0% found this document useful (0 votes)
6 views52 pages

Computer Graphics: Rasterisation & Quantisation

Uploaded by

Mhm
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
6 views52 pages

Computer Graphics: Rasterisation & Quantisation

Uploaded by

Mhm
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Computer Graphics

WBCS019-05, 2025/26

Rasterisation & Quantisation

Jiřı́ Kosinka [Link]@[Link]


What is Computer Graphics?

Scene Scene
capture design

Scene Modelling
description

Computer
Rendering
graphics

Digital display
camera
Image image Image
capture display

Animation

Video

2/52
What is Computer Graphics?
3D geometry,
lights,
textures

Scene
description

Computer Rasterisation Raytracing


graphics (OpenGL, DirectX)

Digital
image

Coloured
pixels

3/52
What is Computer Graphics?
3D geometry,
lights,
textures

Scene
description

Computer Rasterisation Raytracing


graphics (OpenGL, DirectX)

Digital
image

Coloured
pixels

4/52
Digital Image
• Vector images • Raster (bitmap) images
• Described by ‘vectors’ • Array of picture elements, pixels
• Rely on (smooth) geometry primitives • Rely on rasterisation
• Typical example: fonts • Typical example: .jpg, .bmp
• Resolution independent • Fixed resolution
• Need to be rasterised for display • Need to be resampled for display

In this course: image = digital image = raster image


5/52
Digital Image
• Vector images • Raster (bitmap) images
• Described by ‘vectors’ • Array of picture elements, pixels
• Rely on (smooth) geometry primitives • Rely on rasterisation
• Typical example: fonts • Typical example: .jpg, .bmp
• Resolution independent • Fixed resolution
• Need to be rasterised for display • Need to be resampled for display

In this course: image = digital image = raster image


6/52
How Many Pixels Do We Need?

1×1

7/52
How Many Pixels Do We Need?

2×2

8/52
How Many Pixels Do We Need?

4×4

9/52
How Many Pixels Do We Need?

8×8

10/52
How Many Pixels Do We Need?

16 × 16

11/52
How Many Pixels Do We Need?

16 × 16

12/52
How Many Pixels Do We Need?

32 × 32

13/52
How Many Pixels Do We Need?

32 × 32

14/52
How Many Pixels Do We Need?

64 × 64

15/52
How Many Pixels Do We Need?

128 × 128

16/52
How Many Pixels Do We Need?

256 × 256

17/52
How Many Pixels Do We Need?

512 × 512

18/52
How Many Pixels Do We Need?

1024 × 1024

19/52
Resolution
• Horizontal × vertical; in pixels (or dots)

• Image resolution

• Display resolution (← image resampling); LCD, (O)LED, . . .

• HD: 1366 × 768


• FHD: 1920 × 1080

• Size vs. density

• Number of dots/pixels per a unit of length


• DPI: dots per inch; binary devices (ink-jet printers)
• PPI: pixels per inch

20/52
Resolution
• Horizontal × vertical; in pixels (or dots)

• Image resolution

• Display resolution (← image resampling); LCD, (O)LED, . . .

• HD: 1366 × 768


• FHD: 1920 × 1080

• Size vs. density

• Number of dots/pixels per unit of length


• DPI: dots per inch; binary devices (printers)
• PPI: pixels per inch

21/52
What’s Your Phone’s PPI?
• Find out your phone’s PPI
• You may already know it.
• If not, compute it.
• If you are too lazy (= don’t know the maths;-), go here: [Link]
and enter resolution and diagonal.
• If you don’t know these,
go to your phone’s specs.

• and share your result.

22/52
Visual Acuity

Brightspace → Content → Additional Items → Test Image


[Link]/~jiri/[Link]

23/52
Visual Acuity
tan α2 = w
2d
25.4
w= PPI ; w in mm

d pixel
eye α
2
w
2
α =?

24/52
Visual Acuity
tan α2 = w
2d
25.4
w= PPI ; w in mm

d pixel
eye α
2
w
2
α =?
my old phone: 256PPI; 36cm

25/52
PPI and Viewing Distance

• Assuming average eyesight (20/20 vision): α = one arcminute, i.e.,


α = 1′ = 1/60 degrees, we obtain d = 8732
PPI ; d in cm.

d in cm 44 35 29 25 22 19 17

PPI 200 250 300 350 400 450 500

26/52
PPI and Viewing Distance

• Assuming average eyesight (20/20 vision): α = one arcminute, i.e.,


α = 1′ = 1/60 degrees, we obtain d = 8732
PPI ; d in cm.

d in cm 44 35 29 25 22 19 17

PPI 200 250 300 350 400 450 500

• Based on our simplifying assumptions (but actually rather accurate in this range).

27/52
Field of View
• Field of view (FOV), or field of vision
• Total: ∼ 200 degrees
• Binocular: ∼ 124 degrees
• Comfortable: ∼ 40 degrees

28/52
Homework
• Compute the ultimate display resolution
• You will need to make some assumptions (distance, eyesight, ...)

• Is Retina display ‘good’ enough?


• If so, what for?
• If not, why?

29/52
Rasterisation & Quantisation
• Raster image: map from a 2D array of pixels to some colour space (e.g. RGB)

• Quantisation: only discrete values are allowed

• But how many values do we need?

30/52
How many colours do we need?

31/52
How many colours do we need?

32/52
How many colours do we need?

33/52
How many colours do we need?

34/52
How many colours do we need?

16

35/52
How many colours do we need?

32

36/52
How many colours do we need?

64

37/52
How many colours do we need?

128

38/52
How many colours do we need?

256

39/52
How many colours do we need?

8 bits per channel; RGB; 24 bits per pixel

40/52
How many colours do we need?

64 colours 64 colours

41/52
How many colours do we need?

8 colours 8 colours

42/52
How many colours do we need?

8 colours 8 colours
‘best fitting’ palette RGB cube corner colours
(1 bit per channel)

43/52
Palette Selection

blue channel removed palette computation


Wikipedia Photoshop

44/52
Bit Depth
• Intensity
• The eye can adjust over a range of 107 : 1 (or even 1010 : 1)
• At any one time, ‘only’ ∼ 100 : 1
• Displays offer contrast ratios in the thousands (LCD) or even higher (LED)
• Humans distinguish ∼ 2% change in intensity ⇒ 8 bits typically sufficient
• Three channels ⇒ 256 · 256 · 256 = 16, 777, 216 colours
• The human eye can distinguish about 10, 000, 000 colours

45/52
Bit Depth
• Intensity
• The eye can adjust over a range of 107 : 1 (or even 1010 : 1)
• At any one time, ‘only’ ∼ 100 : 1
• Displays offer contrast ratios in the thousands (LCD) or even higher (LED)
• Humans distinguish ∼ 2% change in intensity ⇒ 8 bits typically sufficient
• Three channels ⇒ 256 · 256 · 256 = 16, 777, 216
• The human eye can distinguish about 10, 000, 000 colours

46/52
Bit/Colour Depth
• 8-bit fixed-range RGB
• Web and email applications
• Consumer photographs

• 16-bit fixed-range grey-scale


• Medical imaging

• 16-bit fixed-range RGB


• Professional photography and printing
• Image editing

• 32-bit floating-point RGB


• HDR: high dynamic range images

47/52
Summary
• Computers/displays/printers are digital ⇒ we need to

&

rasterise space quantise colour,


• but so that we get away with it;-)
• Disclaimer: no analogue cats were rasterised or quantised during the making of this presentation

48/52
Summary
• Computers/displays/printers are digital ⇒ we need to

&

rasterise space quantise colour,


• but so that we get away with it;-)
• Disclaimer: no analogue cats were rasterised or quantised during the making of this presentation.

49/52
Summary
• Computers/displays/printers are digital ⇒ we need to

&

rasterise space quantise colour,


• but so that we get away with it;-)
• Disclaimer: no analogue cats were rasterised or quantised during the making of this presentation.

50/52
Feedback via Poll Everywhere
• Navigate to [Link]/jirikosinka874

• No need to use real name (but be nice;-)

• What did you like?

• What can be improved?

• What should be covered in tutorials?

• Anything else?

51/52
Computer Graphics
WBCS019-05, 2025/26

Rasterisation & Quantisation

Jiřı́ Kosinka [Link]@[Link]

You might also like