1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-26 19:01:44 +02:00
FFmpeg/libavcodec/qtrleenc.c
Andreas Rheinhardt 56e9e0273a avcodec/encode: Always use intermediate buffer in ff_alloc_packet2()
Up until now, ff_alloc_packet2() has a min_size parameter:
It is supposed to be a lower bound on the final size of the packet
to allocate. If it is not too far from the upper bound (namely,
if it is at least half the upper bound), then ff_alloc_packet2()
already allocates the final, already refcounted packet; if it is
not, then the packet is not refcounted and its data only points to
a buffer owned by the AVCodecContext (in this case, the packet will
be made refcounted in encode_simple_internal() in libavcodec/encode.c).
The goal of this was to avoid data copies and intermediate buffers
if one has a precise lower bound.

Yet those encoders for which precise lower bounds exist have recently
been switched to ff_get_encode_buffer() (which automatically allocates
final buffers), leaving only two encoders to actually set the min_size
to something else than zero (namely aliaspixenc and hapenc). Both of
these encoders use a very low lower bound that is not helpful in any
nontrivial case.

This commit therefore removes the min_size parameter as well as the
codepath in ff_alloc_packet2() for the allocation of final buffers.
Furthermore, the function has been renamed to ff_alloc_packet() and
moved to encode.h alongside ff_get_encode_buffer().

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2021-06-08 12:52:50 +02:00

416 lines
14 KiB
C

/*
* Quicktime Animation (RLE) Video Encoder
* Copyright (C) 2007 Clemens Fruhwirth
* Copyright (C) 2007 Alexis Ballier
*
* This file is based on flashsvenc.c.
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "libavutil/imgutils.h"
#include "avcodec.h"
#include "bytestream.h"
#include "encode.h"
#include "internal.h"
/** Maximum RLE code for bulk copy */
#define MAX_RLE_BULK 127
/** Maximum RLE code for repeat */
#define MAX_RLE_REPEAT 128
/** Maximum RLE code for skip */
#define MAX_RLE_SKIP 254
typedef struct QtrleEncContext {
AVCodecContext *avctx;
int pixel_size;
AVFrame *previous_frame;
unsigned int max_buf_size;
int logical_width;
/**
* This array will contain at ith position the value of the best RLE code
* if the line started at pixel i
* There can be 3 values :
* skip (0) : skip as much as possible pixels because they are equal to the
* previous frame ones
* repeat (<-1) : repeat that pixel -rle_code times, still as much as
* possible
* copy (>0) : copy the raw next rle_code pixels */
signed char *rlecode_table;
/**
* This array will contain the length of the best rle encoding of the line
* starting at ith pixel */
int *length_table;
/**
* Will contain at ith position the number of consecutive pixels equal to the previous
* frame starting from pixel i */
uint8_t* skip_table;
/** Encoded frame is a key frame */
int key_frame;
} QtrleEncContext;
static av_cold int qtrle_encode_end(AVCodecContext *avctx)
{
QtrleEncContext *s = avctx->priv_data;
av_frame_free(&s->previous_frame);
av_free(s->rlecode_table);
av_free(s->length_table);
av_free(s->skip_table);
return 0;
}
static av_cold int qtrle_encode_init(AVCodecContext *avctx)
{
QtrleEncContext *s = avctx->priv_data;
if (av_image_check_size(avctx->width, avctx->height, 0, avctx) < 0) {
return AVERROR(EINVAL);
}
s->avctx=avctx;
s->logical_width=avctx->width;
switch (avctx->pix_fmt) {
case AV_PIX_FMT_GRAY8:
if (avctx->width % 4) {
av_log(avctx, AV_LOG_ERROR, "Width not being a multiple of 4 is not supported\n");
return AVERROR(EINVAL);
}
s->logical_width = avctx->width / 4;
s->pixel_size = 4;
break;
case AV_PIX_FMT_RGB555BE:
s->pixel_size = 2;
break;
case AV_PIX_FMT_RGB24:
s->pixel_size = 3;
break;
case AV_PIX_FMT_ARGB:
s->pixel_size = 4;
break;
default:
av_log(avctx, AV_LOG_ERROR, "Unsupported colorspace.\n");
break;
}
avctx->bits_per_coded_sample = avctx->pix_fmt == AV_PIX_FMT_GRAY8 ? 40 : s->pixel_size*8;
s->rlecode_table = av_mallocz(s->logical_width);
s->skip_table = av_mallocz(s->logical_width);
s->length_table = av_mallocz_array(s->logical_width + 1, sizeof(int));
if (!s->skip_table || !s->length_table || !s->rlecode_table) {
av_log(avctx, AV_LOG_ERROR, "Error allocating memory.\n");
return AVERROR(ENOMEM);
}
s->previous_frame = av_frame_alloc();
if (!s->previous_frame) {
av_log(avctx, AV_LOG_ERROR, "Error allocating picture\n");
return AVERROR(ENOMEM);
}
s->max_buf_size = s->logical_width*s->avctx->height*s->pixel_size*2 /* image base material */
+ 15 /* header + footer */
+ s->avctx->height*2 /* skip code+rle end */
+ s->logical_width/MAX_RLE_BULK + 1 /* rle codes */;
return 0;
}
/**
* Compute the best RLE sequence for a line
*/
static void qtrle_encode_line(QtrleEncContext *s, const AVFrame *p, int line, uint8_t **buf)
{
int width=s->logical_width;
int i;
signed char rlecode;
/* This will be the number of pixels equal to the previous frame one's
* starting from the ith pixel */
unsigned int skipcount;
/* This will be the number of consecutive equal pixels in the current
* frame, starting from the ith one also */
unsigned int av_uninit(repeatcount);
/* The cost of the three different possibilities */
int total_skip_cost;
int total_repeat_cost;
int base_bulk_cost;
int lowest_bulk_cost;
int lowest_bulk_cost_index;
int sec_lowest_bulk_cost;
int sec_lowest_bulk_cost_index;
const uint8_t *this_line = p->data[0] + line * p->linesize[0] + width * s->pixel_size;
/* There might be no earlier frame if the current frame is a keyframe.
* So just use a pointer to the current frame to avoid a check
* to avoid NULL - s->pixel_size (which is undefined behaviour). */
const uint8_t *prev_line = s->key_frame ? this_line
: s->previous_frame->data[0]
+ line * s->previous_frame->linesize[0]
+ width * s->pixel_size;
s->length_table[width] = 0;
skipcount = 0;
/* Initial values */
lowest_bulk_cost = INT_MAX / 2;
lowest_bulk_cost_index = width;
sec_lowest_bulk_cost = INT_MAX / 2;
sec_lowest_bulk_cost_index = width;
base_bulk_cost = 1 + s->pixel_size;
for (i = width - 1; i >= 0; i--) {
int prev_bulk_cost;
this_line -= s->pixel_size;
prev_line -= s->pixel_size;
/* If our lowest bulk cost index is too far away, replace it
* with the next lowest bulk cost */
if (FFMIN(width, i + MAX_RLE_BULK) < lowest_bulk_cost_index) {
lowest_bulk_cost = sec_lowest_bulk_cost;
lowest_bulk_cost_index = sec_lowest_bulk_cost_index;
sec_lowest_bulk_cost = INT_MAX / 2;
sec_lowest_bulk_cost_index = width;
}
/* Deal with the first pixel's bulk cost */
if (!i) {
base_bulk_cost++;
lowest_bulk_cost++;
sec_lowest_bulk_cost++;
}
/* Look at the bulk cost of the previous loop and see if it is
* a new lower bulk cost */
prev_bulk_cost = s->length_table[i + 1] + base_bulk_cost;
if (prev_bulk_cost <= sec_lowest_bulk_cost) {
/* If it's lower than the 2nd lowest, then it may be lower
* than the lowest */
if (prev_bulk_cost <= lowest_bulk_cost) {
/* If we have found a new lowest bulk cost,
* then the 2nd lowest bulk cost is now farther than the
* lowest bulk cost, and will never be used */
sec_lowest_bulk_cost = INT_MAX / 2;
lowest_bulk_cost = prev_bulk_cost;
lowest_bulk_cost_index = i + 1;
} else {
/* Then it must be the 2nd lowest bulk cost */
sec_lowest_bulk_cost = prev_bulk_cost;
sec_lowest_bulk_cost_index = i + 1;
}
}
if (!s->key_frame && !memcmp(this_line, prev_line, s->pixel_size))
skipcount = FFMIN(skipcount + 1, MAX_RLE_SKIP);
else
skipcount = 0;
total_skip_cost = s->length_table[i + skipcount] + 2;
s->skip_table[i] = skipcount;
if (i < width - 1 && !memcmp(this_line, this_line + s->pixel_size, s->pixel_size))
repeatcount = FFMIN(repeatcount + 1, MAX_RLE_REPEAT);
else
repeatcount = 1;
total_repeat_cost = s->length_table[i + repeatcount] + 1 + s->pixel_size;
/* skip code is free for the first pixel, it costs one byte for repeat and bulk copy
* so let's make it aware */
if (i == 0) {
total_skip_cost--;
total_repeat_cost++;
}
if (repeatcount > 1 && (skipcount == 0 || total_repeat_cost < total_skip_cost)) {
/* repeat is the best */
s->length_table[i] = total_repeat_cost;
s->rlecode_table[i] = -repeatcount;
}
else if (skipcount > 0) {
/* skip is the best choice here */
s->length_table[i] = total_skip_cost;
s->rlecode_table[i] = 0;
}
else {
/* We cannot do neither skip nor repeat
* thus we use the best bulk copy */
s->length_table[i] = lowest_bulk_cost;
s->rlecode_table[i] = lowest_bulk_cost_index - i;
}
/* These bulk costs increase every iteration */
lowest_bulk_cost += s->pixel_size;
sec_lowest_bulk_cost += s->pixel_size;
}
/* Good! Now we have the best sequence for this line, let's output it. */
/* We do a special case for the first pixel so that we avoid testing it in
* the whole loop */
i=0;
this_line = p-> data[0] + line*p->linesize[0];
if (s->rlecode_table[0] == 0) {
bytestream_put_byte(buf, s->skip_table[0] + 1);
i += s->skip_table[0];
}
else bytestream_put_byte(buf, 1);
while (i < width) {
rlecode = s->rlecode_table[i];
bytestream_put_byte(buf, rlecode);
if (rlecode == 0) {
/* Write a skip sequence */
bytestream_put_byte(buf, s->skip_table[i] + 1);
i += s->skip_table[i];
}
else if (rlecode > 0) {
/* bulk copy */
if (s->avctx->pix_fmt == AV_PIX_FMT_GRAY8) {
int j;
// QT grayscale colorspace has 0=white and 255=black, we will
// ignore the palette that is included in the AVFrame because
// AV_PIX_FMT_GRAY8 has defined color mapping
for (j = 0; j < rlecode*s->pixel_size; ++j)
bytestream_put_byte(buf, *(this_line + i*s->pixel_size + j) ^ 0xff);
} else {
bytestream_put_buffer(buf, this_line + i*s->pixel_size, rlecode*s->pixel_size);
}
i += rlecode;
}
else {
/* repeat the bits */
if (s->avctx->pix_fmt == AV_PIX_FMT_GRAY8) {
int j;
// QT grayscale colorspace has 0=white and 255=black, ...
for (j = 0; j < s->pixel_size; ++j)
bytestream_put_byte(buf, *(this_line + i*s->pixel_size + j) ^ 0xff);
} else {
bytestream_put_buffer(buf, this_line + i*s->pixel_size, s->pixel_size);
}
i -= rlecode;
}
}
bytestream_put_byte(buf, -1); // end RLE line
}
/** Encode frame including header */
static int encode_frame(QtrleEncContext *s, const AVFrame *p, uint8_t *buf)
{
int i;
int start_line = 0;
int end_line = s->avctx->height;
uint8_t *orig_buf = buf;
if (!s->key_frame) {
unsigned line_size = s->logical_width * s->pixel_size;
for (start_line = 0; start_line < s->avctx->height; start_line++)
if (memcmp(p->data[0] + start_line*p->linesize[0],
s->previous_frame->data[0] + start_line * s->previous_frame->linesize[0],
line_size))
break;
for (end_line=s->avctx->height; end_line > start_line; end_line--)
if (memcmp(p->data[0] + (end_line - 1)*p->linesize[0],
s->previous_frame->data[0] + (end_line - 1) * s->previous_frame->linesize[0],
line_size))
break;
}
bytestream_put_be32(&buf, 0); // CHUNK SIZE, patched later
if ((start_line == 0 && end_line == s->avctx->height) || start_line == s->avctx->height)
bytestream_put_be16(&buf, 0); // header
else {
bytestream_put_be16(&buf, 8); // header
bytestream_put_be16(&buf, start_line); // starting line
bytestream_put_be16(&buf, 0); // unknown
bytestream_put_be16(&buf, end_line - start_line); // lines to update
bytestream_put_be16(&buf, 0); // unknown
}
for (i = start_line; i < end_line; i++)
qtrle_encode_line(s, p, i, &buf);
bytestream_put_byte(&buf, 0); // zero skip code = frame finished
AV_WB32(orig_buf, buf - orig_buf); // patch the chunk size
return buf - orig_buf;
}
static int qtrle_encode_frame(AVCodecContext *avctx, AVPacket *pkt,
const AVFrame *pict, int *got_packet)
{
QtrleEncContext * const s = avctx->priv_data;
int ret;
if ((ret = ff_alloc_packet(avctx, pkt, s->max_buf_size)) < 0)
return ret;
if (avctx->gop_size == 0 || !s->previous_frame->data[0] ||
(s->avctx->frame_number % avctx->gop_size) == 0) {
/* I-Frame */
s->key_frame = 1;
} else {
/* P-Frame */
s->key_frame = 0;
}
pkt->size = encode_frame(s, pict, pkt->data);
/* save the current frame */
av_frame_unref(s->previous_frame);
ret = av_frame_ref(s->previous_frame, pict);
if (ret < 0) {
av_log(avctx, AV_LOG_ERROR, "cannot add reference\n");
return ret;
}
if (s->key_frame)
pkt->flags |= AV_PKT_FLAG_KEY;
*got_packet = 1;
return 0;
}
const AVCodec ff_qtrle_encoder = {
.name = "qtrle",
.long_name = NULL_IF_CONFIG_SMALL("QuickTime Animation (RLE) video"),
.type = AVMEDIA_TYPE_VIDEO,
.id = AV_CODEC_ID_QTRLE,
.priv_data_size = sizeof(QtrleEncContext),
.init = qtrle_encode_init,
.encode2 = qtrle_encode_frame,
.close = qtrle_encode_end,
.pix_fmts = (const enum AVPixelFormat[]){
AV_PIX_FMT_RGB24, AV_PIX_FMT_RGB555BE, AV_PIX_FMT_ARGB, AV_PIX_FMT_GRAY8, AV_PIX_FMT_NONE
},
.caps_internal = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_INIT_CLEANUP,
};