swscale : treat float input data as uint 16bpc
authorMartin Vignali <martin.vignali@gmail.com>
Mon, 20 Aug 2018 13:28:00 +0000 (15:28 +0200)
committerMartin Vignali <martin.vignali@gmail.com>
Wed, 22 Aug 2018 09:36:09 +0000 (11:36 +0200)
commit3af1c4ea7d18c54738cd93f09b376d2228fdbce0
tree4500d2a78adc7554d76d337f517b8421338e95a0
parent900487043b6e531fe3edf8c8d38288ef915f6f25
swscale : treat float input data as uint 16bpc

Currently float are converted to 16b uint in input part
using src depth (32 bits) in hScale16To19 and hScale16to15,
make an invalid shift for the data

So shift the value when using float input
like 16 bpc uint.
libswscale/swscale.c