Split screen Burnout or else!