]> granicus.if.org Git - libvpx/commit
Fix test on maximum downscaling limits
authorAdrian Grange <agrange@google.com>
Wed, 18 Jun 2014 21:34:24 +0000 (14:34 -0700)
committerAdrian Grange <agrange@google.com>
Tue, 24 Jun 2014 17:26:09 +0000 (10:26 -0700)
commit8357292a5a9ff0cc09616c11c17cf4bf61503b3b
tree735122b7d88200e9646707e2b09cd1000fb9636b
parent4efcf838330a18a4b7e8530e52b5892c5223a5ba
Fix test on maximum downscaling limits

There is a normative scaling range of (x1/2, x16)
for VP9. This patch fixes the maximum downscaling
tests that are applied in the convolve function.

The code used a maximum downscaling limit of x1/5
for historic reasons related to the scalable
coding work. Since the downsampling in this
application is non-normative it will revert to
using a separate non-normative scaler.

Change-Id: Ide80ed712cee82fe5cb3c55076ac428295a6019f
vp9/common/vp9_convolve.c
vp9/encoder/vp9_encoder.c