brown@m.cs.uiuc.edu (11/10/89)
Written 1:03 pm Nov 8, 1989 by jrn@csd4.csd.uwm.edu in comp.graphics: >I changed the color palette to >all shades of grey. What I got was 4 sections of colors where each section >had the same colors as the all other sections. > >I did something like this > > for i:=0 to 255 do > SetRGBPalette(i,i,i,i); > >I would assume that I should have gotten 256 varying shades of grey. >But I didn't. I got 4 sections of 64 shades of grey, were each section >had the same 64 colors. The problem that you've run into is a limitation of the VGA hardware: the VGA digital to analog converters are only six bit devices. While you're specifying the desired value as an eight bit quantity, only the top six bits are used--the bottom two are ignored. Because of this, there are only 64 distinct shades of red, 64 of blue, and 64 of green available (64*64*64 = 262144 or the total number of colors in the VGA palette). >How would I get 256 DIFFERENT shades of grey? With the VGA, I'm afraid that you can't. William Brown brown@cs.uiuc.edu University of Illinois at Urbana-Champaign
kendalla@pooter.WV.TEK.COM (Kendall Auel;685-2425;61-028;;pooter) (11/10/89)
In article <824@uwm.edu> jrn@csd4.csd.uwm.edu (James Ray Norton) writes: > >I've got another question. > >I am using Turbo Pascal v5.0 and I did a test. I changed the color palette to >all shades of grey. What I got was 4 sections of colors where each section >had the same colors as the all other sections. > As mentioned earlier, the VGA has 6-bit DAC's so you cannot get 256 shades of gray from hardware. But you can get 256 shades of gray by using a simple 2x2 dither pattern, and I personally think the shading looks BETTER with dithering than with 8-bit DAC's. (When shading a region). For vector drawing, you are better staying away from dithering and just using the 64 available shades. Kendall Auel kendalla@pooter.wv.tek.com
mcdonald@uxe.cso.uiuc.edu (11/10/89)
>>How would I get 256 DIFFERENT shades of grey? > With the VGA, I'm afraid that you can't. Not for the hardware hacker. In fact it is easy. First you take a pair of scissors and cut your monitor cable in half ...... The rest is too long to post, but it can be done.
magore@watserv1.waterloo.edu (Mike Gore, Institute Computer Research - ICR) (11/11/89)
Hi, In article <824@uwm.edu> jrn@csd4.csd.uwm.edu (James Ray Norton) writes: [ re: VGA palette ] >I've got another question. ... > for i:=0 to 255 do > SetRGBPalette(i,i,i,i); > >I would assume that I should have gotten 256 varying shades of grey. >But I didn't. I got 4 sections of 64 shades of grey, were each section >had the same 64 colors. > >How would I get 256 DIFFERENT shades of grey? Although the DACs on each of THE R B G channels are 6 bits you can fudge the palettes to get a better aproximation the 64 with just a bit of color offset for some entries. If you keep each of the 64 palette entries for R G & B within +/-1 of each other and then consider the sum of R G & B as the actual output you can get 192 values out of 256. This works because near output values of 0 the colors resulting from the offsets are so dark as not to be very visible and for high values of output the color offset does not contribute much to the overall brightness. I have used this method with good results. For better results one should adapt the palette for the monitor being used since each channel of R G & B on most all monitors have slightly differing transfer functions for each channel anyway. In otherwords given a pure gray signal you will still likely get slight changes of color in the "grey" with just changes in brightness alone - tweeking the palette can avoid most of this. Good luck!, # Mike Gore, Technical Support, Institute for Computer Research # Internet: magore@watserv1.waterloo.edu or magore@watserv1.uwaterloo.ca # UUCP: uunet!watmath!watserv1!magore # These ideas/concepts do *not* imply views held by the University of Waterloo.
mfinegan@uceng.UC.EDU (michael k finegan) (11/11/89)
For those who wanted a grey scale palette on the VGA - here is
a combination of C and assembler that loads a LUT into the VGA
(256 color) palette. I used MASM and MSC v. 5.1, but should
work, with minor changes, for turboc, etc.
The same ideas work for 'superVGA' - just change the set_size_color()
routine to select your favorite mode and resolution. You could try
loading a modified palette, as someone suggested, with colors that
were close to grey - the dark greys are very similar.
Hope this is what you wanted ...
Mike Finegan
mfinegan@uceng.UC.EDU
------------------------ cut here for start of C file --------------------------
/* vgapals.c
*
* (c) 1989 Michael K. Finegan, Jr.
*
* Simplistic, and inefficient, way to display
* unsigned char image on 320x200 vga display,
* but illustrates use of LoadPals() with display
* of centered 200 x 256 image - a grey scale ramp.
*
* assuming masm 'loadpals.asm;' was previously executed:
* compiled using `cl /AL /c vgapals.c followed by
* `link vgademo+loadpals;'
*
* for small model - change all "far" to "near" in loadpals.asm ...
*/
#include <stdio.h>
#include <conio.h>
#include <dos.h>
#include <io.h>
#define FALSE 0
#define TRUE 1
#define SAVE_LOAD 1
#define RESTORE 0
/* used by LoadPals */
extern unsigned char far NewLut[], far OldLut[];
main()
{
int i, x, y;
void LoadPals(); /* LoadPals is large model (far return) assembler */
int GetChar(); /* also far return assembler ... */
void setvmode(), set_size_colors(), writepix();
/* Load palette with reduced (64 level) gray scale (R = G = B).
*
* Only the 6 least signifigant bits of R,G,B bytes are
* used for the output Look Up Table.
*
* NOTE: you could load your LUT (256*3 bytes), maybe
* previously saved to file, into NewLut, instead ...
*/
for(i=0;i<256;i++) {
NewLut[3*i + 0] = (unsigned char)(i/4); /* R */
NewLut[3*i + 1] = (unsigned char)(i/4); /* G */
NewLut[3*i + 2] = (unsigned char)(i/4); /* B */
}
setvmode(0x13); /* 0x13 <==> 256 color 320 by 200 VGA */
set_size_colors(320,200,256);
LoadPals(SAVE_LOAD);
/*
* draw a grey scale 'ramp'
*
* "x + 32" shifts 256 column image to center of screen.
*/
for(y=0;y<200;y++) {
for(x=0;x<256;x++)
writepix(x + 32,y,(unsigned char)x);
}
GetChar(); /* i.e. wait for keystroke before quitting ... */
LoadPals(RESTORE);
setvmode(0x3); /* 0x3 <==> text; should restore previous mode ... */
}
void set_size_colors(max_x,max_y,colors)
int max_x, max_y, colors;
{
union REGS inregs, outregs;
inregs.x.ax = 0x7e;
inregs.x.bx = (unsigned)max_x;
inregs.x.cx = (unsigned)max_y;
inregs.x.dx = (unsigned)colors;
int86(0x10,&inregs,&outregs);
}
void setvmode(mode)
int mode;
{
union REGS inregs, outregs;
inregs.h.ah = 0;
inregs.h.al = mode;
int86(0x10,&inregs,&outregs);
}
void writepix(x,y,value)
int x, y;
unsigned char value;
{
union REGS inregs, outregs;
/* simple video bios call, could write directly to memory ...
*
* image processing coordinate system - 0,0 @ upper left corner
*/
inregs.h.al = value;
inregs.h.ah = 0x0C; /* write dot */
inregs.h.bh = 0; /* assume page 0 */
inregs.x.cx = (unsigned)x;
inregs.x.dx = (unsigned)y;
int86(0x10,&inregs,&outregs);
}
------------------------ cut here for end of C file ----------------------------
------------------------ cut here for start of asm file ------------------------
TITLE 'BIOS load of palette'
NAME LoadPals
PAGE 55,132
_DATA SEGMENT
_NewLut db 800 dup (?)
_OldLut db 800 dup (?)
_DATA ENDS
_BSS SEGMENT
_BSS ENDS
CONST SEGMENT
CONST ENDS
DGROUP GROUP CONST,_BSS,_DATA
_TEXT SEGMENT byte public 'CODE'
ASSUME cs:_TEXT,ds:DGROUP,es:DGROUP,ss:DGROUP
PUBLIC _NewLut, _OldLut
; (c) 1989 Michael K. Finegan, Jr.
;
; LoadPals Load VGA 256 color palette
;
; Call with AX = [3,2,1,0] (for MSC, LoadPals(AX) )
; Where:
; 3 ==> Just Load pals
; 1 ==> Save current pals, then Load new pals
; 2 ==> Just Restore pals
; 0 ==> Save current pals, then Restore old pals
; Assume:
; NewLut filled with new LUT; OldLut holds saved LUT,
; depending on calling sequence.
;
; LoadPals(1) followed by LoadPals(2) leaves palette unchanged ...
;
PUBLIC _LoadPals
_LoadPals PROC far
test ax,1
jz restore
cmp ax,3
je load
mov ah,10h ; set palette reg.s/intensity/blink function
mov al,17h ; read a block of DAC color registers
mov bx,0 ; start with register 0
mov cx,100h ; and read all 256
push es
mov dx, seg _OldLut
mov es,dx
mov dx, offset _OldLut
push bp ; int 10h can modify
int 10h ; assume same data segment
pop bp
pop es
load:
mov ah,10h ; set palette reg.s/intensity/blink function
mov al,12h ; update a block of DAC color registers
mov bx,0 ; start with register 0
mov cx,100h ; and change all 256
push es
mov dx, seg _NewLut
mov es,dx
mov dx, offset _NewLut
push bp ; int 10h can modify
int 10h ; assume same data segment
pop bp
pop es
jmp finished
restore:
; *** check values written to registers to debug ***
cmp ax,2
je just_restore
mov ah,10h ; set palette reg.s/intensity/blink function
mov al,17h ; read a block of DAC color registers
mov bx,0 ; start with register 0
mov cx,100h ; and read all 256
push es
mov dx, seg _NewLut
mov es,dx
mov dx, offset _NewLut
push bp ; int 10h can modify
int 10h ; assume same data segment
pop bp
pop es
just_restore:
; restore the pal values read into _Outputlist by a previous
; call to _LoadPals(RESTORE)
mov ah,10h ; set palette reg.s/intensity/blink function
mov al,12h ; update a block of DAC color registers
mov bx,0 ; start with register 0
mov cx,100h ; and change all 256
push es
mov dx, seg _OldLut
mov es,dx
mov dx, offset _OldLut
push bp ; int 10h can modify
int 10h ; assume same data segment
pop bp
pop es
finished:
ret
_LoadPals ENDP
; GetChar reads char from keyboard, with no Cntrl-Brk
; check, no echo, and the <scan code,char> returned in <ah,al>.
PUBLIC _GetChar
_GetChar PROC far
mov ah,0
INT 16h
ret
_GetChar ENDP
_TEXT ENDS
END
------------------------ cut here for end of asm file --------------------------mjb@nucleus.UUCP (Mark Bobak) (11/11/89)
In article <824@uwm.edu> jrn@csd4.csd.uwm.edu (James Ray Norton) writes: > >I am using Turbo Pascal v5.0 and I did a test. I changed the color palette to >all shades of grey. What I got was 4 sections of colors where each section >had the same colors as the all other sections. > (Text deleted...) >How would I get 256 DIFFERENT shades of grey? To the best of my knowledge, the VGA will support only 64 shades of grey. -- Mark Bobak The Nucleus Public Access Unix, Clarkston, Mi. mjb@nucleus.mi.org mjb@m-net.ann-arbor.mi.us
U5569462@ucsvc.ucs.unimelb.edu.au (DAVID CLUNIE) (11/12/89)
> I would assume that I should have gotten 256 varying shades of grey. > But I didn't. I got 4 sections of 64 shades of grey, were each section > had the same 64 colors. > > How would I get 256 DIFFERENT shades of grey? As others have already pointed out you can't because of the 6 bit DAC's in the vga. Interestingly, someone mailed me a palette that aims to simulate more than 64 shades of grey by using colors very close to grey. I have tried it and really can't perceive the difference between it and just using a 6 bit grey scale. Which brings me to the point that I have read somewhere recently (can't remember where) that the human eye CAN'T distinguish any more than 64 different shades of grey. Is this so ? Do people beleive it ? Regards ... David
nn86302@tut.fi (Neuvo Niilo Anselmi) (11/14/89)
In article <391@ucsvc.ucs.unimelb.edu.au> U5569462@ucsvc.ucs.unimelb.edu.au (DAVID CLUNIE) writes: > >Which brings me to the point that I have read somewhere recently (can't >remember where) that the human eye CAN'T distinguish any more than 64 different >shades of grey. Is this so ? Do people beleive it ? > Nope. Nope. I'm not 100% certain about grey scales, but with other colors I can see the difference. Try drawing two squares (on a VGA(Ah forbidden word)) which are colored with two different colors very close(like colors (50,0,0) and (49,0,0)) to each other and you should be able to see a 'line' in the junction of the boxes. -- NN NN NN NN NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN NNNN N NN N NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN NNNNNN N NN N NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN NNNNNNNN NN NN NN NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
myers@hpfcdj.HP.COM (Bob Myers) (11/15/89)
>Which brings me to the point that I have read somewhere recently (can't >remember where) that the human eye CAN'T distinguish any more than 64 different >shades of grey. Is this so ? Do people beleive it ? As a first approximation, this probably isn't too far off for most people, at least within the confines of the average-sized computer display. One factor which affects the discrimination of the gray levels or colors is the area involved; for example, the human eye has much poorer spatial resolution for color than for monochrome images, meaning that you will be unable to distinguish very small areas which are of different colors but similar in apparent brightness. On the other hand, those same details will be apparent if they are rendered as differring in brightness. This fact is used in the NTSC color encoding process (and others), in which the color bandwidth is purposely limited relative to the luminance bandwidth. Bob Myers KC0EW HP Graphics Tech. Div.| Opinions expressed here are not Ft. Collins, Colorado | those of my employer or any other myers%hpfcla@hplabs.hp.com | sentient life-form on this planet.
brian@hpfcdj.HP.COM (Brian Rauchfuss) (11/15/89)
/ hpfcdj:comp.graphics / U5569462@ucsvc.ucs.unimelb.edu.au (DAVID CLUNIE) / 4:27 pm Nov 11, 1989 / >Which brings me to the point that I have read somewhere recently (can't >remember where) that the human eye CAN'T distinguish any more than 64 different >shades of grey. Is this so ? Do people beleive it ? > >Regards ... David >---------- Everyone around here can distinquish 256 shades of gray, but a dark room is required. 64 shades are very easy to distinquish. Brian Rauchfuss
fleming@balboa (Dennis Paul Fleming) (11/23/89)
In article <17400014@hpfcdj.HP.COM> brian@hpfcdj.HP.COM (Brian Rauchfuss) writes: > >Everyone around here can distinquish 256 shades of gray, but a dark room is >required. 64 shades are very easy to distinquish. > >Brian Rauchfuss Your ability to distinguish probably depends on two gray levels being neighbors. If they are seperated then one's ability is curtailed. There are two experements which point this out: 1) Grossberg in his book on Neural Nets looks at an image with the folowing greylevel pattern: |\ | \ | \ ------------------------\ | \----------------------- \ | \| \ The resulting perception is that the 2 grey patches on either side of the dicontinuity appear to be different grey levels. 2) McCann and Land in their paper on the Retinex theory Journal of Optical Scientist (?) had an experiment where light was cast onto a sheet of paper with two areas of different reflectance. The energy comming off of the paper was as follows: _____________________________ \ |\ | | | \ | \ | | | \ | \ / | | | \ | \ )- | | | \ | \ \ | 40% | 80% | \ | \ | reflectance | reflectance | \ | \ | | | \| \ |___________________________| The result of this experiment was that when a pencil or other obstruction was places at the dicontinuity both grey areas appeared the same value. because of the position of the light and the relative values the total energy coming from each grea patch was the same, and so percieved as such. Dennis Fleming Dept of Electrical and Computer Engineering University of California, Irvine