@FioraAeterna @stephentyrone transforms *provided you don't put random rounding stages in the middle of the frickin' datapath*."
-
-
Replying to @rygorous
@FioraAeterna@stephentyrone Also it's *way* smaller, easier to design and in practice faster than having a 32-wide dedicated DCT unit.1 reply 0 retweets 0 likes -
Replying to @rygorous
@FioraAeterna@stephentyrone And then H.265 just said "here's a 32x32 matrix of 7-bit int coeffs. It has symmetries if you're into that."1 reply 0 retweets 1 like -
-
Replying to @rygorous
@rygorous@stephentyrone let's let everyone specify their own coefficients in each frame header just to be sure1 reply 0 retweets 0 likes -
Replying to @FioraAeterna
@FioraAeterna@stephentyrone ahh, that's what we in the 'biz call "The Vorbis".3 replies 0 retweets 1 like -
Replying to @rygorous
@rygorous@stephentyrone or we can specify a megabyte of tables in the spec as an appendix, then it's technically all static!1 reply 0 retweets 0 likes -
Replying to @FioraAeterna
@rygorous@stephentyrone I believe this is called "the twinvq"1 reply 0 retweets 0 likes -
Replying to @FioraAeterna
@FioraAeterna@stephentyrone also more recently employed by Brotli.2 replies 0 retweets 0 likes -
Replying to @rygorous
@rygorous@stephentyrone oh yes, the "let's make the decompressor big, and not count it in the filesize" school of compression algorithms2 replies 0 retweets 0 likes
@FioraAeterna @rygorous tbf the decompressor size is asymptotically 0% of the size, the more data you decompress.
-
-
Replying to @delroth_0 replies 0 retweets 0 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.