assuming i modelled a 40 foot cargo container.
if i crop the width to 2.54 meters (or not - cause i have bits spare) and modelled everything at 1 cm precision, would it compress into a 8x11x10 or 9x11x10 binary format without distortion? i mean… it would make sense.
it’s a theoretical compression ratio of 33%.
also… if i’d model something smaller at the same precision. would it compress 3 dimensions into a 16 bit integer?
also… how much more instructions i would have to run to decompress this? there is binary data and binary exponent math and sign extension to convert this to a float3 vector on the gpu.
where’s the limit or balance when the computed math outweights the vertex density?
or… how fast can nvidia make this conversion?