Since the while loop's condition already check
rc->frames_to_key < cpi->oxcf.key_freq,
it impossible to have "frames_to_key >= 2 * cpi->oxcf.key_freq"
and "frames_to_key > cpi->oxcf.key_freq".
Hence, these logics are removed.
Change-Id: I9dfc2ba36e1012718c857fc710036e2d30acd3b8
* changes:
Rename num_show_frames by num_coding_frames
Use compute_arf_boost() in define_gf_group()
Localize av_err mean_mod_score in define_gf_group
Move code of deciding gop size into brackets
When Checking for AVX Support, only the CPU's Capabilities and YMM
Register support by the OS were queried. In case of AVX-512, that is
insufficient, and ZMM Register support by the OS needs querying,
otherwise the OS will raise an Illegal Operation Exception if the CPU
is capable of AVX-512 but the OS is not.
Change-Id: I3444b19156d5743841de96cecbdaac19cc3f2b3f
The behavior is the same as that of detect_transition_still,
only we void using cpi and twopass->stats_in
Change-Id: I07722c817d98d8e4991a0a883235a582db8b5c3c
It's behavior is the same as that of calc_arf_boost()
But, we avoid using cpi and twopass->stats_in
Change-Id: I31cf7889abf43effcca9004a9d55f4b424ce388a
Note the last packet is cumulative first pass stats.
So the number of frames is packet number minus one
Change-Id: I5f617e7eeb63d17204beaaeb6422902ec076caeb
Move the logics of computing
gf_group_err, gf_group_raw_error, gf_group_noise,
gf_group_skip_pct, gf_group_inactive_zone_rowsa,
gf_group_inter, gf_group_motion
into one for loop
The behavior stays the same.
Change-Id: Idbc338a88469bf7a2786c831880e8aba8ed4feb5
This is part of the change aims at replacing
stats_in/stats_in_start/stats_in_end by first_pass_info.
Change-Id: Ibcd2a08e57cb749fe68996f33fe3a5e7f92b1758
when the best filter selected is not EIGHTTAP_SMOOTH, and
reuse_inter_pred is 0, pred buffer was not pointing to the right place.
Change-Id: I5b519fedd2d892bf140879faa74b463a161e253b
with vp9-highbitdepth off.
Unit Test: SSE2/Trans16x16DCT , VP9/LevelTest.TestTargetLevel20Large, VP9/CpuSpeedTest
implicit conversion from type 'int32_t' (aka 'int') of value -32851
(32-bit, signed) to type 'tran_low_t' (aka 'short') changed the value to
32685 (16-bit, signed)
BUG=webm:1615
BUG=webm:1647
Change-Id: I9ef064dc9ac734379628565ff6505b0876984123
Unit test: VP8/InvalidFileTest
implicit conversion from type 'int' of value -45844 (32-bit, signed) to
type 'short' changed the value to 19692 (16-bit, signed)
BUG=webm:1615
BUG=webm:1644
Change-Id: Id5d470f706d68e24f7a1e689526c9ecd3a8e8db8
Unit Test: VP9/InvalidFileTest
implicit conversion from type 'int' of value -65536 (32-bit, signed) to
type 'int16_t' (aka 'short') changed the value to 0 (16-bit, signed)
BUG=webm:1615
BUG=webm:1645
Change-Id: I4ce0c6abf8b5bf43ee43e958ad75d9fa28b23eee
From unit test: AVX/VP9QuantizeTest; SSSE3/VP9QuantizeTest ...
implicit conversion from type 'int' of value -139812 (32-bit, signed)
to type 'tran_low_t' (aka 'short') changed the value to -8740 (16-bit,
signed)
BUG=webm:1615
Change-Id: I730946ac6c7a250dcbcfd8a2712c0f1150ddb4fd
From unit test: VP9MultiThreaded/InvalidFileTest
implicit conversion from type 'int' of value 83144 (32-bit, signed) to
type 'tran_low_t' (aka 'short') changed the value to 17608 (16-bit,
signed)
BUG=webm:1615
BUG=webm:1648
Change-Id: I4170494c328596ace66432c8563c55f31745cf76
Fix some speed feature settings for speed 4
in real-time mode.
Use rd pickmode (i.e.,nonrd_pick_mode=0), but
use variance partitioning. Allow aq-mode=3 to
work at speed 4 and modify some other speed settings.
This makes it much faster than the current speed 4,
and still better quality than speed 5.
Change-Id: I94ec43ccac022030a75b5a528703be0c37f9a35c
