Age | Commit message (Collapse) | Author |
|
In non-rd pickmode for screen content:
this logic to reset segment should only be for cyclic_refresh
mode on, so add that condition explicitly.
There may be other uses of segments, like ROI, so we
should condition this reset logic on cyclic_refresh,
as it was intended for that mode only.
Change-Id: I954e6cee968fbca35b34286c4a7ca2531c8e9823
|
|
|
|
For CBR mode: clamp the Q to worst/best quality in
adjust_q_cbr().
Under certain conditions, when the worst/best quality is
suddenly changed by a large amount mid-stream, the Q
adjustment from the final Q from adjust_q_cbr may not respect
the worst/best qualiy limits.
Change-Id: I3776129325d89882d422b22e6247d44660dd90ac
|
|
Increases denoising for noisy content.
Change-Id: Iff8573f8dca7b177ef53ee6682d691e6cd8e2740
|
|
When compiled for High Bitdepth SSE can overflow 32-bit unsigned
integer, so change it to 64 bit. Also fixing unit/int mismatch of sum
BUG=webm:1601
Change-Id: Ib576ed1d5579b0c2b4661058aa64119560b652bf
|
|
Lower the frame_motion and consec_zeromv thresholds,
this make the noise estimation and denoiser have more effect
on noisy clips.
Change-Id: I49cf5d78a04d00fcf8538bee6f3b2980efe6b3b5
|
|
|
|
usage was removed with:
c1b024b48 Modify map read/write to sync logic in row_mt case
Change-Id: I515fe397083079a4f11702e67c322fd04bdcf410
|
|
|
|
BUG=webm:1031
Change-Id: I28f4e8cdec170393b2d22cd8cb0b73a32204e09c
|
|
~10% speed up with no quality change for speed 8.
7% quality gain for speed 9 with no speed change.
Change-Id: I7eaaa4b82f7b082c9b15aa1d7624765ecc5082e7
|
|
Change-Id: I3e521b62a2f99902c4be80fe25d3869121673e43
|
|
Change-Id: Ib2bd9a74473ccb00e9ad71e0b186c8ddc0ee7b3c
|
|
Change-Id: I3326f0912627981fd604b16ddbf668d2262d4287
|
|
Change-Id: I38c5d4de93bebfd3f46bcc01716a0cc4a76af950
|
|
Change-Id: I98c32aba4c9e81380b588dcdbfa991468487ce73
|
|
The visited is not set to 1 after an item is pushed into the heap.
This may cause one item being pushed into the heap multiple
times, which may incur buffer overflow and memory corruption.
Change-Id: I443f1e5693856bb4066542403f98492d4daec69d
|
|
|
|
Add last_q[] to layer context, and add limit on
Q change from previous layer/frame. For now put
hard limit of 12 for decrease.
For 1 pass CBR sreen content mode.
Change-Id: Ifb972c9b6831440c80b1cb07a054c577ece930ec
|
|
|
|
Also write it to opsnr.stt when internal stats is enabled.
Removed some redundant code in vpxenc.c and vp9cx_set_ref.c
Change-Id: I3700137fff0be92a23e4ab75713db72da1dc4076
|
|
Remove it from runtime flag.
Add new struct for rd ml partition.
BUG=webm:1599
Change-Id: I883edbba83c65b7e557b8832419e212cffc85997
|
|
For 1 pass cbr screen-content mode: quantity should
only be updated on delta frames.
Change-Id: I16fc47b2805c7527ab4ff25bd8b5a5bd9c2b8976
|
|
|
|
Put rc stats related code into a separate function.
Change-Id: I11808bb947079b5fd9e53dfa5894bf227ed0c4c6
|
|
* changes:
Add tpl_bsize to VP9_COMP
Compute future_rd_diff in predict_mv_mode
|
|
Make it same as vpxenc so easier to run on borg.
Change-Id: Ie19db6e828ced773cba9aef715c8fbd0f4715b27
|
|
|
|
ClangTidy reported 16 issues. All are around typecasting and
straightforward
Change-Id: Ie8f9fc2ba7992dd44fef65b121fe65966a1a1297
|
|
|
|
|
|
Write height and width of top layer to ivf header in SVC.
vpxdec Can't decode it correctly when output is y4m.
Change-Id: I9b2f1d54696611a30e252bdfd182897d191d92b5
|
|
Remove the block variance and skip flags from the input features. They
do not seem to reduce the average loss of the model. Also decrease the
number of hidden nodes. The model size is reduced significantly.
Compression quality and speed are both neutral.
Change-Id: Ic62f73c4f4c0a3148285f575747f0423ff568c64
|
|
BUG=webm:1270
Change-Id: I7d56667d946196bbbe355303de805422e40b0763
|
|
Change-Id: Ibbe12dade04b218a41de9b65bbedba0054a69d83
|
|
The future_rd_diff computes the future rd difference between
new mv mode and ref mv mode.
Change-Id: I4bf9e2ad34257ba9cfec95419c2c5eca469584e9
|
|
|
|
|
|
|
|
VP9 encoder still inserts key frame periodically when VPX_KF_DISABLED is
set in non SVC for 1-pass CBR.
BUG=webm:1592
Change-Id: Ie99d7c5b95230d739e263a2d87879693c53f620e
|
|
|
|
Y4M reader has its own allocation.
Change-Id: Ie02440a183126072ea773860f4e9dc9b412772f5
|
|
Block "for (int i;;)" style declarations.
Use --std=gnu89 to avoid enforcing c89-style comments.
Change-Id: Ia7d1405eac647d04e92513c047773695e8d7dc6e
|
|
|
|
Change-Id: Ib8ca37f1b58e9903e7efa29689a0a49f14b4d73a
|
|
|
|
|
|
Include the sizes of the above and left partition block as additional
features.
This affects speed 0 and 1.
Compression change is almost neutral(about 0.03% on average).
Average encoding speedup is 3~6% depending on QP and resolution.
Change-Id: I8bddfadf6072ae757c124da0819302850d8c6fe7
|
|
bug=webm:1595
Change-Id: I7fbb16444a8526eb9479007772fbf52b09ff8338
|
|
|