Why does Anti-Aliasing slow down frame rates?
This is a simple. Anti-Aliasing is done by specific
calculations to choose which pixels are drawn additional & in what shade to
accomplish the desired result. The graphics card usually takes charge of this
task. Due to these additional calculations the graphics card has to give a
quantity of its processing power to finishing these calculations. This affects
the amount of processing power left to render the scene.
Usually the frame rates are affected by the amount of pixels
per second your graphics card can render, providing the CPU & other parts
can keep up. When you are using anti-aliasing the amount of pixels that your
graphics card can render will lower by a degree & therefore cut your frame
rates in games down.
What is FSAA?
FSAA stands for Full Scene Anti Aliasing & is used as
the most common term for Anti-Aliasing in the gaming world. FSAA refers to a
process of Anti-Aliasing that affects the whole screen not a certain picture. When
playing a game with FSAA turned on the graphics card will run an Anti-Aliasing
algorithm on every frame from top to bottom removing jagged edges from the
whole scene. This can have a giant hit on performance in games & has only
been overcome recently by having much more powerful graphics cards on the
market.
Types of Anti Aliasing
There are main types of Anti-Aliasing techniques,
Multisampling & Supersampling. There are varieties of each as graphics card
manufacturers find improved ways of getting the same result. Finding ways to
limit the performance hit is the main priority. Here they will look at the
general picture of the techniques. First they will concentrate on
Supersampling.
Supersampling
Supersampling is a process of Anti-Aliasing by taking the
corners of each pixel & generating what would be the average color. This is
then the displayed pixel on the screen. By doing this you are effectively
smudging the picture & averaging out the color along a curve.
This basic graphic shows you what would happen when
Supersampling is used in the squares in the center of this picture. Three of
the four squares are in both red & white areas & so will be displayed
as a shade of orange. One of the squares is fully in the white zone & so
that pixel will be displayed as a pure white pixel. If of the samples was taken
from an area that was full in the red zone then that pixel would be displayed
as fully red.
Obviously samples are much more complex than this, having
plenty of more colors to average out; however the principles stay the same. A
mathematical calculation is done to choose the average color for each pixel.
Effectively Supersampling renders the scene times larger than the true scene
& is then scaled down time the calculations are complete. This process has
a giant performance hit but does give the best results.
continued in part 3, stay tuned and subscribe to us for updates!
0 komentar:
Posting Komentar