intro to gràfics en temps real

8/18/2019 intro to Gràfics en Temps Real

http://slidepdf.com/reader/full/intro-to-grafics-en-temps-real 1/17

1

1

Course Introduction

Gustavo Patow

GGG - UdG

Those slides are based on the” Advanced Computer Graphics Advanced Computer Graphics” course by Tomas Akenine-Möller

2

Three major parts of thiscourse:

Real-Time Rendering

Performance is extremely important.

Really, real-time is soft

Try to increase the level of realism

Shadows

Reflections/refractions

Trees, clouds, water, crowds...



2

3

What are the results that are

achievable with

Computer Graphics?

4

State-of-the-ArtReal-Time Rendering2001



3

5

Which one is a real photo,and which one is CG?

6

Complexity: 8 milliontriangles…



4

7

Photo-realistic Rendering…

8

The Graphics Rendering Pipeline

The pipeline is the ”engine” that createsimages from 3D scenes

Three conceptual stages of the pipeline:

Application (executed on the CPU)

Geometry

Rasterizer

Application Geometry Rasterizer

3D

sceneinput

Image

output



5

9

Let’s take a closer look

The programmer ”sends” downprimitives to be rendered through thepipeline (using API calls)

The geometry stage does per-primitiveand per-vertex operations

The rasterizer stage does per-pixeloperations


10

Back to the pipeline:The APPLICATION stage

Executed on the CPU Means that the programmer decides what

happens here

Examples: Collision detection

Speed-up techniques

Animation

Most important task: send renderingprimitives (e.g. triangles) to the graphicshardware




6

11

Rendering Primitives

Use graphics hardware for real time…

These can render points, lines, triangles.

A surface is thus an approximation by anumber of such primitives.


12

You say that you render a”3D scene” , but what is it? First, of all to take a picture, it takes a camera – a

virtual one.

Decides what should end up in the final image

A 3D scene is: Geometry (triangles, lines, points, and more)

Light sources

Material properties of geometry

Textures (images to glue onto the geometry)

A triangle consists of 3 vertices

A vertex is 3D position, and mayinclude normals and more.




7

13

The GEOMETRY stage

Task: ”geometrical” operations on the inputdata (e.g. triangles)

Allows: Modify object geometry Move objects (matrix multiplication) Move the camera (matrix multiplication)

Compute lighting at vertices of triangle Project onto screen (3D to 2D) Clipping (avoid triangles outside screen) Map to window


14

The GEOMETRYstage in more detail

TheThe modelmodel transformtransform

Originally, an object is in ”model space”

Move, orient, and transform geometrical objects into

”world space” Example, a sphere is defined with origin at (0,0,0) with

radius 1

Translate, rotate, scale to make it appear elsewhere

Done per vertex with a 4x4 matrix multiplication!

The user can apply different matrices over time toanimate objects




8

15

GEOMETRYThe view transform

You can move the camera in the samemanner

But apply inverse transform to objects,so that camera looks down negative z-axis

z x


16

GEOMETRY Virtual Camera

Defined by position, direction vector, upvector, field of view, near and far plane.

pointdir

near far

fov

(angle)

Create image of geometry inside gray region

Used by OpenGL, DirectX, ray tracing, etc.




9

17

GEOMETRYProjection

Two major ways to do it

Orthogonal (useful in few applications)

Perspective (most often used)

Mimics how humans perceive the world, i.e.,objects’ apparent size decreases with distance


18

GEOMETRYProjection

Also done with a matrix multiplication!

Pinhole camera (left), analog used in

CG (right)




10

19

GEOMETRYLighting

Compute ”lighting” at vertices

Try to mimic how light in nature behaves Hardware uses empirical models, hacks, and some

real theory

Much more about this in later lecture


light

Geometry

blue

red green

Rasterizer

20

GEOMETRYClipping and Screen Mapping

Square (cube) after projection

Clip primitives to square


Screen mapping, scales and translates squareso that it ends up in a rendering window

These ”screen space coordinates” together with Z (depth) are sent to the rasterizer stage



11

21

GEOMETRYSummary


model space world space world space

compute lighting

camera space

projection

image space

clip map to screen

22

Animate objects and camera

Can animate in many different ways with 4x4matrices

Example:

Before displaying a torus on screen, a matrix thatrepresents a rotation can be applied

The result is that the torus is rotated

Same thing with camera (this is possible sincemotion is relative)




12

23

The RASTERIZER stage

Main task: take output from GEOMETRYand turn into visible pixels on screen


Also, add textures and various other per-pixeloperations

And visibility is resolved here: sorts theprimitives in the z-direction

24

The RASTERIZERin more detail

Scan-conversion Find out which pixels are inside the primitive

Texturing Put images on triangles

Interpolation over triangle

Z-buffering Make sure that what is visible from the camera

really is displayed

Double buffering

And more…




13

25

The RASTERIZERScan conversion

Triangle vertices from GEOMETRY is input

Find pixels/fragments inside the triangle

Or on a line, or on a point

Do per-pixel operations on thesefragments:

Interpolation

Texturing

Z-buffering

And more…


26

The RASTERIZERInterpolation

Interpolate colors over the triangle

Called Gouraud interpolation


blue

red green



14

27

The RASTERIZER Interpolation


u1

u2

u3

We want to evaluate f(u)

f(u)??

f(u1)

u1

u2

u3

f(u3) f(u2)

f(u)~ (f(u1)+f(u2)+f(u3))/3

f(u)~ f((u1+u2+u3)/3)

Gouraud Shading

Phong Shading

We know:

28

The RASTERIZERTexturing

One application of texturing is to ”glue” imagesonto geometrical object

Uses and other applications More realism

Bump mapping

Pseudo reflections

Store lighting

Almost infinitely many uses


+ =



15

29

The RASTERIZERZ-buffering

The graphics hardware is pretty stupid It ”just” draws triangles

However, a triangle that is covered by a moreclosely located triangle should not be visible

Assume two equally large tris at differentdepths


Triangle 1 Triangle 2 Draw 1 then 2

incorrect

Draw 2 then 1

correct

30

Would be nice to avoid sorting…

The Z-buffer (aka depth buffer) solves this

Idea:

Store z (depth) at each pixel When scan-converting a triangle, compute z at each

pixel on triangle

Compare triangle’s z to Z-buffer z-value

If triangle’s z is smaller, then replace Z-buffer andcolor buffer

Else do nothing

Can render in any order

The RASTERIZERZ-buffering




16

31

The RASTERIZERDouble buffering

The monitor displays one image at atime

So if we render the next image toscreen, then rendered primitives pop up

And even worse, we often clear thescreen before generating a new image

A better solution is ”double buffering”


32

Use two buffers: one front and oneback

The front buffer is displayed The back buffer is rendered to

When new image has been created inback buffer, swap front and back

The RASTERIZERDouble buffering




17

33

And, lately… Programmable shading has become a hot topic

Geometry & Vertex shaders

Pixel shaders

Point & Hull shaders

Adds more control and much more possibilities for theprogrammer


HARDWARE

Vertex

Shader

programs

Pixel Shader

program

Geometry

Shader

programs

Tessellation

programs

intro to gràfics en temps real

Documents