A small gain (0.1 - 0.2%) with this experiment on derflr/hevcmr.
The DST2 can be implemened very efficiently using sign flipping
of odd indexed inputs, followed by DCT, followed by reversal of
the output. This is how it is implemented in this patch.
SIMD optimization is pending.
Change-Id: Ic2fc211ce0e6b7c6702974d76d6573f55cc4da0e
This is a port of 01bb4a318dc0f9069264b7fd5641bc3014f47f32
This commit also fixes a bug where FLIPADST transforms when combined
with a DST (that is FLIPADST_DST and DST_FLIPADST) did not actually did
a flipped transform but a straight ADST instead. This was due to the C
implementation that it fell back on not implementing flipping. This is
now fixed as well and FLIPADST_DST and DST_FLIPADST does what it is
supposed to do.
Change-Id: I89c67ca1d5e06808a1567c51e7d6bec4998182bd
Adds an additional transform in the ext_tx experiment that
is a 2d DST1-DST1 combination.
To enable use --enable-ext-tx --enable-dst1.
This needs to be later extended to combine DST1 with DCT
or ADST.
Change-Id: I6d29f1b778ef8294bcfb6a512a78fc5eda20723b
The wavelets implemented are 2/6, 5/3 and 9/7 each with
a lifting based scheme for even block sizes. The 9/7
one is a double implementation currently.
This is to start experiments with:
1. Replacing large transforms (32x32 and 64x64) with wavelets
or wavelet-dct hybrids that can hopefully localize errors better
spatially. (Will also need alternate entropy coder)
2. Super-resolution modes where the higher sub-bands may be
selectively skipped from being conveyed, while a smart
reconstruction recovers the lost frequencies.
The current patch includes two types of 32x32 and 64x64
transforms: one where only wavelets are used, and another
where a single level wavelet decomposition is followed
by a lower resolution dct on the low-low band.
Change-Id: I2d6755c4e6c8ec9386a04633dacbe0de3b0043ec
This change is made in preparation for a
subsequent patch which adds acceleration
for the highbitdepth transform functions.
The highbitdepth transform functions attempt
to use 16/32bit sse instructions where possible,
but fallback to using the C implementations if
potential overflow is detected. For this reason
the dct routines are made global so they can be
called from the acceleration functions in the
subsequent patch.
Change-Id: Ia921f191bf6936ccba4f13e8461624b120c1f665
vp9_quantize_rect did illegal shifts but didn't use the results.
The shift |a << b| is unfortunately undefined if |a < 0|, but the
more verbose |a * (1 << b)| generates the same machine code.
Change-Id: I7ceac66fa20a700630cf8ed008949146b161dab4
Implements vertical, horizontal, and tm dpcm intra prediction for
blocks in tx_skip mode. Typical coding gain on screen content video
is 2%~5%.
Change-Id: Idd5bd84ac59daa586ec0cd724680cef695981651
This patch improves the non-transform coding mode. At this
point, the coding gain on screen content videos is about
12% for lossless, an 15% for lossy case.
1. Encode tx_skip flags with context. Y tx_skip flag context is
whether the prediction mode is inter or intra. UV flag context
is Y flag.
2. Transform skipping is less helpful when the Q-index is high.
So it is enabled only when the Q-index is smaller than a
threshold. Currently the threshold is set as 255 for intra blocks,
and 0 for inter blocks.
3. The shift of the prediction residue, when copying them to the
coeff buffer, is set as 3 when the Q-index is larger than a
threshold (currently set as 0), and 2 otherwise.
Change-Id: I372973c7518cf385f6e542b22d0f803016e693b0
Non-transform option is enabled in both intra and inter modes.
In lossless case, the average coding gain on screen content
clips is 11.3% in my test.
Change-Id: I2e8de515fb39e74c61bb86ce0f682d5f79e15188
Preliminary 64x64 transform implementation.
Includes all code changes.
All mismatches resolved.
Coding results for derf and stdhd are within noise. stdhd is slightly
higher, derf is slightly lower.
To be further refined.
Change-Id: I091c183f62b156d23ed6f648202eb96c82e69b4b
Adds various high bitdepth transform functions and tests.
Much of the changes are related to using typedefs tran_low_t
and tran_high_t for the final transform cofficients and intermediate
stages of the transform computation respectively rather than fixed
types int16_t/int. When vp9_highbitdepth configure flag is off,
these map tp int16_t/int32_t, but when the flag is on, they map
to int32_t/int64_t to make space for needed extra precision.
Change-Id: I3c56de79e15b904d6f655b62ffae170729befdd8
This commit enables a fast path computational flow for forward
transformation. It checks the sse and variance of prediction
residuals and decides if the quantized coefficients are all
zero, dc only, or more. It then selects the corresponding coding
path in the forward transformation and quantization stage.
It is currently enabled in rtc coding mode. Will do it for rd
coding mode next.
In speed -6, the runtime for pedestrian_area 1080p at 1000 kbps
goes down from 14234 ms to 13704 ms, i.e., about 4% speed-up.
Overall coding performance for rtc set is changed by -0.18%.
Change-Id: I0452da1786d59bc8bcbe0a35fdae9f623d1d44e1
This commit adjusts the forward 16x16 DCT computation steps to
simplify the register level operations. It fixes the corresponding
sse2 version accordingly.
Change-Id: I72a9c25b8ca9442fc5e113f47cd701ae55aa7f08
Separate the rounding and right shift operations of forward transform
from those of inverse transform. Take out the assertion check from
inverse transforms. If the transform coefficients were constructed to
cause intermediate steps of inverse transform overflow, the codec will
just let it overflow without breaking the decoding flow.
Change-Id: I73cfc3706c4e840fc543a77cbc4cdb0b05d07730
Adding these functions to encapsulate tx_type check. Changing TX_TYPE to
int to match the declaration in vo9_rtch.h.
Change-Id: I6f3a2df6e35595ca73b6aaa9e3909ee7bc3fd16f
For consistency with idct function names. Renames:
vp9_short_fdct4x4 -> vp9_fdct4x4
vp9_short_walsh4x4 -> vp9_fwht4x4
Change-Id: Id15497cc1270acca626447d846f0ce9199770f58
Just making fdct consistent with iht/idct/fht functions which all use
stride (# of elements) as input argument.
Change-Id: I0ba3c52513a5fdd194f1e7e2901092671398985b
Just making fdct consistent with iht/idct/fht functions which all use
stride (# of elements) as input argument.
Change-Id: Ibc944952a192e6c7b2b6a869ec2894c01da82ed1
Just making fdct consistent with iht/idct/fht functions which all use
stride (# of elements) as input argument.
Change-Id: I2d95fdcbba96aaa0ed24a80870cb38f53487a97d
Just making fdct consistent with iht/idct/fht functions which all use
stride (# of elements) as input argument.
Change-Id: Id623c5113262655fa50f7c9d6cec9a91fcb20bb4
Renames:
fdct4_1d -> fdct4
fadst4_1d -> fadst4
fdct8_1d -> fdct8
fadst8_1d -> fadst8
fdct16_1d -> fdct16
fadst16_1d -> fadst16
"_1d" suffix is redundant, so removing it. The same will happen with idct
in the next change sets.
Change-Id: Ibf421cd2f569146c6079269df7a31819c098265e
This commit fixed the potential overflow issue in the SSE2
implementation of 32x32 forward DCT. It resolved the corrupted
coded frames in the border of scenes.
Change-Id: If87eef2d46209269f74ef27e7295b6707fbf56f9
This commit enables 8x8 DCT and hybrid transform unit tests. It
also tunes the forward hybrid transform rounding opertions for
more precise round-trip performance.
Change-Id: If05c1ce59d75d641b9c6c91527d02d3a6ef498c3
This commit makes use of dual fdct32x32 versions for rate-distortion
optimization loop and encoding process, respectively. The one for
rd loop requires only 16 bits precision for intermediate steps.
The original fdct32x32 that allows higher intermediate precision (18
bits) was retained for the encoding process only.
This allows speed-up for fdct32x32 in the rd loop. No performance
loss observed.
Change-Id: I3237770e39a8f87ed17ae5513c87228533397cc3
The commit changed to use a new variant of Walsh-Hadamard Transform
by Tim Terriberry. This new variant has the best compression among a
number of variants that developed by Tim.
Change-Id: Icb3a88515463cfc644b17ca046fcd139db2557e9
