| Message ID | 20200904231716.16182-3-andreas.rheinhardt@gmail.com |
|---|---|
| State | Accepted |
| Commit | fefe69c5c9c6d4cfa2b1329a72a805ffd89a97c3 |
| Headers | show |
| Series | [FFmpeg-devel,1/8] avcodec/jpeglsenc: Don't use put bits API for byte-aligned writes | expand |
| Context | Check | Description |
|---|---|---|
| andriy/default | pending | |
| andriy/make | success | Make finished |
| andriy/make_fate | success | Make fate finished |
On Sat, Sep 05, 2020 at 01:17:11AM +0200, Andreas Rheinhardt wrote: > ls_encode_line() encodes a line of input, going from left to right. In > order to calculate a predicted value it uses the left and upper-left > value of the output picture (that is, it uses how a decoder would see > the already encoded part of the picture), unless this is the very first > pixel of this line in which case one uses the first pixel of the last > (upper) line and the line before the last line. Therefore the loop > contained a check for whether this is the beginning of a new line. This > commit moves said check out of the loop by initializing these values > before the loop and by updating these values at the end of the loop > body; already read/calculated values are reused for this (the prediction > also needs the value of the upper pixel and this can be reused for the > upper left value of the next iteration of the loop). > > Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com> > --- > libavcodec/jpeglsenc.c | 13 ++++++------- > 1 file changed, 6 insertions(+), 7 deletions(-) > probably ok
diff --git a/libavcodec/jpeglsenc.c b/libavcodec/jpeglsenc.c index 67897ce8ae..b9c04e979c 100644 --- a/libavcodec/jpeglsenc.c +++ b/libavcodec/jpeglsenc.c @@ -145,16 +145,14 @@ static inline void ls_encode_line(JLSState *state, PutBitContext *pb, int stride, int comp, int bits) { int x = 0; - int Ra, Rb, Rc, Rd; + int Ra = R(last, 0), Rb, Rc = last2, Rd; int D0, D1, D2; while (x < w) { int err, pred, sign; /* compute gradients */ - Ra = x ? R(cur, x - stride) : R(last, x); Rb = R(last, x); - Rc = x ? R(last, x - stride) : last2; Rd = (x >= w - stride) ? R(last, x) : R(last, x + stride); D0 = Rd - Rb; D1 = Rb - Rc; @@ -194,9 +192,9 @@ static inline void ls_encode_line(JLSState *state, PutBitContext *pb, Ra = av_clip(pred + err * state->twonear, 0, state->maxval); else Ra = av_clip(pred - err * state->twonear, 0, state->maxval); - W(cur, x, Ra); } else - W(cur, x, R(in, x)); + Ra = R(in, x); + W(cur, x, Ra); if (err < 0) err += state->range; @@ -236,12 +234,13 @@ static inline void ls_encode_line(JLSState *state, PutBitContext *pb, Ra = av_clip(pred + err * state->twonear, 0, state->maxval); else Ra = av_clip(pred - err * state->twonear, 0, state->maxval); - W(cur, x, Ra); } else - W(cur, x, R(in, x)); + Ra = R(in, x); + W(cur, x, Ra); ls_encode_regular(state, pb, context, err); } + Rc = Rb; x += stride; } }
ls_encode_line() encodes a line of input, going from left to right. In order to calculate a predicted value it uses the left and upper-left value of the output picture (that is, it uses how a decoder would see the already encoded part of the picture), unless this is the very first pixel of this line in which case one uses the first pixel of the last (upper) line and the line before the last line. Therefore the loop contained a check for whether this is the beginning of a new line. This commit moves said check out of the loop by initializing these values before the loop and by updating these values at the end of the loop body; already read/calculated values are reused for this (the prediction also needs the value of the upper pixel and this can be reused for the upper left value of the next iteration of the loop). Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com> --- libavcodec/jpeglsenc.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-)