MrBesen/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
2c05ee092b
ffmpeg/libavcodec/utvideoenc.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

668 lines
20 KiB
C
Raw Blame History

/*
* Ut Video encoder
* Copyright (c) 2012 Jan Ekström
*
* 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
*/
/**
* @file
* Ut Video encoder
*/
#include "libavutil/imgutils.h"
#include "libavutil/intreadwrite.h"
#include "libavutil/opt.h"
#include "avcodec.h"
#include "encode.h"
#include "internal.h"
#include "bswapdsp.h"
#include "bytestream.h"
#include "put_bits.h"
#include "mathops.h"
#include "utvideo.h"
#include "huffman.h"
typedef struct HuffEntry {
uint16_t sym;
uint8_t len;
uint32_t code;
} HuffEntry;
/* Compare huffman tree nodes */
static int ut_huff_cmp_len(const void *a, const void *b)
{
const HuffEntry *aa = a, *bb = b;
return (aa->len - bb->len)*256 + aa->sym - bb->sym;
}
/* Compare huffentry symbols */
static int huff_cmp_sym(const void *a, const void *b)
{
const HuffEntry *aa = a, *bb = b;
return aa->sym - bb->sym;
}
static av_cold int utvideo_encode_close(AVCodecContext *avctx)
{
UtvideoContext *c = avctx->priv_data;
int i;
av_freep(&c->slice_bits);
for (i = 0; i < 4; i++)
av_freep(&c->slice_buffer[i]);
return 0;
}
static av_cold int utvideo_encode_init(AVCodecContext *avctx)
{
UtvideoContext *c = avctx->priv_data;
int i, subsampled_height;
uint32_t original_format;
c->avctx = avctx;
c->frame_info_size = 4;
c->slice_stride = FFALIGN(avctx->width, 32);
switch (avctx->pix_fmt) {
case AV_PIX_FMT_GBRP:
c->planes = 3;
avctx->codec_tag = MKTAG('U', 'L', 'R', 'G');
original_format = UTVIDEO_RGB;
break;
case AV_PIX_FMT_GBRAP:
c->planes = 4;
avctx->codec_tag = MKTAG('U', 'L', 'R', 'A');
original_format = UTVIDEO_RGBA;
avctx->bits_per_coded_sample = 32;
break;
case AV_PIX_FMT_YUV420P:
if (avctx->width & 1 || avctx->height & 1) {
av_log(avctx, AV_LOG_ERROR,
"4:2:0 video requires even width and height.\n");
return AVERROR_INVALIDDATA;
}
c->planes = 3;
if (avctx->colorspace == AVCOL_SPC_BT709)
avctx->codec_tag = MKTAG('U', 'L', 'H', '0');
else
avctx->codec_tag = MKTAG('U', 'L', 'Y', '0');
original_format = UTVIDEO_420;
break;
case AV_PIX_FMT_YUV422P:
if (avctx->width & 1) {
av_log(avctx, AV_LOG_ERROR,
"4:2:2 video requires even width.\n");
return AVERROR_INVALIDDATA;
}
c->planes = 3;
if (avctx->colorspace == AVCOL_SPC_BT709)
avctx->codec_tag = MKTAG('U', 'L', 'H', '2');
else
avctx->codec_tag = MKTAG('U', 'L', 'Y', '2');
original_format = UTVIDEO_422;
break;
case AV_PIX_FMT_YUV444P:
c->planes = 3;
if (avctx->colorspace == AVCOL_SPC_BT709)
avctx->codec_tag = MKTAG('U', 'L', 'H', '4');
else
avctx->codec_tag = MKTAG('U', 'L', 'Y', '4');
original_format = UTVIDEO_444;
break;
default:
av_log(avctx, AV_LOG_ERROR, "Unknown pixel format: %d\n",
avctx->pix_fmt);
return AVERROR_INVALIDDATA;
}
ff_bswapdsp_init(&c->bdsp);
ff_llvidencdsp_init(&c->llvidencdsp);
if (c->frame_pred == PRED_GRADIENT) {
av_log(avctx, AV_LOG_ERROR, "Gradient prediction is not supported.\n");
return AVERROR_OPTION_NOT_FOUND;
}
/*
* Check the asked slice count for obviously invalid
* values (> 256 or negative).
*/
if (avctx->slices > 256 || avctx->slices < 0) {
av_log(avctx, AV_LOG_ERROR,
"Slice count %d is not supported in Ut Video (theoretical range is 0-256).\n",
avctx->slices);
return AVERROR(EINVAL);
}
/* Check that the slice count is not larger than the subsampled height */
subsampled_height = avctx->height >> av_pix_fmt_desc_get(avctx->pix_fmt)->log2_chroma_h;
if (avctx->slices > subsampled_height) {
av_log(avctx, AV_LOG_ERROR,
"Slice count %d is larger than the subsampling-applied height %d.\n",
avctx->slices, subsampled_height);
return AVERROR(EINVAL);
}
/* extradata size is 4 * 32 bits */
avctx->extradata_size = 16;
avctx->extradata = av_mallocz(avctx->extradata_size +
AV_INPUT_BUFFER_PADDING_SIZE);
if (!avctx->extradata) {
av_log(avctx, AV_LOG_ERROR, "Could not allocate extradata.\n");
return AVERROR(ENOMEM);
}
for (i = 0; i < c->planes; i++) {
c->slice_buffer[i] = av_malloc(c->slice_stride * (avctx->height + 2) +
AV_INPUT_BUFFER_PADDING_SIZE);
if (!c->slice_buffer[i]) {
av_log(avctx, AV_LOG_ERROR, "Cannot allocate temporary buffer 1.\n");
return AVERROR(ENOMEM);
}
}
/*
* Set the version of the encoder.
* Last byte is "implementation ID", which is
* obtained from the creator of the format.
* Libavcodec has been assigned with the ID 0xF0.
*/
AV_WB32(avctx->extradata, MKTAG(1, 0, 0, 0xF0));
/*
* Set the "original format"
* Not used for anything during decoding.
*/
AV_WL32(avctx->extradata + 4, original_format);
/* Write 4 as the 'frame info size' */
AV_WL32(avctx->extradata + 8, c->frame_info_size);
/*
* Set how many slices are going to be used.
* By default uses multiple slices depending on the subsampled height.
* This enables multithreading in the official decoder.
*/
if (!avctx->slices) {
c->slices = subsampled_height / 120;
if (!c->slices)
c->slices = 1;
else if (c->slices > 256)
c->slices = 256;
} else {
c->slices = avctx->slices;
}
/* Set compression mode */
c->compression = COMP_HUFF;
/*
* Set the encoding flags:
* - Slice count minus 1
* - Interlaced encoding mode flag, set to zero for now.
* - Compression mode (none/huff)
* And write the flags.
*/
c->flags = (c->slices - 1) << 24;
c->flags |= 0 << 11; // bit field to signal interlaced encoding mode
c->flags |= c->compression;
AV_WL32(avctx->extradata + 12, c->flags);
return 0;
}
static void mangle_rgb_planes(uint8_t *dst[4], ptrdiff_t dst_stride,
uint8_t *const src[4], int planes, const int stride[4],
int width, int height)
{
int i, j;
int k = 2 * dst_stride;
const uint8_t *sg = src[0];
const uint8_t *sb = src[1];
const uint8_t *sr = src[2];
const uint8_t *sa = src[3];
unsigned int g;
for (j = 0; j < height; j++) {
if (planes == 3) {
for (i = 0; i < width; i++) {
g = sg[i];
dst[0][k] = g;
g += 0x80;
dst[1][k] = sb[i] - g;
dst[2][k] = sr[i] - g;
k++;
}
} else {
for (i = 0; i < width; i++) {
g = sg[i];
dst[0][k] = g;
g += 0x80;
dst[1][k] = sb[i] - g;
dst[2][k] = sr[i] - g;
dst[3][k] = sa[i];
k++;
}
sa += stride[3];
}
k += dst_stride - width;
sg += stride[0];
sb += stride[1];
sr += stride[2];
}
}
#undef A
#undef B
/* Write data to a plane with median prediction */
static void median_predict(UtvideoContext *c, uint8_t *src, uint8_t *dst,
ptrdiff_t stride, int width, int height)
{
int i, j;
int A, B;
uint8_t prev;
/* First line uses left neighbour prediction */
prev = 0x80; /* Set the initial value */
for (i = 0; i < width; i++) {
*dst++ = src[i] - prev;
prev = src[i];
}
if (height == 1)
return;
src += stride;
/*
* Second line uses top prediction for the first sample,
* and median for the rest.
*/
A = B = 0;
/* Rest of the coded part uses median prediction */
for (j = 1; j < height; j++) {
c->llvidencdsp.sub_median_pred(dst, src - stride, src, width, &A, &B);
dst += width;
src += stride;
}
}
/* Count the usage of values in a plane */
static void count_usage(uint8_t *src, int width,
int height, uint64_t *counts)
{
int i, j;
for (j = 0; j < height; j++) {
for (i = 0; i < width; i++) {
counts[src[i]]++;
}
src += width;
}
}
/* Calculate the actual huffman codes from the code lengths */
static void calculate_codes(HuffEntry *he)
{
int last, i;
uint32_t code;
qsort(he, 256, sizeof(*he), ut_huff_cmp_len);
last = 255;
while (he[last].len == 255 && last)
last--;
code = 0;
for (i = last; i >= 0; i--) {
he[i].code = code >> (32 - he[i].len);
code += 0x80000000u >> (he[i].len - 1);
}
qsort(he, 256, sizeof(*he), huff_cmp_sym);
}
/* Write huffman bit codes to a memory block */
static int write_huff_codes(uint8_t *src, uint8_t *dst, int dst_size,
int width, int height, HuffEntry *he)
{
PutBitContext pb;
int i, j;
int count;
init_put_bits(&pb, dst, dst_size);
/* Write the codes */
for (j = 0; j < height; j++) {
for (i = 0; i < width; i++)
put_bits(&pb, he[src[i]].len, he[src[i]].code);
src += width;
}
/* Pad output to a 32-bit boundary */
count = put_bits_count(&pb) & 0x1F;
if (count)
put_bits(&pb, 32 - count, 0);
/* Flush the rest with zeroes */
flush_put_bits(&pb);
/* Return the amount of bytes written */
return put_bytes_output(&pb);
}
static int encode_plane(AVCodecContext *avctx, uint8_t *src,
uint8_t *dst, ptrdiff_t stride, int plane_no,
int width, int height, PutByteContext *pb)
{
UtvideoContext *c = avctx->priv_data;
uint8_t lengths[256];
uint64_t counts[256] = { 0 };
HuffEntry he[256];
uint32_t offset = 0, slice_len = 0;
const int cmask = ~(!plane_no && avctx->pix_fmt == AV_PIX_FMT_YUV420P);
int i, sstart, send = 0;
int symbol;
int ret;
/* Do prediction / make planes */
switch (c->frame_pred) {
case PRED_NONE:
for (i = 0; i < c->slices; i++) {
sstart = send;
send = height * (i + 1) / c->slices & cmask;
av_image_copy_plane(dst + sstart * width, width,
src + sstart * stride, stride,
width, send - sstart);
}
break;
case PRED_LEFT:
for (i = 0; i < c->slices; i++) {
sstart = send;
send = height * (i + 1) / c->slices & cmask;
c->llvidencdsp.sub_left_predict(dst + sstart * width, src + sstart * stride, stride, width, send - sstart);
}
break;
case PRED_MEDIAN:
for (i = 0; i < c->slices; i++) {
sstart = send;
send = height * (i + 1) / c->slices & cmask;
median_predict(c, src + sstart * stride, dst + sstart * width,
stride, width, send - sstart);
}
break;
default:
av_log(avctx, AV_LOG_ERROR, "Unknown prediction mode: %d\n",
c->frame_pred);
return AVERROR_OPTION_NOT_FOUND;
}
/* Count the usage of values */
count_usage(dst, width, height, counts);
/* Check for a special case where only one symbol was used */
for (symbol = 0; symbol < 256; symbol++) {
/* If non-zero count is found, see if it matches width * height */
if (counts[symbol]) {
/* Special case if only one symbol was used */
if (counts[symbol] == width * (int64_t)height) {
/*
* Write a zero for the single symbol
* used in the plane, else 0xFF.
*/
for (i = 0; i < 256; i++) {
if (i == symbol)
bytestream2_put_byte(pb, 0);
else
bytestream2_put_byte(pb, 0xFF);
}
/* Write zeroes for lengths */
for (i = 0; i < c->slices; i++)
bytestream2_put_le32(pb, 0);
/* And that's all for that plane folks */
return 0;
}
break;
}
}
/* Calculate huffman lengths */
if ((ret = ff_huff_gen_len_table(lengths, counts, 256, 1)) < 0)
return ret;
/*
* Write the plane's header into the output packet:
* - huffman code lengths (256 bytes)
* - slice end offsets (gotten from the slice lengths)
*/
for (i = 0; i < 256; i++) {
bytestream2_put_byte(pb, lengths[i]);
he[i].len = lengths[i];
he[i].sym = i;
}
/* Calculate the huffman codes themselves */
calculate_codes(he);
send = 0;
for (i = 0; i < c->slices; i++) {
sstart = send;
send = height * (i + 1) / c->slices & cmask;
/*
* Write the huffman codes to a buffer,
* get the offset in bytes.
*/
offset += write_huff_codes(dst + sstart * width, c->slice_bits,
width * height + 4, width,
send - sstart, he);
slice_len = offset - slice_len;
/* Byteswap the written huffman codes */
c->bdsp.bswap_buf((uint32_t *) c->slice_bits,
(uint32_t *) c->slice_bits,
slice_len >> 2);
/* Write the offset to the stream */
bytestream2_put_le32(pb, offset);
/* Seek to the data part of the packet */
bytestream2_seek_p(pb, 4 * (c->slices - i - 1) +
offset - slice_len, SEEK_CUR);
/* Write the slices' data into the output packet */
bytestream2_put_buffer(pb, c->slice_bits, slice_len);
/* Seek back to the slice offsets */
bytestream2_seek_p(pb, -4 * (c->slices - i - 1) - offset,
SEEK_CUR);
slice_len = offset;
}
/* And at the end seek to the end of written slice(s) */
bytestream2_seek_p(pb, offset, SEEK_CUR);
return 0;
}
static int utvideo_encode_frame(AVCodecContext *avctx, AVPacket *pkt,
const AVFrame *pic, int *got_packet)
{
UtvideoContext *c = avctx->priv_data;
PutByteContext pb;
uint32_t frame_info;
uint8_t *dst;
int width = avctx->width, height = avctx->height;
int i, ret = 0;
/* Allocate a new packet if needed, and set it to the pointer dst */
ret = ff_alloc_packet(avctx, pkt, (256 + 4 * c->slices + width * height)
* c->planes + 4);
if (ret < 0)
return ret;
dst = pkt->data;
bytestream2_init_writer(&pb, dst, pkt->size);
av_fast_padded_malloc(&c->slice_bits, &c->slice_bits_size, width * height + 4);
if (!c->slice_bits) {
av_log(avctx, AV_LOG_ERROR, "Cannot allocate temporary buffer 2.\n");
return AVERROR(ENOMEM);
}
/* In case of RGB, mangle the planes to Ut Video's format */
if (avctx->pix_fmt == AV_PIX_FMT_GBRAP || avctx->pix_fmt == AV_PIX_FMT_GBRP)
mangle_rgb_planes(c->slice_buffer, c->slice_stride, pic->data,
c->planes, pic->linesize, width, height);
/* Deal with the planes */
switch (avctx->pix_fmt) {
case AV_PIX_FMT_GBRP:
case AV_PIX_FMT_GBRAP:
for (i = 0; i < c->planes; i++) {
ret = encode_plane(avctx, c->slice_buffer[i] + 2 * c->slice_stride,
c->slice_buffer[i], c->slice_stride, i,
width, height, &pb);
if (ret) {
av_log(avctx, AV_LOG_ERROR, "Error encoding plane %d.\n", i);
return ret;
}
}
break;
case AV_PIX_FMT_YUV444P:
for (i = 0; i < c->planes; i++) {
ret = encode_plane(avctx, pic->data[i], c->slice_buffer[0],
pic->linesize[i], i, width, height, &pb);
if (ret) {
av_log(avctx, AV_LOG_ERROR, "Error encoding plane %d.\n", i);
return ret;
}
}
break;
case AV_PIX_FMT_YUV422P:
for (i = 0; i < c->planes; i++) {
ret = encode_plane(avctx, pic->data[i], c->slice_buffer[0],
pic->linesize[i], i, width >> !!i, height, &pb);
if (ret) {
av_log(avctx, AV_LOG_ERROR, "Error encoding plane %d.\n", i);
return ret;
}
}
break;
case AV_PIX_FMT_YUV420P:
for (i = 0; i < c->planes; i++) {
ret = encode_plane(avctx, pic->data[i], c->slice_buffer[0],
pic->linesize[i], i, width >> !!i, height >> !!i,
&pb);
if (ret) {
av_log(avctx, AV_LOG_ERROR, "Error encoding plane %d.\n", i);
return ret;
}
}
break;
default:
av_log(avctx, AV_LOG_ERROR, "Unknown pixel format: %d\n",
avctx->pix_fmt);
return AVERROR_INVALIDDATA;
}
/*
* Write frame information (LE 32-bit unsigned)
* into the output packet.
* Contains the prediction method.
*/
frame_info = c->frame_pred << 8;
bytestream2_put_le32(&pb, frame_info);
/*
* At least currently Ut Video is IDR only.
* Set flags accordingly.
*/
pkt->flags |= AV_PKT_FLAG_KEY;
pkt->size = bytestream2_tell_p(&pb);
/* Packet should be done */
*got_packet = 1;
return 0;
}
#define OFFSET(x) offsetof(UtvideoContext, x)
#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
static const AVOption options[] = {
{ "pred", "Prediction method", OFFSET(frame_pred), AV_OPT_TYPE_INT, { .i64 = PRED_LEFT }, PRED_NONE, PRED_MEDIAN, VE, "pred" },
{ "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRED_NONE }, INT_MIN, INT_MAX, VE, "pred" },
{ "left", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRED_LEFT }, INT_MIN, INT_MAX, VE, "pred" },
{ "gradient", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRED_GRADIENT }, INT_MIN, INT_MAX, VE, "pred" },
{ "median", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRED_MEDIAN }, INT_MIN, INT_MAX, VE, "pred" },
{ NULL},
};
static const AVClass utvideo_class = {
.class_name = "utvideo",
.item_name = av_default_item_name,
.option = options,
.version = LIBAVUTIL_VERSION_INT,
};
const AVCodec ff_utvideo_encoder = {
.name = "utvideo",
.long_name = NULL_IF_CONFIG_SMALL("Ut Video"),
.type = AVMEDIA_TYPE_VIDEO,
.id = AV_CODEC_ID_UTVIDEO,
.priv_data_size = sizeof(UtvideoContext),
.priv_class = &utvideo_class,
.init = utvideo_encode_init,
.encode2 = utvideo_encode_frame,
.close = utvideo_encode_close,
.capabilities = AV_CODEC_CAP_FRAME_THREADS,
.pix_fmts = (const enum AVPixelFormat[]) {
AV_PIX_FMT_GBRP, AV_PIX_FMT_GBRAP, AV_PIX_FMT_YUV422P,
AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV444P, AV_PIX_FMT_NONE
},
.caps_internal = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_INIT_CLEANUP,
};
Reference in New Issue View Git Blame Copy Permalink