Many lens effects have come in to fashion in video games over the years, and some have faded out. Possibly the most commonly known lens effect in realtime rendering was the lens flare. If your early 2000s video game didn’t have lens flare it probably wasn’t a real game. They died out over the coming decade but other lens effects have started to see popularity in being implemented as well as sensor artifacts. A brief list of lens effects would probably include bloom, depth of field and chromatic aberration and a list of sensor artifacts would prominently include grain and motion blur. Each of these has seen inclusion in many games and there’s a lot of general distaste for these effects, but the one I’d like to cover in this article is motion blur.
What is Motion Blur?
Before going in to how motion blur is a bad idea, I think it’s important we cover what motion blur actually is. Motion blur is the result of light sensors in cameras and our eyes having an exposure time (it’s a little more complex with our eyes but the idea is roughly the same). Light is collected onto the sensor after focusing by the lens in order to create the image we observe. Because this process happens over a period of time, objects or the camera might move and this creates a smearing effect where the object may be on the left of the field of view at the start and on the right by the end of the exposure time.
For video game implementations, these two distinct types of motion blur are quite important too. Even though the exact cause of them is the same (relative movement of objects on screen), object and camera blur are quite different in terms of how they affect the perception of the scene for a player.
Why Camera Motion Blur is Bad
Firstly I’d like to cover camera motion blur, or as I also like to call it “Panning Blur”. This is the kind of blur caused by moving the camera and it is the most common type of motion blur you see in games, primarily because it’s very simple to do (though less simple to do without artifacting). The game simply needs to know the position/rotation of the camera on the current frame and the previous frame. Knowing this, we can calculate where a pixel is on the current and previous frame and then we just blur the image between these two points. This technical implementation is accurate enough to real life (if we assume a linear interpolation of position/rotation during that time period as well as ignoring obscurance problems) but unfortunately I believe regardless of this, it is a poor idea.
To explain why this is, I’d like to compare to two scenarios. Try looking around your room. You’ll probably notice two things. 1. Your eyes tend to skip around when looking. It’s actually extremely difficult to pan your eyes slowly between two positions without skipping. 2. You tend to blink when you do skip large distances. This helps your eyes adjust quickly to new focal lengths. Both of these observations mean that seeing panning motion blur in real life is actually pretty rare. You’ll probably only notice it if you really forcefully try not to flick your vision around. The second scenario would be looking at movies or other rendered content. Again, I would make the claim that smooth panning/rotating shots are quite rare. Movies much more often cut when they want to jump long distances and even when they don’t, a director has a much better idea of where the viewer is going to be focusing their vision at any point in time.
Now let’s look at video games. In video games ( especially console games controlled with a stick ), slow, linear panning camera motions are basically modus operadi. Combine this with camera motion blur and you have an image that is basically in a perpetual state of blur and herein lies the problem with the assumptions that camera motion blur makes. They are technically accurate if you assume the player has their eyes glued to the middle of the screen but you can’t make that assumption and even if you do it fails to mimic the quick glances you make with your eyes in real life.
Spyro may well be the absolute worst example of motion blur I’ve ever seen in a game. A 30fps game combined with quick, smooth camera motion with some of the longest streaks I’ve ever seen in a game. Most of the time you’re looking at it all you’re seeing is blur. God forbid the player want to see anything around them.
A counterpoint might be that it helps reduce the jerky nature of 30fps gameplay and all I can say to that is in my opinion it doesn’t. It still looks just as jerky to me, just with extra blur.
Object Motion Blur Is (typically) Good
Despite the aforementioned problems with camera motion blur, I think object based motion blur IS good because it tends to more accurately map to how we perceive the world, if used sparingly. A good example of this would be a game like the latest Spiderman game.
When characters move around so quickly, a lack of blur tends to make them look pretty jerky. Since the developer can be pretty sure the player isn’t staring directly at Spiderman’s moving body parts, but rather at Spiderman as a whole, you don’t get the breakdown of the effect caused by their being a mismatch between assumptions made by the game over the player’s gaze. What you’re left with is a character that can move around extremely quickly without jerking around and a much better sense of motion. It looks fantastic.
When Camera Blur Can Look Good
There are a couple of situations I think camera blur can be done appropriately. Racing games which use motion blur at the edge of the screen to give a better feeling of motion can make the assumption that the player is looking either at their car or shortly in front of them. In this situation you’re avoiding the mismatch I mentioned before as well as avoiding the problem of environments zooming past the player so fast that you lose a sense of momentum from scenery objects jumping. I also think in certain places where the camera might make a quick jump across an RTS map for example, camera motion blur can be appropriate and give a better sense of where the camera is actually moving.
I’d really like to see developers either all agree to include motion blur toggles in their game or recognise the inherent problems I’ve pointed out in this article. Of course it might help if games actually ran at 60fps more often, where these effects are much less notable….
Depth of field shares much of the same issues and I might write about that shortly to explain my views on that particular effect.
