Valve Goes Cinematic Discussion

C

CefiroZ

Limp Gawd
Joined
Jan 17, 2004
Messages
218
I want to make it clear as I begin that I'm not trying to start a big fight (if one is possible) with this post.

I am curious as to what people think about the "Valve Goes Cinematic" article that was posted on HardOCP a little while ago. I, for one, am a bit conflicted as to what my thoughts are. I am all for making games look better and I hope these tools can be used to create more immersive environments. However, I see some problems, as mentioned in the article. Maybe I want to look an a poorly detailed rock, how will the game engine know to bring it into focus? What if there is lots of stuff happening all over the place? What if I'm looking somewhere other than where my gun is pointed? I see some of these blur effects as being prohibitive to realism. I know if I were in a real-world room with hostiles I'd want my vision to be crystal clear over my entire vision, I wouldn't focus my vision on a specific point while losing focus on the surroundings. I guess this all depends on the implementation that the hardware/software developers go for, but I'm not sure it can be done in a truly natural fashion.

I must say that the Color Correction could be a great thing to set tones for a scene or create lighting transitions, etc. The film grain doesn't seem very useful, maybe only for period pieces or odd situations, but not a common thing. Right? Wrong?

My other problem with this is that is has not worked before. 3dfx tried it back in the day to no success. Obviously there were problems with 3dfx, but the community never caught on to their cinematic ideas, so what makes Valve different? 3dfx's implemenation could be argued that it is a proprietary technology and thus was not wise to support, however, couldn't the same thing be said for this new stuff since it needs Source? How open will these technologies be for others to use? Are these going to be a part of OpenGL/Directx?

Sorry if the above is random and/or jumbled, I just typed as I was thinking (a deadly combination, I know).
 
If there was a small camera that fit on top of your monitor that could tell what part of the screen your pupils were looking at, the "focus" effect could be cool. Otherwise, nah, I don't think it would work too well.
 
I can see these effects working fine for singleplayer games for the most part to help add to the experience or to guide you to certain parts of the scene. But for multiplayer I don't see how Depth of Field would work well for gamers.

We'll just have to see what happens. Right now neither method from Valve is currently playable in real time, the performance just isn't there yet.

Color Correction and Film Grain though don't really take any performance hit. I see great potential for Color Correction.
 
I think it's interesting that Valve is experimenting, but countless others do the same, and I find it a little strange how much PR is being put out for their rendering experimentations.

What's particularly strange is that Valve's depth of field and (particularly) motion blur have serious consequences on performance, and combining these features and achieving playable framerates is a fair amount of time down the road. I believe Bit-Tech mentioned that for the video footage, the render time for each frame was around two seconds.

Some of these things we'll be seeing in future Valve games, but I think it's almost misleading to implement them in Day of Defeat and start the presses when there's no possiblity for some bits to actually happen. Maybe I'm used to the old ways, where graphics gurus like John Carmack tooled around in the background, keeping his tricks secret from everyone's eyes and ears, but I don't quite understand much of this Valve PR.
 
phide said:
I think it's interesting that Valve is experimenting, but countless others do the same, and I find it a little strange how much PR is being put out for their rendering experimentations.

What's particularly strange is that Valve's depth of field and (particularly) motion blur have serious consequences on performance, and combining these features and achieving playable framerates is a fair amount of time down the road. I believe Bit-Tech mentioned that for the video footage, the render time for each frame was around two seconds.

Some of these things we'll be seeing in future Valve games, but I think it's almost misleading to implement them in Day of Defeat and start the presses when there's no possiblity for some bits to actually happen. Maybe I'm used to the old ways, where graphics gurus like John Carmack tooled around in the background, keeping his tricks secret from everyone's eyes and ears, but I don't quite understand much of this Valve PR.
Click to expand...

What was interesting to me was that sitting at that table along with me were all hardware editors, no gaming websites were present. This was definitely focused at the hardware guys, which is cool, I love hearing about new 3D effects, it makes me warm and fuzzy inside :)
 
My first thougths when i read the article were " wow, 3dfx was five or six years ahead" and " I really hope everything except color correction is used in cut scenes only"

I mean some depth of field is realistic, but those pictures made it look like someone had smeared vasiline on the sides of my monitor. Motion blur isn't the worst idea, but becomes very annoying and degrades detail level( i guess that is thier point). Look at games like Grand theft auto 3 and need for speed undrground. They both allowed for motin blur and it was horrible. In both games i turned it off. In real life honestly how fast does somethign have to be going befor it blurs? I guess it depends on it's size, and how focused it is, but i mean some guy running 10 miles an hour across a field is not going fast enough to blur. Granted valves pictures had the camera panning too, but the second picture of the building why is that blured?

I can see as a cut scene cinematic effect, these would be very welcome, as cut scenes are little movies, but eyes work differently from a camera, and why i'm playing a first person shooter, i want the image to mimic my vision.

put it this way, If i want to play blurry, non detailed textured games i'll spend 300 dollars on a console, and save myself the extra $1000 a computer would have cost me.
 
Motion blur is incredibly awesome when done right. It needs next-gen hardware to do it right though, which is why it probably looked like crap in GTA3. Perfect Dark Zero's blur effects look great in my opinion, and make the game look better than if it didn't have it.
 
I feel that the motion blur in Need for Speed Most Wanted is done pretty good. It isn't too intrusive, they've set it to engage at higher racing speeds in this new game, which is better.
 
Superunknown98 said:
My first thougths when i read the article were " wow, 3dfx was five or six years ahead" and " I really hope everything except color correction is used in cut scenes only".
Click to expand...

That was more or less the same conclusion that I came to, along with Color Correction having a lot of potential.

I am curious why Valve is pushing this. As phide mentioned, why hasn't Valve just let this develop in the background till it's ready and then just plop it into a game for us to see? One could think that they are pushing it so hard because it can't be done without the proper hardware and they want to cause a buzz to get card manufacturers to move in Valve's direction. I suppose this would give them a competitive edge if it happened, but as the consesus seems to be, do we even need this stuff?

Personally I think that advanced hardware should be making our play experiences better in the future in a different way. I want more creative games with high fun factors and more realistic physics, artificial intelligence, etc. How about making the graphics hardware better at displaying things accurately than worrying about wasting time making them render blurry images? Unless a game is going for gritty realism these effects seem somewhat limited in use, how will cartoon-ish graphics benefit from stuff like this: not much.

I hope Valve keeps their heads in the right place: creating the most fun games possible. I hope they don't turn into the "make a ho-hum game to highlight our awesome graphics engine" company.
 
How about making the graphics hardware better at displaying things accurately than worrying about wasting time making them render blurry images? Unless a game is going for gritty realism these effects seem somewhat limited in use, how will cartoon-ish graphics benefit from stuff like this: not much
Click to expand...

That is exactly what i mean. I want more detail and focus, larger textures, more distance draw, and more polygons. I know thats what hurts performance, and there are many way in which shaders or bump mapping can keep the detail high and polygons low, but don't do it by simply blurring the scene. Fisrt person games are supossed to be from a persons perspecative (eye balls), not what life would look like if i was watcing it though a tv tube.
 
Superunknown98 said:
That is exactly what i mean. I want more detail and focus, larger textures, more distance draw, and more polygons. I know thats what hurts performance, and there are many way in which shaders or bump mapping can keep the detail high and polygons low, but don't do it by simply blurring the scene. Fisrt person games are supossed to be from a persons perspecative (eye balls), not what life would look like if i was watcing it though a tv tube.
Click to expand...

You do realize that in real life there is a ton of motion blur, right? It is through a "tv tube," actually, where there is an unrealistic absence of blur in most cases. Shift your head from side to side really quickly or run your hand up and down in front of your face and you'll see what I mean. There is actually quite a bit of blur because, in reality, our eyeballs don't have vertical sync! But instead of "tearing" an image, our brain interprets the image as being blurry because of the way we process images coming in.

In terms of larger textures, more polygons, etc... I believe that is being done too. It's not just special effects such as motion blur, which actually hurt perfromance very little...
 
I do realize that in real life, we do see blur, but it is not to the extent in which the valve demos were. The guy shooting at the window, was all blured, and he nor any other object was moving( as far as i can tell). as i said in my first post, I really don't see how blurring a scene so less detail is needed, so the game runs faster, is good. I also don't want my video games to look like movies where every thing is blurred ( becasue of the 24 frames a second). I want highly detailed objects, highly shaded textures, and motion only on objects the are moving fast. Unless the whole world is shaking, the rest of the scene should be mothin blur free. Depth of perception, is ok, but i don't know how it will work, or does it know where i'm looking, i dont always look straight forward.

When people watch a pixar movie what does everyone say? "ohh look at how real those characters are!" not "wow look at the realistic motion blur/ depth perception!" Polygon count and texture/shader detail is much more important in making a scene look realistic. Blur/ depth perception is something that will become important later, but right now I would rather be able to see a characters facial expressions, than see him wave his hands with motion blur.
 
You must log in or register to reply here.
Back
Top