]> granicus.if.org Git - libvpx/commit
hybrid transform of 16x16 dimension
authorJingning Han <jingning@google.com>
Wed, 29 Aug 2012 18:25:38 +0000 (11:25 -0700)
committerJingning Han <jingning@google.com>
Thu, 30 Aug 2012 23:52:25 +0000 (16:52 -0700)
commitde6dfa6bb0cd680ca446a15b52b2a026859eb1e6
tree50bc09ddb7d72155a8a37079cf33c25fdf4a60bc
parentc59e36fc76ce3c5dd0b9a40e8f95a225248ce018
hybrid transform of 16x16 dimension

Enable ADST/DCT of dimension 16x16 for I16X16 modes. This change provides
benefits mostly for hd sequences.

Set up the framework for selectable transform dimension.

Also allowing quantization parameter threshold to control the use
of hybrid transform (This is currently disabled by setting threshold
always above the quantization parameter. Adaptive thresholding can
be built upon this, which will further improve the coding performance.)

The coding performance gains (with respect to the codec that has all
other configuration settings turned on) are

derf:   0.013
yt:     0.086
hd:     0.198
std-hd: 0.501

Change-Id: Ibb4263a61fc74e0b3c345f54d73e8c73552bf926
38 files changed:
configure
vp8/common/blockd.h
vp8/common/coefupdateprobs.h
vp8/common/default_coef_probs.h
vp8/common/entropy.c
vp8/common/entropy.h
vp8/common/generic/systemdependent.c
vp8/common/idct.h
vp8/common/idctllm.c
vp8/common/invtrans.c
vp8/common/invtrans.h
vp8/common/loopfilter.c
vp8/common/onyxc_int.h
vp8/common/reconinter.c
vp8/decoder/decodframe.c
vp8/decoder/dequantize.c
vp8/decoder/dequantize.h
vp8/decoder/detokenize.c
vp8/decoder/detokenize.h
vp8/decoder/generic/dsystemdependent.c
vp8/decoder/onyxd_int.h
vp8/encoder/bitstream.c
vp8/encoder/block.h
vp8/encoder/dct.c
vp8/encoder/dct.h
vp8/encoder/encodeframe.c
vp8/encoder/encodeintra.c
vp8/encoder/encodemb.c
vp8/encoder/encodemb.h
vp8/encoder/generic/csystemdependent.c
vp8/encoder/onyx_if.c
vp8/encoder/onyx_int.h
vp8/encoder/quantize.c
vp8/encoder/quantize.h
vp8/encoder/ratectrl.c
vp8/encoder/rdopt.c
vp8/encoder/tokenize.c
vp8/encoder/tokenize.h