1. Graphics Dr. Dimitrios S. Nikolopoulos CSL/UIUC
Lecture 14
Graphics
Dr. Dimitrios S. Nikolopoulos
CSL/UIUC

2. Outline Color theory Video Hardware Mode 13h Graphics File Formats
ECE291 Lecture 13

3. Color theory
TV’s, Computer Monitors, Projectors… It’s all Red, Green, Blue
Why RGB? These are the primary colors for visible light. You can get all the colors by mixing them.
Red + Green = Yellow
Green + Blue = Cyan
Blue + Red = Purple
ECE291 Lecture 13

4. Color Theory - Biology
How can mixing a Low Frequency (red) with High Frequency (blue) make Really High Frequency (purple) ?
It can’t according to the principles of signal processing. It is a biological phenomenon
Our eyes have RGB sensors. These are the frequencies we can see best
The people who invented color photography and color TV used this fact.
ECE291 Lecture 13

5. Color Theory - Biology
If three-eyed aliens with different color receptors visited us, they’d “see” the same three frequencies coming from the monitor, but they wouldn’t be able to see any pictures.
Most SciFi movies forget that…
The same goes for your dog that wonders why you spend hours looking at a blank screen.
ECE291 Lecture 13

6. Full Color Images Each pixel has a RGB value.
Color depth is defined by the number of bits used to assign colors to pixels
Full color is when we use a given number of bits to specify the intensity of each color component.
24-bit color means 1 R byte, 1 G byte, 1 B byte.
Each determines the color intensity (0 = none, 255 = full)
16-bit color means 5 bits red, 5 bits green, 5 bits blue. The extra bit is system-dependent.
ECE291 Lecture 13

7. 24-bit Full Color Images
24-bit color allows 256 different intensities for each color component. This allows for 224 or 16,777,216 different colors.
That’s way more than our eyes can distinguish.
You don’t need 16M colors to enjoy all this beauty in the world no matter what the produces of consumer electronics tell you…
This 24-bit color is sometimes called Full or True color.
True color requires 3 bytes for every pixel.
ECE291 Lecture 13

8. 16-bit Full Color Images
16-bit color allows 32 different intensities for each color component. This allows for 215 or 32,768 different colors (assuming 5 bits green)
Not quite as good as True color, 16-bit color is also referred to as High color.
High color requires 2 bytes for every pixel.
ECE291 Lecture 13

9. Full Color Images Positives of high and true color images
More colors means higher quality pictures
Easier to do image manipulation (math)
Make everything a little redder
Blur an image by averaging RGB values of adjacent pixels
Negatives of high and true color images
Take up a lot more memory
Ex: A 320x200 pixel image with 24-bit color needs
320 x 200 x 3 = 192,000 bytes of memory
ECE291 Lecture 13

10. Color Palettes
Unfortunately we’re still in real mode (1 Megabyte of memory no matter what…)
While we’d like to do everything in full color, in real mode we need to reduce the amount of memory our images use.
The color palette contains the 3-byte RGB values of 256 different colors.
The image data is then just a bunch of indexes into this palette. It’s a look-up table!
ECE291 Lecture 13

11. Color Palettes Positives of palettes
Less memory and bandwidth needed to store and move pictures
You can do some cool stuff with palettes
Invisible pixels – assign one palette entry to be “clear” and when you’re copying one image over another, don’t copy the “clear” pixels.
You can cycle the palette for 60’s psychedelic effects.
You can fade out a screen just by subtracting from each entry in the palette table until they’re all zero
ECE291 Lecture 13

12. Color Palettes Negatives of palettes
Fewer colors (at most 256) mean lower quality pictures
Harder to do manipulation
Average blurring is harder because new color isn’t in palette
All “real” pictures come in full color and have to be converted
This is hard because you have to choose a good palette
For each “full color” pixel, find closest palette entry
When you put images together, they share the same palette
ECE291 Lecture 13

13. Examples of using Palettes
8-bit (256-color) bitmap image (BMP)
Each file has a specific format
56 bytes header
256*4 bytes color palette
Each color has 3 RGB bytes and one clearness or unused byte
640*480*1 bytes of data
308,280 bytes
4-bit (16-color) bitmap image (BMP)
54 byte header
16*4 bytes color palette
640*480*0.5 bytes image data
153,720 bytes
ECE291 Lecture 13

14. Text Mode Video Revisited
Text mode video is like a form of palette
Each ASCII byte is an index into a font table with a bitmap for each character
Each Mode Byte is an index into a type of palette
Or you can think of the mode byte as 3-bit full color (kind of)
ECE291 Lecture 13

15. Video Hardware You’ve got a chunk of memory
You write to the memory through memory mapped I/O (MP3)
In a analog RGB monitor, three electron beams sweep across all rows 60 to 85 times a second to zap the phosphor elements to make them glow
The video card adjusts the intensity of the three electron beams in real time as they are sweeping to correspond to the intensities specified in the VRAM
ECE291 Lecture 13

16. Video Hardware
To do this, the video card reads out a whole row of RGB values in the VRAM and shifts them “out” in a big shift register.
The byte that “falls off the edge” gets converted into an intensity for the electron beam.
With a palette, there is an additional step. Before data shifts out, it looks up the “real” RGB value in palette memory.
Take a look at Windows->Control Panel->Display
ECE291 Lecture 13

17. Mode 13h In mode 13h you draw graphics using palettes
Mode 13h is a 320 x 200 x 256-color graphics display mode that was very popular for DOS programming due to its relatively high number of colors (256 vs 16 for other VGA modes) and simple addressing.
320 x 200 = pixels. With 1 byte per pixel this easily fits in a single real-mode segment.
It’s linear just like text mode. The pixels increase in memory as you go from left to right, while each row immediately follows the previous one.
Offset = row*320 + col
ECE291 Lecture 13

18. Mode 13h - How to get there
The following instructions get you into Mode 13h video
mov ah, 0 ; int 10h’s set mode subfunction
mov al, 13h ; The aptly named Mode 13h
int 10h ; Call the Video BIOS interrupt
ECE291 Lecture 13

19. Mode 13h How to draw a pixel
The following instructions draw a pixel on the screen
VidGrSeg equ 0A000h
Location dw (200/2)*320+(320/2) ; Center
Color db 98 ; Some random palette index
mov ax, VidGrSeg
mov es, ax ; Use extra segment to address VRAM
mov bx, [Location] ; Offset to center of screen
mov al, [Color] ; Palette Entry Number
mov [es:bx], al ; Write it to the screen
ECE291 Lecture 13

20. Default Palette Entries 0-15 are RGB colors [0,0,0,0,I,R,G,B]
Entries grayscale [0,0,0,1,g3,g2,g1,g0]
Entries contain user-defined colors
ECE291 Lecture 13

21. Mode 13h How to change the palette
Method 1: Load an entire palette
Write 0 to port 3C7h
Write red, green, blue bytes to port 3C9h for all 256 colors
Method 2: Change just one palette entry using VBIOS call
BX = Palette Number (0 – 255)
CH = Green (0 – 63)
CL = Blue (0 – 63)
DH = Red (0 – 63)
AX = 1010h ; Set VGA Palette Registers command
INT 10h ; vBIOS call
ECE291 Lecture 13

22. Mode 13h How to change the palette
; Change a palette entry to light purple
Red db 39
Green db 0
Blue db 63
Mov bx, 0 ; changing palette entry 0
Mov ch, [Green]
Mov cl, [Blue]
Mov dh, [Red]
Mov ax, 1010h
Int 10h
ECE291 Lecture 13

23. Notes on graphics modes
Mode 13h is outdated but remained popular due to its advantages (standardized, linear accessing, does not consume a lot of memory etc.)
Some efforts to standardize more high-resolution graphics did not succeed (VESA, DirectX)
13h appears to be your only option in real-mode for graphics
ECE291 Lecture 13

24. Notes on graphics modes
In protected mode you will be able to work with high-resolution graphics
Although it would be possible to use one of the standards (VESA) it is much more convenient to use the ECE291 libraries (pmodelib)
Many advantages
Designed to work with protected mode
Support for true color resolutions available on the video card
Linear addressing, just like what you’ve learned in text-mode video and mode 13h
ECE291 Lecture 13

25. Graphics File Formats Raster Formats
Arrays of pixels (what we’ve been talking about)
Raw pixels (.bmp)
Run-Length encoding (.pcx)
LZW encoding (.gif)
Transform encoding (.jpg)
Deflate (zip) compression (.png)
ECE291 Lecture 13

26. Graphics File Formats Vector Formats Like PostScript or CorelDraw
A vector file might have something like:
“Draw line from (24, 56) to (87, 12)”
“Here’s what an ‘A’ looks like….”
“Not draw an ‘A’ at (75, 96)”
Eventually converted to raster format by using a rasterizer
Printers are very high resolution rasterizers (for example)
ECE291 Lecture 13

27. Graphics File Formats Included in image files…
Header: Most file formats start out with some kind of info about the image. For graphics files this info could be:
Size (horizontal and vertical pixels)
Color depth (bits per pixel)
Version (makes life difficult…GIF89A, GIF97A, …)
Compression Technique
Palette: If you’re not using true color, you must include a palette for the image in the image file
Data: Raw or compressed, raster or vector information.
ECE291 Lecture 13

28. BMP 16 colors
Header 118 bytes (54 for attributes, 64 for colors,4 per color)
Example: 32x32 pixel image
118 bytes header
32x32x0.5 bytes per pixel=512 bytes data
Image data goes backwards from bottom to top
ECE291 Lecture 13

29. GIF and PNG GIF uses a compression algorithm patented by UNISYS
If a program uses GIF compression it must get a license from UNISYS
PNG requires no license, it uses the not patented ZIP compression format
ECE291 Lecture 13

30. PCX Originally used for the PC paintbrush Run-length encoding
Idea: Exploit the fact that many pictures have blobs of the same color
We can encode by storing the color and how many times it is repeated on the screen (run length)
Algorithm
Read a byte
If its two h.o bits are set it’s a length byte, read the length
Next byte is a color byte so copy it length times to the screen
Else, it is a color byte, copy it to the screen
ECE291 Lecture 13

31. PCX problems
What if you have a picture with a single color which starts with a 11 (like the length byte)
You have to use the length byte (length=1) and then the color code
So you actually occupy more room than needed
Solution, put the colors that tend to be used as single colors towards the beginning of the palette in memory, so that their h.o. bits are 00
Read the lab manual
ECE291 Lecture 13