3D Gaming*

Sounds like they are using some sort of 2D image plus Z-buffer technique to create the two seperate views. There are issues with this because of being able to view behind objects in one view and not the other, but if they know where these issues are happening maybe they are able to render the missing information seperately after the initial conversion..

 

Things would look pretty flat with that option. Cardboard cutout 3D gaming just isn't worth it, unless you are trying to represent cardboard cutouts.

 

I think so. Look at this article from the same site.

http://www.videogamer.com/news/crysis_2_3d_more_pleasing_than_killzone_3_3d.html

 

Dunno but read this if to be trusted.

http://neogaf.net/forum/showpost.php?p=22080891&postcount=75

http://www.videogamer.com/news/crytek_magic_prevents_crysis_2_3d_performance_issues.html

 

If their 3D is always on, this means that the method may not be effective sometimes.

OTOH, when you read KZ3 interviews, sometimes the developers think full 3D effect + post processing in select scenes may cause headache (Apparently too much for the brain to process).

Both parties need to adjust their techniques to give the best results.

If Crysis's depth method is accepted by the gamers, the "real 3D" camp will adopt the technique (no loss to them !), and then add fuller 3D effects whenever the simpler one fails to deliver, like right now. There is no real "us vs them" boundary here.

As for verticality (e.g., jumping down a skyscapper), I agree it's the easiest way to show the effect because usually there's nothing in between: Just the faraway ground level and where the player is standing. The depth/volume will be very apparent. KZ2 has good vertical MP maps too, I expect KZ3 to show the vertical 3D effect well also.

MotorStorm's and Red Dead Redemption's wide open landscape will be very suitable for 3D. I think in this kind of level, if it's the billboard type 3D, then we may notice it.

 

Wall Street Journal has another overview on 3DTV:
http://online.wsj.com/article/SB100...4703714698146.html?mod=WSJ_newsreel_lifeStyle

Too much scattered info to summarize. Have to read it yourself.

 

When you only have the depth buffer and on camera to make 3d , the deeper the better.That's obvious.Seeing how small their paralax is makes that quote really easy to understand.

This confirms to me that crysis 3d will be on the cheap side of stereoscopic gaming.
And i disagree with you Patsu , the stereoscopy looks way better when there is intermediate objects.

 

When I'm standing on top of a skyscraper looking down ? Is it because there're more points of reference for the depth ?

I was expecting the side of the skyscaper to provide that continual sense of height/depth (all the way down to the ground floor).

 

Actually according to certain sites it is actually THE best 3d experience at e3

http://www.1up.com/do/feature?pager.offset=2&cId=3179985

 

I can understand that. The sense of height/volume is accentuated from the top of a skyscraper. If Crysis has more open space, then it's also a big showcase.

But for a mature 3D presentation, we also need a variety of techniques for close up melee, elementals, post-processing, subtle depth, etc.

I expect the developers to mix and match techniques until they find a middle ground over time.

 

Arthur(aegies) also seemed really hot on it(he was positive about it, but did note its weaknesses).

Seems sort of hit or miss. Some people loved it, other didn't.


I also would bet money that Crytek is going to continue tweaking it. If they can maintain the depth on the really vertical showcases but make the closer stuff look a bit better, then they might very well end up having the best solution.

Keep in mind, their image quality is going to be a lot better than KZ3 because they aren't being forced to cut the resolution in half... And their framerate will also be more stable since they aren't doing so much extra work...

 

Frankly ,i don't understand how you could praise a 3d without interaxial input and narrow parallax, over the reall stuff.
Good for them ,but that's just wrong.I toyed too much with 3d now to buy that.I need interaxial and wide paralax.

Patsu:I just meant that a good 3d is not related to openness or draw distance or world scale.
It's related to filling the depth between foreground and background: intermediate depth.
The more intermediate objects the better.

 

I see. I think at the beginning, people will appreciate the really obvious effects until they play more 3D games to enjoy the subtle ones.

Yes, based on the other 3D developers' feedback, the more subtle and realistic 3D effects require rendering 2 images, which tend to lower the resolution due to double in workload. If Crytek can find a depth only solution that works in all cases, it would be a welcomed addition. However, it may take time to find. In the mean time, people will also find ways to optimize/simplify the dual image approach.

 

It's not the subtle effects that need 2 different pictures , it's the biggest ! (wide parallax positives and negatives !)

 

Hmm... we will need side by side comparison. I can't see it ! @_@

 

And what makes you certain that their technique is not better or impossible without having to render two images.

Its like saying that the ps3 cannot do MSAA because it does not have the hardware for it.

The point is that it does not have to, when it can accomplish better result with MLAA.

I think at this point we should be more open minded to what developers can achieve especially when they think outside the box.

 

Well said.

 

I think _phil_ is saying the depth-based approach has been studied before, and is generally inferior to the 2 image approach.

MLAA studies showed that it is equivalent to 16xAA or above. So the situation is different. May be another example would work better.

For an obvious scenario, I thought the depth approach and dual image approach should provide very apparent effect. I can't really "visualize" the differences between them without a side-by-side comparison, or a research paper showing the shortfalls. Right now, we know the depth-based 3D effect in Crysis may not be consistent, but the 2-image approach may lower the resolution.

 

How exactly are they different, my examples I mean. MSAA and MLAA are to ways of achieving the same thing.

FYI, they do use two images, its just that they do not have to render two images.

Read the interview on videogamer, it explains their process a bit.

 

I was told on paper, depth-based 3D effect is inferior to dual image.

But on paper, MLAA is equivalent to higher level of MSAA when it works.

EDIT: I thought the videogamer article is lacking details and also inaccurate. e.g., it claimed that 360 can't do 3D when we already have stereoscopic 3D game on the market for 360. Depth-based approach does not need 2 images (no need to render twice), but I have also seen discussion that mention it's not as good as "real" 3D effects.

Nonetheless, it may be possible to have a beautiful depth-based 3D game.