manipulating bytes - GameCreators Forum

Author

Message

SoftMotion3D

AGK Developer

19

Years of Service

User Offline

Joined: 24th Aug 2005

Location: Calgary,Alberta

Posted: 22nd Nov 2016 04:28 Edited at: 22nd Nov 2016 04:47

Link

just what the title says!

i was doing some work to see if i could infact store more information inside a single byte.
as everyone knows... i hope... a byte can be in the range of 0-255

you can store allot in a single byte surprisingly.
you could have 8 on off switches
you could have two 0-15 valued numbers
you could have 4 0-3 valued numbers

and more combinations....

what id like to do is store map data for a scene in the smallest possible compact way to conserve space and memory.

after writting this i also realized that it would be easy to hide code from others in a way of bit shifting and swapping...lol

comments... let me know what you think

Attachments

Login to view attachments

Back to top

Profile PM Email Website

Markus

Valued Member

20

Years of Service

User Offline

Joined: 10th Apr 2004

Location: Germany

Posted: 22nd Nov 2016 07:20

Link

Quote: " map data"

which size?
in relationship to image data size i don't think you save memory, but your attitude is good.
i prefer real var names.
often code optimize end up in kludge and you lost overview.

AGK (Steam) V2.0.21 : Windows 10 Pro 64 Bit : AMD (16.6) Radeon R7 265 : Mac mini OS Sierra (10.12)

Back to top

Profile PM Email

BatVink

Moderator

21

Years of Service

User Offline

Joined: 4th Apr 2003

Location: Gods own County, UK

Posted: 22nd Nov 2016 08:10

Link

This is how Sprite groups work, so you would be extending what is already part of AGK.
Personally, I get confused with bit-shifting, but if you can get your head around it comfortably then it's a good way to go.
I don't worry too much about file sizes these days. Reducing 10K to 2K doesn't make much difference. But if you are reducing 10MB to 2 MB then it might be worth considering.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Quidquid latine dictum sit, altum sonatur
TutCity is being rebuilt

Back to top

Profile PM Email

CJB

Valued Member

20

Years of Service

User Offline

Joined: 10th Feb 2004

Location: Essex, UK

Posted: 22nd Nov 2016 16:46

Link

This was my whole idea of introducing Byte Code to the DB family of products... originally I had intended it to be used to make a download of a game (in code) as tiny as possible, with the use of a browser plugin "player" to decode and interpret the source, downloading media from remote locations as required. Bytecode also serves as an excellent way to obfuscate source code.

Win-win.

V2 T1 (Mostly)
Phone Tap!
Uzmadesign

Back to top

Profile PM Email Website

SoftMotion3D

AGK Developer

19

Years of Service

User Offline

Joined: 24th Aug 2005

Location: Calgary,Alberta

Posted: 22nd Nov 2016 17:19

Link

This could also be good for networking a game. You could pack more information in your sent messages keeping the chunks smaller.

Im going to do one more... extracting a byte from a short.

Also Processing midi sometimes has its data coded in 3 bytes of data. Extracting the number from 3 bytes is currently not possible with akg basic so i will add functions for this also.

Back to top

Profile PM Email Website

Jack

20

Years of Service

User Offline

Joined: 4th Oct 2004

Location: [Germany]

Posted: 24th Nov 2016 12:48

Link

Quote: "i was doing some work to see if i could infact store more information inside a single byte."

I am quite astonished by the fact that I work on nearly the same problems as you do, at the same time.

Have you found a way to store the information of 4 bytes into an integer? This borders me, because I want to store r,g,b,a values in it,
but I think this is not possible

This is for an online image transmitter.

Sorry for going a little bit off topic, but this is also a nice formula to play around, if you try to store information in strings
Tuppers formula
Tuppers Sandbox

[/url]

Back to top

Profile PM Email Website

damothegreat

User Banned

Posted: 24th Nov 2016 13:07 Edited at: 24th Nov 2016 13:13

Link

R - possible colours 0-255
G - possible colours 0-255
B - possible colours 0-255
A - possible transparency 0-255

= 4 single Bytes

Integer come in 4 bytes anyway

So maybe R = 255, G = 255, B = 0 , A = 255

So maybe could do:-

INT = R + (G * 255) + ( ( G * 255) + ( B * 0 ) ) + ( ( G * 255 ) + ( B * 0 ) + ( A * 255 ) )

or something along those lines

to convert back jyst reverse by minus and divides etc

Maybe completely wrong, but its all a learning curve from the top programming geniuses here

Damo

Using Tier 1 AppGameKit V2
Started coding with AMOS
Anything is possible if put mind to it

Back to top

Profile

damothegreat

User Banned

Posted: 24th Nov 2016 14:30

Link

Try this

// Project: RGB_to_INT 
// Created: 2016-11-24

// show all errors
SetErrorMode(2)

// set window properties
SetWindowTitle( "RGB_to_INT" )
SetWindowSize( 1024, 768, 0 )

// set display properties
SetVirtualResolution( 1024, 768 )
SetOrientationAllowed( 1, 1, 1, 1 )
SetSyncRate( 30, 0 ) // 30fps instead of 60 to save battery

RGBvalue as integer
red as integer
green as integer
blue as integer
alpha as integer

do
	
	red=251
	green=252
	blue=253
	alpha=254
	
	print (" seperate RGB values = R " + str(red) + " G " + str(green) + " B " + str(blue) + " A " + str(alpha))
	
	rgbvalue = (red << 24) + (green << 16) + (blue << 8) + alpha

//RGBvalue = red + (green * 255) + (( green * 255 ) + (blue * 255 )) + (( green * 255) + (blue * 255) + (alpha *255 ))

print (" INT = " + str(rgbvalue))
	
	
	newred = rgbvalue >> 24 && val("ff",16)
	newgreen = rgbvalue >> 16 && val("ff",16)
	newblue = rgbvalue >> 8  && val("ff",16)
	newalpha = rgbvalue && val("ff",16)
	
	//newred = (( alpha / 255 ) - (blue / 255) - (green / 255)) - ((blue / 255) - (green / 255)) - (green / 255)
	
	
	print ("convertred back " + str(newred) + " G " + str(newgreen) + " B " + str(newblue) + " A " + str(newalpha))

Print( ScreenFPS() )
    Sync()
loop

+ Code Snippet



// Project: RGB_to_INT 
// Created: 2016-11-24

// show all errors
SetErrorMode(2)

// set window properties
SetWindowTitle( "RGB_to_INT" )
SetWindowSize( 1024, 768, 0 )

// set display properties
SetVirtualResolution( 1024, 768 )
SetOrientationAllowed( 1, 1, 1, 1 )
SetSyncRate( 30, 0 ) // 30fps instead of 60 to save battery


RGBvalue as integer
red as integer
green as integer
blue as integer
alpha as integer

do
	
	red=251
	green=252
	blue=253
	alpha=254
	
	print (" seperate RGB values = R " + str(red) + " G " + str(green) + " B " + str(blue) + " A " + str(alpha))
	
	rgbvalue = (red << 24) + (green << 16) + (blue << 8) + alpha
	


	//RGBvalue = red + (green * 255) + (( green * 255 ) + (blue * 255 )) + (( green * 255) + (blue * 255) + (alpha *255 ))

	print (" INT = " + str(rgbvalue))
	
	
	newred = rgbvalue >> 24 && val("ff",16)
	newgreen = rgbvalue >> 16 && val("ff",16)
	newblue = rgbvalue >> 8  && val("ff",16)
	newalpha = rgbvalue && val("ff",16)
	
	//newred = (( alpha / 255 ) - (blue / 255) - (green / 255)) - ((blue / 255) - (green / 255)) - (green / 255)
	
	
	print ("convertred back " + str(newred) + " G " + str(newgreen) + " B " + str(newblue) + " A " + str(newalpha))

    Print( ScreenFPS() )
    Sync()
loop

Damo

Using Tier 1 AppGameKit V2
Started coding with AMOS
Anything is possible if put mind to it

Back to top

Profile

Jack

20

Years of Service

User Offline

Joined: 4th Oct 2004

Location: [Germany]

Posted: 24th Nov 2016 15:37

Link

@damo
Impressive! Thank you!

[/url]

Back to top

Profile PM Email Website

damothegreat

User Banned

Posted: 24th Nov 2016 15:40

Link

ur welcome

Using Tier 1 AppGameKit V2
Started coding with AMOS
Anything is possible if put mind to it

Back to top

Profile

Ortu

DBPro Master

17

Years of Service

User Offline

Joined: 21st Nov 2007

Location: Austin, TX

Posted: 24th Nov 2016 19:27 Edited at: 24th Nov 2016 20:30

Link

More bitwise fun:

+ Code Snippet

myNumber as integer
myNumber = setByte(0, myNumber, 42)
myNumber = setByte(1, myNumber, 128)
myNumber = setByte(2, myNumber, 255)
myNumber = setByte(3, myNumber, 1)


do
    print(getByte(0, myNumber))
    print(getByte(1, myNumber))
    print(getByte(2, myNumber))
    print(getByte(3, myNumber))
    Sync()
loop


function getByte(rByte as integer, rData as integer)
	byteVal = (rData >> (rByte * 8)) && 0xff
endfunction byteVal

function setByte(rByte as integer, rData as integer, rVal as integer)
	rData = rData || (rVal << (rByte * 8))
endfunction rData

A single player RPG featuring a branching, player driven storyline of meaningful choices and multiple endings alongside challenging active combat and intelligent AI.

Back to top

Profile PM Email Website

Kevin Picone

22

Years of Service

User Offline

Joined: 27th Aug 2002

Location: Australia

Posted: 24th Nov 2016 23:18

Link

Regarding ARGB colour packing and depacking, the math just boils down some shifts and masks, the same goes for pulling nibbles and bits etc

NOTE: mock up code

MyColour = rgb(255,50,100,200) ; Assuming colour levels are in A,R,G,B order.. They can be in in order but R,G,B,A is likely

A = ( MyColour >>24) and 255
R = ( MyColour >>16) and 255
G = ( MyColour >> 8) and 255
B = ( MyColour >>0) and 255 ; The shift here isn't needed here

Masking A would only be needed with the shift is arthimetic, which it shouldn't be in AGK

To roll your own RGB function we mask the inputs and then shift and OR into position, you can use addition also.

Function MyRGB(A,R,G,B)
NewColour = ((A and 255) << 24) or ((R and 255) << 16) or ((A and 255) << 8) or (A and 255 )
EndFunction NewColour

PlayBASIC To HTML5/WEB - Convert PlayBASIC To Machine Code

Back to top

Profile PM Website

SoftMotion3D

AGK Developer

19

Years of Service

User Offline

Joined: 24th Aug 2005

Location: Calgary,Alberta

Posted: 25th Nov 2016 00:30

Link

Ok so im building a new quarter float word.

It uses 1 bit for the sign and 2 bits for the quarter and the rest for the whole part of the number. This gives me a number range of -8191.75 to 8191.75

I think i can make adjustments to my 3d level and model editor to save this data and then to load back into agk as words instead of floats ... then reconstructing the model at run time. A normal float would be 4 bytes so this should cut my 3d scene and model data in half to 2 bytes rounding to the nearest quarter for the decimal place.

I will post with my new quarterfloat word function when completed.... if anyone interested in using it.

I know microsoft developed a half float using 16bits but i think i like the quarter approach better for the numbers i need.

Does anyone feel like taking a crack at coding a 16 bit float? Im no good with exponent math. What would that range be compared to mine?

Back to top

Profile PM Email Website

Kevin Picone

22

Years of Service

User Offline

Joined: 27th Aug 2002

Location: Australia

Posted: 25th Nov 2016 23:58

Link

you can try using Float to Fixed Point

PlayBASIC To HTML5/WEB - Convert PlayBASIC To Machine Code

Back to top

Profile PM Website

SoftMotion3D

AGK Developer

19

Years of Service

User Offline

Joined: 24th Aug 2005

Location: Calgary,Alberta

Posted: 26th Nov 2016 18:10 Edited at: 26th Nov 2016 18:13

Link

@kevin... I like that idea thanks!

ok so here is my next addition

a byte float. This number range is -1.27 to 1.27 . im hoping to use this to store 3d texture alignment mapping. It only has two decimal places but I think it should work with certain atlas texture sizes.

ill report back after I test it.

here are the functions

function setbytefloat(value#)
nval#=value#
if nval#=>1.270 then nval#=1.270
if nval#=<-1.270 then nval#=-1.270

if nval#<0.000 
	nvalue=128-(nval#*100.0)
else
	nvalue=(nval#*100)
endif
	
bit1=0
bit2=0
bit3=0
bit4=0
bit5=0
bit6=0
bit7=0
bit8=0
working=nvalue
//find current binary
test=working/128
if test=>1 then bit8=1
if bit8=1 then working=working-128
test=working/64
if test=>1 then bit7=1
if bit7=1 then working=working-64
test=working/32
if test=>1 then bit6=1
if bit6=1 then working=working-32
test=working/16
if test=>1 then bit5=1
if bit5=1 then working=working-16
test=working/8
if test=>1 then bit4=1
if bit4=1 then working=working-8
test=working/4
if test=>1 then bit3=1
if bit3=1 then working=working-4
test=working/2
if test=>1 then bit2=1
if bit2=1 then working=working-2
test=working/1
if test=>1 then bit1=1
if bit1=1 then working=working-1

newbyte=(bit8*128)+(bit7*64)+(bit6*32)+(bit5*16)+(bit4*8)+(bit3*4)+(bit2*2)+(bit1*1)

endfunction newbyte

function readbytefloat(byte)

bit1=0
bit2=0
bit3=0
bit4=0
bit5=0
bit6=0
bit7=0
bit8=0
working=byte
//find current binary
test=working/128
if test=>1 then bit8=1
if bit8=1 then working=working-128
test=working/64
if test=>1 then bit7=1
if bit7=1 then working=working-64
test=working/32
if test=>1 then bit6=1
if bit6=1 then working=working-32
test=working/16
if test=>1 then bit5=1
if bit5=1 then working=working-16
test=working/8
if test=>1 then bit4=1
if bit4=1 then working=working-8
test=working/4
if test=>1 then bit3=1
if bit3=1 then working=working-4
test=working/2
if test=>1 then bit2=1
if bit2=1 then working=working-2
test=working/1
if test=>1 then bit1=1
if bit1=1 then working=working-1
negative#=1.0
if bit8=1 then negative#=-1.0

newbyte#=((bit7*64)+(bit6*32)+(bit5*16)+(bit4*8)+(bit3*4)+(bit2*2)+(bit1*1))/100.00

nvalue#=negative#*newbyte#
endfunction nvalue#

+ Code Snippet

function setbytefloat(value#)
nval#=value#
if nval#=>1.270 then nval#=1.270
if nval#=<-1.270 then nval#=-1.270

if nval#<0.000 
	nvalue=128-(nval#*100.0)
else
	nvalue=(nval#*100)
endif
	
bit1=0
bit2=0
bit3=0
bit4=0
bit5=0
bit6=0
bit7=0
bit8=0
working=nvalue
//find current binary
test=working/128
if test=>1 then bit8=1
if bit8=1 then working=working-128
test=working/64
if test=>1 then bit7=1
if bit7=1 then working=working-64
test=working/32
if test=>1 then bit6=1
if bit6=1 then working=working-32
test=working/16
if test=>1 then bit5=1
if bit5=1 then working=working-16
test=working/8
if test=>1 then bit4=1
if bit4=1 then working=working-8
test=working/4
if test=>1 then bit3=1
if bit3=1 then working=working-4
test=working/2
if test=>1 then bit2=1
if bit2=1 then working=working-2
test=working/1
if test=>1 then bit1=1
if bit1=1 then working=working-1

newbyte=(bit8*128)+(bit7*64)+(bit6*32)+(bit5*16)+(bit4*8)+(bit3*4)+(bit2*2)+(bit1*1)

endfunction newbyte


function readbytefloat(byte)

bit1=0
bit2=0
bit3=0
bit4=0
bit5=0
bit6=0
bit7=0
bit8=0
working=byte
//find current binary
test=working/128
if test=>1 then bit8=1
if bit8=1 then working=working-128
test=working/64
if test=>1 then bit7=1
if bit7=1 then working=working-64
test=working/32
if test=>1 then bit6=1
if bit6=1 then working=working-32
test=working/16
if test=>1 then bit5=1
if bit5=1 then working=working-16
test=working/8
if test=>1 then bit4=1
if bit4=1 then working=working-8
test=working/4
if test=>1 then bit3=1
if bit3=1 then working=working-4
test=working/2
if test=>1 then bit2=1
if bit2=1 then working=working-2
test=working/1
if test=>1 then bit1=1
if bit1=1 then working=working-1
negative#=1.0
if bit8=1 then negative#=-1.0

newbyte#=((bit7*64)+(bit6*32)+(bit5*16)+(bit4*8)+(bit3*4)+(bit2*2)+(bit1*1))/100.00

nvalue#=negative#*newbyte#
endfunction nvalue#

Back to top

Profile PM Email Website

Kevin Picone

22

Years of Service

User Offline

Joined: 27th Aug 2002

Location: Australia

Posted: 27th Nov 2016 03:33

Link

if you've something in mind, then a look up table of 256 32bit values should do. So each byte is just indexes into a precalc table.

PlayBASIC To HTML5/WEB - Convert PlayBASIC To Machine Code

Back to top

Profile PM Website

SoftMotion3D

AGK Developer

19

Years of Service

User Offline

Joined: 24th Aug 2005

Location: Calgary,Alberta

Posted: 27th Nov 2016 19:41

Link

Yes that will work for sure!

If i have 256 possible atlas textures then i can calculate the offsets just from a single byte value.

Thanks again

Back to top

Profile PM Email Website

Sorry your browser is not supported!

AppGameKit Classic Chat / manipulating bytes

Attachments