Well, LAC is proposing an 8 bit storage format, and it's easy to show that 8 bits are not enough. you can try it out with a gradient in gimp and compare expansions to 8 and 16 bit. you can easily make out letterboxes in the former.
HOW OFTEN IS DA AVERAGE PERSON SITTIN AT DER COMPUTER PULLIN DER HAIR OUT BECOS AN IMAGE IS ENCODED WITH 8 BIT COMPONENTS RATHER THAN 16 BIT? BECOS DEY CAN MAKE OUT LETTERBOXES IN GRADIENTS? NOW DATS A RARE OCCURRENCE!
MAYBE DEY'RE PISSED OFF WITH COLOUR SPACES, BUT PROGRAMMERS R PROBABLY FUCKIN UP DA COLOUR TRANSLATION IN DA FIRST PLACE.
HOW MANY COLOUR SPACES DOES UR MONITOR OR GRAPHICS HARDWARE USE AT ANY GIVEN TIME? 1 FUCKIN COLOUR SPACE.
HOW MANY COLOUR SPACES DOES UR PRINTER USE AT ANY GIVEN TIME? 1 FUCKIN COLOUR SPACE.
NOW IF DEY BOTH USE DIFFERENT ONES, SAY RGB ON DA MONTIOR N CYMK ON DA PRINTER, DON'T TRANSLATE TO CYMK N TRY TO RENDER DAT SHIT ON UR FUCKIN MONITOR. PRINT OUT SHIT-LOADS OF SHEETS OF COLOUR IMAGES UNTIL DA RGB COLOUR SPACE RENDERED ON DA MONITOR MATCHES DA CORRESPONDING CYMK COLOUR SPACE RENDERED ON DA PAPER. K.I.S.S! IT ONLY GETS COMPLICATED WHEN U UNCONSCIOUSLY ADD A BUNCH OF STUPID FACTORS INTO DA PROBLEM.
We all know PBM and it's capability of storing images in a non-binary format, but it has drawbacks as well, e.g., it doesn't compress well.
FROM quotes.cat-v.org:
So much complexity in software comes from trying to make one thing do two things.
— Ryan Singer
DON'T DIP UR HAND INTO DAT BUCKET OF CRABS, U'LL GET NIPPED. EITHER UR SHOOTING FOR SIMPLICITY N A FILE FORMAT DAT'S EASY TO HANDLE, OR UR TRYING TO DO TWO THINGS DAT ARE IN DIRECT CONTRADICTION TO EACH OTHER N TRYING TO FIND A BALANCE ON DA TOP OF DA FUCKIN FENCE. PICK A SIDE OF DA FENCE N FUCKIN STICK WITH IT. STOP SHOOTIN FOR DA "PERFECT BALANCE". MANY HAVE TRIED, U HAVE A BUNCH OF STUPID FILE FORMATS AS A RESULT, DEY ALL SUCK, N SO DOES FARBFELD. "DO ONE THING N DO IT WELL."
EITHER IT COMPRESSES WELL OR IT REPRESENTS WELL. IF IT REPRESENTS WELL U NEED MINIMAL SOFTWARE TO READ N WRITE IT. IF IT COMPRESSES WELL U NEED MINIMAL RESOURCES (BANDWIDTH, DISK SPACE, ETC) TO PROCESS IT. PICK 1 N ONLY 1 ...
How often are you going to look at the image data by hand? How often are you going to generate images in the command line? This is a rare use-case
IT'S A RARE USE CASE BECOS IT'S IMPOSSIBLE WITH ALL DA IMAGE FORMATS DAT ARE OUT DER TODAY.
NONE OF DEM HAVE DIS CAPABILITY SO NOBODY DOES IT, YA FUCKIN RETOID. DAT IS WHY IT IS A RARE USE CASE, N IF U USE A FORMAT LIKE DA ONE I DESCRIBED, IT BECOMES AN OPTION, N USIN DESE UTILITIES BECOMES "FAIRLY DAM OFTEN" BECOS U DON"T HAVE TO CLUNK HEAVY DEPENDENCIES AROUND WHEN U WANT TO PROCESS A FILE. U GO FROM HAVING TO USE A PIECE OF SHIT LIBRARY IN EVERY LANGUAGE, TO USIN DA BASIC TEXT PROCESSING UTILITIES, EITHER THOSE DAT R AVAILABLE WITH DA BASIC TEXT PROCESSIN FUNCTIONS OF UR FAVOURITE PROGRAMMIN LANGUAGE (LIKE printf N scanf) OR DA RICH SET OF UTILITIES AVAILABLE WITH UNIX (fold, sort, uniq, grep, wc, sed, tr, awk, ...)
I don't know why everbody gets so obsessed about endianness to be honest.
SHO ME WHERE uint32_t IS IN DA STANDARD.
SHO ME WHERE uint16_t IS IN DA STANDARD.
SHO ME WHERE ntohl, ntohs, N htons ARE IN DA STANDARD.
YAINT RED DA FUCKIN STANDARD.
unsigned long width, height;
unsigned rgba[4];
unsigned char buf[256];
width = buf[8]*0x1000000LU + buf[9]*0x10000LU + buf[10]*0x100LU + buf[11];
height = buf[12]*0x1000000LU + buf[13]*0x10000LU + buf[14]*0x100LU + buf[15];
rgba[i] = buf[i*2]*0x100LU + buf[i*2 + 1];
buf[i*2 + 0] = rgba[i] >> 8;
buf[i*2 + 1] = rgba[i];WE DIDNT NEED DATA SERIALISATION DEPENDENCIES IN DA 80S N WE DONT NEED DEM NOW. UR SOFTWARE SUCKS JUST AS MUCH AS DA BSD socket SHIT. UR NO BETTER THAN DA HYENAS AT DA ISO WHO TURNED DA LANGUAGE DEANIS RICKY LOVED INTO DA PILE OF CRAP DAT IT IS TODAY.