i mean yeah, its a honest loss of information, jpg on the other hand introduces compression artifacts that are basically hallucinations, meaning it pretends to have more data than it actually has and humans compensate for that thru image recognition and fantasy.
It’s really only helpful for formats that will be directly read by hardware (the video chip) and where the “compression” ratio (I would prefer the term quantization) needs to be fixed. For file compression, which was quite mature but CPU- and memory-intensive at the time, the dithering only makes it more difficult to compress further.
Compressed textures on modern GPUs actually use similar compression: a color palette followed by indexes into the palette. But that’s done per 4x4 pixel block.
I’m not really convinced that file compression was “mature” at the time. Text compression was reasonable progressed but image compression was created for a reason besides just a requirement for fixed compression ratio.
But I do agree that dithering was limited in it’s usefulness.
My point was just that dithering existed in print media, so it was one of the first ways they used to reduce image size, they just copied over the same technique
I’m thinking of file compression formats like Zip, LHA and ARJ, which would work particularly well if the image was not dithered and used run-length encoding (e.g. the PIC format of the Atari ST). The PNG format still uses the deflate algorithm which is essentially identical to the compression used by PKZip in 1991.
Back when image compression was just reducing the resolution and color depth.
i mean yeah, its a honest loss of information, jpg on the other hand introduces compression artifacts that are basically hallucinations, meaning it pretends to have more data than it actually has and humans compensate for that thru image recognition and fantasy.
I mean, that’s what dithering is
An artifact that your brain processes as something else.
It’s really only helpful for formats that will be directly read by hardware (the video chip) and where the “compression” ratio (I would prefer the term quantization) needs to be fixed. For file compression, which was quite mature but CPU- and memory-intensive at the time, the dithering only makes it more difficult to compress further.
Compressed textures on modern GPUs actually use similar compression: a color palette followed by indexes into the palette. But that’s done per 4x4 pixel block.
What do you mean “at the time”?
What time are you talking about?
At the time when dithering was commonly used to achieve the illusion of more available colors, i.e. the 80s and the first half of the 90s.
I’m not really convinced that file compression was “mature” at the time. Text compression was reasonable progressed but image compression was created for a reason besides just a requirement for fixed compression ratio.
But I do agree that dithering was limited in it’s usefulness.
My point was just that dithering existed in print media, so it was one of the first ways they used to reduce image size, they just copied over the same technique
I’m thinking of file compression formats like Zip, LHA and ARJ, which would work particularly well if the image was not dithered and used run-length encoding (e.g. the PIC format of the Atari ST). The PNG format still uses the deflate algorithm which is essentially identical to the compression used by PKZip in 1991.