• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

New Source 2 direct-feed screenshots - Taken from Valve's VR presentation

Durante

Member
Is the pixel density really such a good idea
Pixel throughput, in both density, quality and temporal frequency. And yes, if you are doing VR it's a very good idea indeed to sacrifice flashiness and postprocessing for 90Hz, stereoscopy, resolution and high-quality pixels.
 

Easy_D

never left the stone age
Looks good. Valve's always managed to make stuff look "solid" if you know what I mean. Like an actual place, even as far back as HL2.
 
That part in the video with the floating parts and AR texts? People said they're tangible and readable. You're there. So again, it looking drab on these slides is meaningless until you try it in VR.
 

drowsy

Member
What I love about Source in particular is just how damn clean it looks. UE4 and Cryengine and all these other more recent engines have really cool features and complicated lighting models, but it's damn near impossible to get them to look clean. Frostbite is probably the worst offender, to the point that even at 4k with 4xMSAA you end up with a screen full of artifacts. It's not helped by the level designers at DICE really loving their fences and telephone cables.

And with Source 2 being so closely tied to VR, it's obviously going to continue that tradition of focusing on image quality over cool effects. I'm all for it.
 
What I love about Source in particular is just how damn clean it looks. UE4 and Cryengine and all these other more recent engines have really cool features and complicated lighting models, but it's damn near impossible to get them to look clean. Frostbite is probably the worst offender, to the point that even at 4k with 4xMSAA you end up with a screen full of artifacts. It's not helped by the level designers at DICE really loving their fences and telephone cables.

And with Source 2 being so closely tied to VR, it's obviously going to continue that tradition of focusing on image quality over cool effects. I'm all for it.

Well, that was a better explanation for why I like Source visuals over most other engines than mine. The clean visuals are definitely at least a huge part of it. Games in engines like UE4 or CryEngine often just feel so incredibly overdone, especially AAA titles.
 

belmonkey

Member
Well, I'll tell you what that demo looks like on the Source engine: nothing. Because OG Source would curl up and die with a 600k poly model.

Is that 600k poly model a character model? Seems kinda high considering The Order had 100k+ poly character models.
 
Is that 600k poly model a character model? Seems kinda high considering The Order had 100k+ poly character models.

Vlachos in the video says "all these moving parts alone are like 600k triangles polysomething can't here this part, don't know the exact number, but they're a lot of triangles!"
 

Schryver

Member
Now people will look at these, compare them to what they see in their latest sub-1080p 30FPS post-AA'd frame of reference, and conclude that it's not impressive.

While disregarding all of this (and all the other challenges of VR rendering, such as latency and geometry amplification):


Context, my dear friends, is everything.

Thank you as always
 

gofreak

GAF's Bob Woodward
Now people will look at these, compare them to what they see in their latest sub-1080p 30FPS post-AA'd frame of reference, and conclude that it's not impressive.

While disregarding all of this (and all the other challenges of VR rendering, such as latency and geometry amplification)

Can you elaborate on the last bit, geometry amplification?

Doing post-draw-call improvements to geometry (e.g. tesselation) to help offset triangle setup limits that might become more obvious with stereoscopy?

Kind of curious because I thought I read somewhere that tesselation and vr aren't a good mix, but trying to remember where I saw that...
 

RetroStu

Banned
tmtniu.jpg


It kind of looks like an updated version of the scientists lab in Half Life 2.
 

THEaaron

Member
Don't know how anyone can say this is disappointing. We are talking about advanced VR rendering where you have very very different demands compared to traditional rendering.

You need a lot of juice to produce this image for a good VR experience.
 

Durante

Member
Can you elaborate on the last bit, geometry amplification?
That was basically just a poor choice of words on my part. What I meant is that you have 2x (or less than that due to smart tricks, but still more than monoscopic) the geometry load because of stereoscopic rendering, in addition to the pixel load outlined on those slides. But the term "geometry amplification" is used mostly in the context of hull/geometry shading and tessellation, so I see how that can be confusing.
 
You get AO, PBR, and 4x MSAA now and it still doesn't satisfy you? The lighting looks realistic, I'm sure it will look different outside like the plantation demo. I just hope for high-resolution textures because that hurt the longevity of Steam 1 graphics.
 
Do you guys reckon Valve with still support consoles with their next full retail priced efforts such as Half Life 3 or Left 4 Dead 3? Havn't updated my PC in a while and i'm not sure I want to.
 

gofreak

GAF's Bob Woodward
That was basically just a poor choice of words on my part. What I meant is that you have 2x (or less than that due to smart tricks, but still more than monoscopic) the geometry load because of stereoscopic rendering, in addition to the pixel load outlined on those slides. But the term "geometry amplification" is used mostly in the context of hull/geometry shading and tessellation, so I see how that can be confusing.

Ah gotcha, thanks.
 

dave is ok

aztek is ok
Do you guys reckon Valve with still support consoles with their next full retail priced efforts such as Half Life 3 or Left 4 Dead 3? Havn't updated my PC in a while and i'm not sure I want to.
I wouldn't be surprised if L4D3 came to consoles considering it's always done well there.

But so much of Valve's philosophy is based around semi-regular updates and Workshop and things like that - so who knows. They might never do another console game.
 

Raven77

Member
I like the amount of geometry I'm seeing, especially on that computer thing on the left. Overall though this looks pretty generic. Not impressive at all. Will wait to see a non-VR demo though.
 

TimFL

Member
I wouldn't be surprised if L4D3 came to consoles considering it's always done well there.

But so much of Valve's philosophy is based around semi-regular updates and Workshop and things like that - so who knows. They might never do another console game.

Do consoles still enforce that retarded "pay for adding DLC/update" rule? I thought they went away with it this gen.
 

StMeph

Member
That was basically just a poor choice of words on my part. What I meant is that you have 2x (or less than that due to smart tricks, but still more than monoscopic) the geometry load because of stereoscopic rendering, in addition to the pixel load outlined on those slides. But the term "geometry amplification" is used mostly in the context of hull/geometry shading and tessellation, so I see how that can be confusing.

Given how the graphics are duplicated and running in parallel, would VR be one of the few applications where you could get close to full performance out of both cards in an SLI setup?
 

EVIL

Member
Given how the graphics are duplicated and running in parallel, would VR be one of the few applications where you could get close to full performance out of both cards in an SLI setup?

out of the box no, since you are still rendering one image 2160x1200 but nvidia is working on something called VRDirect, which does just that. having 2 970's would be better then one 980
 
Given how the graphics are duplicated and running in parallel, would VR be one of the few applications where you could get close to full performance out of both cards in an SLI setup?

out of the box no, since you are still rendering one image 2160x1200 but nvidia is working on something called VRDirect, which does just that. having 2 970's would be better then one 980

Valve already tested AMD's multi-GPU solution and that almost doubled their framerate.
 

Salsa

Member
honestly looks fine, kinda surprised at the negative reactions. VR shots are never top notch in terms of graphics since you're basically showing off something else
 
honestly looks fine, kinda surprised at the negative reactions. VR shots are never top notch in terms of graphics since you're basically showing off something else

Initially the thread didn't have those VR infos (the vid and the descritption which has Durante's post at the end). So, people saw the thread titled "direct-feed screenshots", came in and either ignored the link provided in the OP or they saw it, but didn't read it, then saw the pics and naturally got disappointed.

edit: okay, the following posts show those edits didn't help one bit :p
 
I wouldn't be surprised if L4D3 came to consoles considering it's always done well there.

But so much of Valve's philosophy is based around semi-regular updates and Workshop and things like that - so who knows. They might never do another console game.

Most priced games still seem to come to consoles such as Counter Strike Global Offensive and Portal 2. With something like Half Life 3 I don't care about updates I just want that damn base game :c Even though i've lost touch with modern PC gaming I think I would upgrade my PC if HL3 were PC exclusive.
 

dr_rus

Member
Source 1 reveal compared to this was night and day different. Can we have that valve back?

You'll have to not buy anything VR related first and after it fails you'll probably get your Valve back.

Which is unfortunate really as I seriously suspect that this is exactly how it'll end up - a lot of effort spent on helm VR tech which won't take off and all of this will go to waste essentially.
 

-COOLIO-

The Everyman
You'll have to not buy anything VR related first and after it fails you'll probably get your Valve back.

Which is unfortunate really as I seriously suspect that this is exactly how it'll end up - a lot of effort spent on helm VR tech which won't take off and all of this will go to waste essentially.

nah.
 

Skinpop

Member
I was really expecting more from them. Even the last year's pics didn't look that groundbreaking: http://www.gamersbook.com/News/Article/ID/378/Source-Engine-2-Screenshot

Valve really missed their oppurtunity to stand out if they released it 2 years ago! Now they really have serious competition that will eclipse them. I hope it will be at least easy to develop with.

As a programmer myself, flashy images doesn't excite me that much, it's the tools and workflow that matters. Consumers being excited by an engine doesn't mean anything. This is gdc, not some hype building carneval.
 
If you are based on - this is VR and not representative of the true quality of the engine- I gave how Source 2 looked in non VR mode: http://www.gamersbook.com/News/Article/ID/378/Source-Engine-2-Screenshot : which still doesn't look that special ! It may have looked good 2-3 years ago !


Also UE4 looks very good even in VR: http://www.neogaf.com/forum/showpost.php?p=154482016&postcount=1073

That presentation has been confirmed to be from 2011. A lot could have changed since then. Also if you look closely those trees are Source 1 assets. And they're probably not the only one in that pic that's older stuff.
 
As a programmer myself, flashy images doesn't excite me that much, it's the tools and workflow that matters. Consumers being excited by an engine doesn't mean anything.

As a developer and programmer too, I agree. An engine interests us the most in its ease of use and the techniques it has in its workflow. Consumers may be impressed by sth that looks great artistically and weak technically and many times they don't consider how sth is great in tech because it didn't appeal to them.
 

Skinpop

Member
As a developer and programmer too, I agree. An engine interests us the most in its ease of use and the techniques it has in its workflow. Consumers may be impressed by sth that looks great artistically and weak technically and many times they don't consider how sth is great in tech because it didn't appeal to them.

yeah, maybe I misunderstood you :)
 
Top Bottom