![[H]ardOCP](/images/blank.gif){kind=small}
- Date:
- Tuesday , December 13, 2005
- Author:
- Brent Justice
- Editor:
- Kyle Bennett
- Google +1
Valve Goes Cinematic
Valve is using cutting edge technologies to push the Source engine to new cinematic heights. We give you an overview of what Valve is developing to add depth to your Source engine gaming experience.
Introduction:
On December 8th, HardOCP had the pleasure of being invited to Valve Software in Bellevue, Washington. We weren’t told what exactly we would see, but we were told that it involved some new technology regarding the Source engine. To our surprise, hardware editors were invited to this event instead of gaming editors. It seemed that what Valve was going to show us had to deal with some new 3D effects in the Source engine that would soon be possible in real-time—given faster video cards.
Valve is located on the 10th floor of the Bank of America complex in downtown Bellevue, Washington. Their offices comprise the entire 10th level. As soon as you enter the reception area, you know you are among gamers. Everywhere you look there is some object or piece of memorabilia to remind you of the history of Valve. In fact, right in the center of the reception area is literally a giant, metal valve.
Valve is dedicated to further pushing the envelope with what can be done in the Source engine. This engine is incredibly flexible and allows Valve to push into new territory. Today we will examine what Valve has in store for gamers as they move into future titles. In a word, think Cinematics.
The Move Toward Cinematic Gameplay:
Valve literally wants to bring cinematic effects to games to increase realism. These effects can make the game feel more realistic and lifelike by adding greater depth to the experience. There are common film effects that movies have been using for some time to create a more immersive movie experience. Valve wants to bring these effects to games, and they are starting with Day of Defeat Source.
This is very interesting what Valve is talking about here because this is something you may not have thought of. In gaming graphics, the goal is to achieve a very detailed environment that immerses you in the game, but, at the same time, you must have enough performance to actually enjoy it. Well, what if you could render effects that hide some of the nasty imperfections game exhibit to help increase your immersion in that game? What if you could render fewer polygons and use an effect that still makes the player model appear like he/she is detailed? With the effects Valve is talking about today, this is the goal.
Valve wants to increase realism by implying that the scene is complex, but in reality it may not be. The goal is also to be able to do this in real time where these effects do not harm performance. Unfortunately, with two of the effects Valve is talking about today— which are Motion Blur and Depth of Field—they are not able to be rendered in real-time on current hardware. On the other hand, the other two effects Valve is talking about today, Color Correction and Film Grain and Dust, are able to be rendered in real-time and will be implemented in Day of Defeat in a Steam update.
Non Real-Time Effects
The current implementation of Valve’s Motion Blur and Depth of Field cannot be rendered in real-time on current hardware. In fact, we learned that the Day of Defeat demo video they used to present these new technologies took almost 2 seconds to render each frame with all the effects in use. The video was put together as a demonstration and was not being rendered in real-time, but it was a very good example of what a game could look like when all the effects are being used.
Valve is implementing Motion Blur and Depth of Field using an Accumulation Buffer. An Accumulation Buffer is nothing new really; it simply means that each frame can be stored for use in the frame using a different render target. Valve states that this method is better than approximations like Image-space depth of field and frame feedback or vector motion blur. In fact, this method of storing frames can most likely take advantage of multi-GPU acceleration very easily using Alternate Frame Rendering.
Motion Blur:
Motion Blur is not a new idea; in fact, it has already been implemented into some recent games. This effect is often so subtle in the game that you don’t even notice it, yet it can make the game feel much smoother to the eye. In a still screenshot, you can see this effect as it appears that in motion objects become blurred. In real time, however, this effect is very lifelike and imitates how the eye perceives motion and how cameras capture motion. Motion Blur improves the smoothness of the gameplay experience.
Depth of Field:
Depth of Field is also not a new idea. There are current approximations of it in use in some 3D demos; ATI’s Ruby demo and NVIDIA’s Toys demo, for example, use Image Space Depth of Field. Depth of field is a technique that can be used to accomplish two things. First, it can help draw your attention to specific points that the game content developer may want you to be focusing on in a game. For example, in the second screenshot above, you can see that no Depth of Field is in use and everything is in focus equally in the scene. In the third screenshot, however, you can see that the gun and bottom part of your screen have become blurry or out of focus. Before you might have been focusing on the gun itself, but now that it is out of focus, your attention is drawn to the center of the screen to the balcony. It just so happens that in this scene of the Day of Defeat video, an enemy soldier appears here. With this shift in focus, you are drawn to look at the balcony just as the soldier appears so you can shoot him.
Now, the problem with this is what if a soldier comes up along the right side of you which is out of focus? This is where balance must be drawn from the game content developer. There is a fine balance between what the game content developer wants you to see and be focused on and what you yourself want to focus on. In a single player game, we can see where this would be very beneficial to the story. But in a multiplayer game, we question how an effect like this could really be used properly for this result when you want to be focused on all aspects of the environment at once. For example, gamers will choose effects that help their gameplay. If they find that objects in the distance are blurry or they can’t see clearly from side to side, they might disable that feature so that they can see the entire world clearly to better their game. Therefore, a gamer in a multiplayer game may choose less realism in order to be the better multiplayer performer.
There is another aspect of Depth of Field that can be beneficial for realism. If you look again at the same screenshots above, this time look at the rock at the bottom right corner of your screen. With no depth of field effects you can see the very un-detailed texture and low polygon nature of this rock. However, with depth of field applied, the rock becomes blurry, which actually gives you the impression that it is more detailed if it were in focus. It draws your attention away from the bad looking rock and you simply don’t notice anymore how plain or low poly the rock is when it is blurred out. It’s like saying hey, this rock isn’t important to this moment in gameplay, so let’s blur it out so you are not focusing on it.
Both Motion Blur and Depth of Field:
Putting both Motion Blur and Depth of Field together can help improve the smoothness and image quality in the game. With both effects, you can hide things like mip-map boundaries, low polygon models, shader aliasing, texture crawling, moiré, and other game imperfections that crop up.