; STAR97.S (NASM version of STAR97.ASM) ; ; [ STARFIELD '97 ] ; ; [ Explanation ] ; ; well, here's an explanation of what's going on. first off we generate the ; starfeild data, star structures are qword aligned for easy offset calc's, ; and a floating random number seed is used. the initial random seed is mov'd ; into the very last star structure, and the next random seed is stored in ; the structure below, and this routine is followed to the very first star ; structure. instead using the modulus of the random number and using it as ; the saved value we mask off all the unwanted bits in the random number, and ; then modify the values to fit in the ranges we need. i was looking for a 4 ; layer field (00000011b), and 16 scales of white (00001111b). nifty both ; numbers are binary constants, lucky me! most things don't work out so well. ; ; [ Excuses ] ; ; ok, your saying "Big woop!, i've seen 20 byte starfields". well this is ; true, so have i. however all those starfields are just random color shit. ; not only is the randomness shitty, but they don't check for a keypress, and ; they don't even look like a starfield, otherwise there not usable at all! ; this routine has parallax style scrolling of a 4 layer starfield, 16 random ; shades of stars, a fairly random placement of stars, it's fast, and it even ; checks for vertical retrace and keypress! ; ; [ Questions / Answers ] ; ; you may be scratching your head on some of the sections of this code, so ; let me give you my reasoning before you call me a dumbshit. ; ; Q: why don't you use 'in ax,40h' as a random number routine rather than ; using imul and such? ; ; A: well 'in ax,40h' sucks, it's not random enough, try it. you'll see my ; reasoning, the method used is much better. btw i got the random routine ; i'm using by simplifying the random routine used by borland turbo c 3.0 ; i just disassembled the lib call and this is pretty much what i got. ; otherwise i had no idea how to make random numbers, this concept was new ; to me!, wow a learning experience. ; ; Q: why don't you use a 4 byte structure, a word for the address, and a byte ; for the speed, and another byte for the color? ; ; A: well in the main loop you'll see that i add the speed to the address, ; and that i store the address in di, well this means that if i wanted to ; add anything to a 16bit register i would have to use another 16bit mem ; location or register, and i can't otherwise it screws up. i wish there ; was some kind of addzx instruction which added the contents of a smaller ; register but internally masked it to the same size as the destination ; with zero's extended. ohh well there are too many instructions anyways. ; ; Q: only 255 stars? ; ; A: it looks ok doesn't it?!, you can use up to something like 8000 stars if ; you change the 'mov cl,NSTARS' to 'mov cx,NSTARS'. however the code size ; will increase two bytes (this was my origional goal 99 bytes), however i ; was told it would be "cooler" to release something that referenced the ; current year?? ; ; Q: my god your a moron, i could make this much smaller, and still retain ; all of the kewl features you talk about. ; ; A: great, please do, and send me your changes to brandt@europa.com. i tried ; to the best of my ability to make it smaller, i even got it to 87 bytes ; using the 'in ax,40h', and commenting out the random number seed setup, ; and the imul, along with the following mov. but like i said a couple of ; questions above 'in ax,40h' looks shitty. ; ; ohh, one last thing i do use the following 80386 instructions: ; ; shl reg16,immediate > 1 ; imul reg16,mem | reg16,reg16 | immediate ; ; deathlock ; brandt@europa.com ; NASM version: Erdogan Tan, 03/10/2016 NSTARS EQU 255 ; number of stars to draw [Bits 16] ; Real Mode (MSDOS) Program [org 100h] ; MSDOS COM File start: mov al,13h ; function 00h, mode 13h int 10h ; bios video interrupt mov dx,03DAh ; vga status port for retrace push 0A000h ; use es for segment overrides pop es ; when writing to the screen ;[ initialize starfield variables ]------------------------------------------ mov cl,NSTARS ; cx = number of stars to initialize mov byte [NSTARS*8+stars+8],1 ; set the begining random number seed loop1: mov bx,cx ; copy cx to bx for indexing shl bx,3 ; multiply bx by 8 (structure size) add bx,stars ; add the start offset of stars to bx imul ax,word [bx+8],4E35h ; multiply to get next random number mov word [bx],ax ; stars[cx].addr = ax and ax,0000001100001111b ; mask off unwanted values add al,14 ; make sure al < 30 & al > 14 mov word [bx+2],ax ; stars[cx].col = ax shr ax,8 ; shift hi byte to low byte inc al ; stars[cx].spd can't be zero mov word [bx+4],ax ; stars[cx].spd = ax loop loop1 ; do this again for the next star ;[ the main loop, holy shit! ]----------------------------------------------- drawstars: vend: in al,dx ; al = vga status byte (03DAh) and al,08h ; keep the vretrace status bit jz short vend ; not equal to 1?, check again mov cl,NSTARS ; lets draw NSTARS to the screen loop2: mov bx,cx ; copy cx to bx for indexing again shl bx,3 ; multiply bx by 8 (structure size) add bx,stars ; add start offset of stars to bx mov di,word [bx] ; di = stars.addr mov [es:di],ah ; erase star (ah isn't ever used) add di,word [bx+4] ; stars.addr = stars.addr + stars.spd mov al,byte [bx+2] ; al = stars.col mov [es:di],al ; draw new star to the screen mov word [bx],di ; save the updated star posistion loop loop2 ; erase/draw all NSTARS to the screen in al,60h ; al = port 60h, keyboard cmp al,1 ; escape key pressed? jnz short drawstars ; was escape pressed, no do it again mov al,03h ; escape was pressed, lets quit int 10h ; al = 03h text mode, int 10 video retn ; com file trick to exit to dos stars: ; start of starfield structure