On Fri, Aug 17, 2012 at 01:35:27PM +0300, Jan Ekström wrote:
> Move common code from utvideodec.c to a new common header utvideo.h.
>
> - Slices work, but hardcoded to one for the time being.
> - All related colorspaces work.
> - Interlaced encoding is not yet supported.
> ---
> Changelog | 1 +
> doc/general.texi | 2 +-
> libavcodec/Makefile | 1 +
> libavcodec/allcodecs.c | 2 +-
> libavcodec/utvideo.h | 96 ++++++
> libavcodec/utvideodec.c | 39 +---
> libavcodec/utvideoenc.c | 735
> +++++++++++++++++++++++++++++++++++++++++++++++
> libavcodec/version.h | 2 +-
> 8 files changed, 838 insertions(+), 40 deletions(-)
> create mode 100644 libavcodec/utvideo.h
> create mode 100644 libavcodec/utvideoenc.c
>
> diff --git a/libavcodec/utvideo.h b/libavcodec/utvideo.h
> new file mode 100644
> index 0000000..de20a03
> --- /dev/null
> +++ b/libavcodec/utvideo.h
> @@ -0,0 +1,96 @@
> +/*
> + * Common Ut Video header
> + * Copyright (c) 2011 Konstantin Shishkov
> + *
> + * This file is part of Libav.
> + *
> + * Libav 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.
> + *
> + * Libav 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 Libav; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301
> USA
> + */
> +
> +#ifndef AVCODEC_UTVIDEO_H
> +#define AVCODEC_UTVIDEO_H
> +
> +/**
> + * @file
> + * Common Ut Video header
> + */
> +
> +#include "libavutil/common.h"
> +#include "avcodec.h"
> +#include "dsputil.h"
> +
> +enum {
> + PRED_NONE = 0,
> + PRED_LEFT,
> + PRED_GRADIENT,
> + PRED_MEDIAN,
> +};
> +
> +enum {
> + COMP_NONE = 0,
> + COMP_HUFF,
> +};
> +
> +/*
> + * "Original format" markers.
> + * Based on values gotten from the official VFW encoder.
> + * They are not used during decoding, but they do have
> + * an informative role on seeing what was input
> + * to the encoder.
> + */
> +enum {
> + UTVIDEO_RGB = MKTAG(0x00, 0x00, 0x01, 0x18),
> + UTVIDEO_RGBA = MKTAG(0x00, 0x00, 0x02, 0x18),
> + UTVIDEO_420 = MKTAG('Y', 'V', '1', '2'),
> + UTVIDEO_422 = MKTAG('Y', 'U', 'Y', '2'),
> +};
> +
> +/* Mapping of libavcodec prediction modes to Ut Video's */
> +static const int pred_order[5] = {
> + PRED_LEFT, PRED_MEDIAN, PRED_MEDIAN, PRED_NONE, PRED_GRADIENT
> +};
> +
> +static const int rgb_order[4] = { 1, 2, 0, 3 }; // G, B, R, A
> +
> +typedef struct UtvideoContext {
> + AVCodecContext *avctx;
> + AVFrame pic;
> + DSPContext dsp;
> +
> + uint32_t frame_info_size, flags, frame_info;
> + int planes;
> + int slices;
> + int compression;
> + int interlaced;
> + int frame_pred;
> +
> + uint8_t *slice_bits, *slice_buffer;
> + int slice_bits_size;
> +} UtvideoContext;
> +
> +typedef struct HuffEntry {
> + uint8_t sym;
> + uint8_t len;
> + uint32_t code;
> +} HuffEntry;
> +
> +/* Compares huffentries' lengths */
Huffman tree nodes?
> +static int 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;
> +}
> +
> +#endif /* AVCODEC_UTVIDEO_H */
> diff --git a/libavcodec/utvideodec.c b/libavcodec/utvideodec.c
> index 27980d8..31f2817 100644
> --- a/libavcodec/utvideodec.c
> +++ b/libavcodec/utvideodec.c
> @@ -32,40 +32,7 @@
> #include "get_bits.h"
> #include "dsputil.h"
> #include "thread.h"
> -
> -enum {
> - PRED_NONE = 0,
> - PRED_LEFT,
> - PRED_GRADIENT,
> - PRED_MEDIAN,
> -};
> -
> -typedef struct UtvideoContext {
> - AVCodecContext *avctx;
> - AVFrame pic;
> - DSPContext dsp;
> -
> - uint32_t frame_info_size, flags, frame_info;
> - int planes;
> - int slices;
> - int compression;
> - int interlaced;
> - int frame_pred;
> -
> - uint8_t *slice_bits;
> - int slice_bits_size;
> -} UtvideoContext;
> -
> -typedef struct HuffEntry {
> - uint8_t sym;
> - uint8_t len;
> -} HuffEntry;
> -
> -static int huff_cmp(const void *a, const void *b)
> -{
> - const HuffEntry *aa = a, *bb = b;
> - return (aa->len - bb->len)*256 + aa->sym - bb->sym;
> -}
> +#include "utvideo.h"
>
> static int build_huff(const uint8_t *src, VLC *vlc, int *fsym)
> {
> @@ -82,7 +49,7 @@ static int build_huff(const uint8_t *src, VLC *vlc, int
> *fsym)
> he[i].sym = i;
> he[i].len = *src++;
> }
> - qsort(he, 256, sizeof(*he), huff_cmp);
> + qsort(he, 256, sizeof(*he), huff_cmp_len);
>
> if (!he[0].len) {
> *fsym = he[0].sym;
> @@ -216,8 +183,6 @@ fail:
> return AVERROR_INVALIDDATA;
> }
>
> -static const int rgb_order[4] = { 1, 2, 0, 3 };
> -
> static void restore_rgb_planes(uint8_t *src, int step, int stride, int width,
> int height)
> {
> diff --git a/libavcodec/utvideoenc.c b/libavcodec/utvideoenc.c
> new file mode 100644
> index 0000000..dcd4b3b
> --- /dev/null
> +++ b/libavcodec/utvideoenc.c
> @@ -0,0 +1,735 @@
> +/*
> + * Ut Video encoder
> + * Copyright (c) 2012 Jan Ekström
> + *
> + * This file is part of Libav.
> + *
> + * Libav 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.
> + *
> + * Libav 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 Libav; 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/intreadwrite.h"
> +#include "avcodec.h"
> +#include "internal.h"
> +#include "bytestream.h"
> +#include "put_bits.h"
> +#include "dsputil.h"
> +#include "mathops.h"
> +#include "utvideo.h"
> +
> +/* 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);
useless parentheses
> +}
> +
> +static av_cold int utvideo_encode_close(AVCodecContext *avctx)
> +{
> + UtvideoContext *c = avctx->priv_data;
> +
> + av_freep(&avctx->coded_frame);
> + av_freep(&c->slice_bits);
> + av_freep(&c->slice_buffer);
> +
> + return 0;
> +}
> +
> +static av_cold int utvideo_encode_init(AVCodecContext *avctx)
> +{
> + UtvideoContext *c = avctx->priv_data;
> +
> + uint32_t original_format;
> +
> + c->avctx = avctx;
> + c->frame_info_size = 4;
> +
> + switch (avctx->pix_fmt) {
> + case PIX_FMT_RGB24:
> + c->planes = 3;
> + avctx->codec_tag = MKTAG('U', 'L', 'R', 'G');
> + original_format = UTVIDEO_RGB;
> + break;
> + case PIX_FMT_RGBA:
> + c->planes = 4;
> + avctx->codec_tag = MKTAG('U', 'L', 'R', 'A');
> + original_format = UTVIDEO_RGBA;
> + break;
> + case 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;
> + avctx->codec_tag = MKTAG('U', 'L', 'Y', '0');
> + original_format = UTVIDEO_420;
> + break;
> + case 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;
> + avctx->codec_tag = MKTAG('U', 'L', 'Y', '2');
> + original_format = UTVIDEO_422;
> + break;
> + default:
> + av_log(avctx, AV_LOG_ERROR, "Unknown pixel format: %d\n",
> + avctx->pix_fmt);
> + return AVERROR_INVALIDDATA;
> + }
> +
> + ff_dsputil_init(&c->dsp, avctx);
> +
> + /* Check the prediction method, and error out if unsupported */
> + if (avctx->prediction_method < 0 || avctx->prediction_method > 4) {
> + av_log(avctx, AV_LOG_WARNING,
> + "Prediction method %d is not supported in Ut Video.\n",
> + avctx->prediction_method);
> + return AVERROR_OPTION_NOT_FOUND;
> + }
> +
> + if (avctx->prediction_method == FF_PRED_PLANE) {
> + av_log(avctx, AV_LOG_ERROR,
> + "Plane prediction is not supported in Ut Video.\n");
> + return AVERROR_OPTION_NOT_FOUND;
> + }
> +
> + /* Convert from libavcodec prediction type to Ut Video's */
> + c->frame_pred = pred_order[avctx->prediction_method];
> +
> + if (c->frame_pred == PRED_GRADIENT) {
> + av_log(avctx, AV_LOG_ERROR, "Gradient prediction is not
> supported.\n");
> + return AVERROR_OPTION_NOT_FOUND;
> + }
> +
> + avctx->coded_frame = avcodec_alloc_frame();
> +
> + if (!avctx->coded_frame) {
> + av_log(avctx, AV_LOG_ERROR, "Could not allocate frame.\n");
> + utvideo_encode_close(avctx);
> + return AVERROR(ENOMEM);
> + }
> +
> + /* extradata size is 4 * 32bit */
> + avctx->extradata_size = 16;
> +
> + avctx->extradata = av_mallocz(avctx->extradata_size +
> + FF_INPUT_BUFFER_PADDING_SIZE);
> +
> + if (!avctx->extradata) {
> + av_log(avctx, AV_LOG_ERROR, "Could not allocate extradata.\n");
> + utvideo_encode_close(avctx);
> + return AVERROR(ENOMEM);
> + }
> +
> + c->slice_buffer = av_malloc(avctx->width * avctx->height +
> + FF_INPUT_BUFFER_PADDING_SIZE);
> +
> + if (!c->slice_buffer) {
> + av_log(avctx, AV_LOG_ERROR, "Cannot allocate temporary buffer 1.\n");
> + utvideo_encode_close(avctx);
> + 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.
> + * Set one slice for now.
> + */
> + c->slices = 1;
> +
> + /* Set compression mode */
> + c->compression = COMP_HUFF;
> +
> + /*
> + * Set the encoding flags:
> + * - Slice count minus 1
> + * - Interlacing (0/1), set to zero for now.
> + * - Compression mode (none/huff)
> + * And write the flags.
> + */
> + c->flags = ((uint8_t)(c->slices - 1)) << 24;
> + c->flags |= 0 << 11;
I'd like less mentions of interlaced mode.
> + c->flags |= c->compression;
> +
> + AV_WL32(avctx->extradata + 12, c->flags);
> +
> + return 0;
> +}
> +
> +static void mangle_rgb_planes(uint8_t *src, int step, int stride, int width,
> + int height)
> +{
> + int i, j;
> + uint8_t r, g, b;
> +
> + for (j = 0; j < height; j++) {
> + for (i = 0; i < width * step; i += step) {
> + r = src[i];
> + g = src[i + 1];
> + b = src[i + 2];
> +
> + src[i] = r - g + 0x80;
> + src[i + 2] = b - g + 0x80;
> + }
> + src += stride;
> + }
> +}
> +
> +/* Write data to a plane, no prediction applied */
> +static void write_plane(uint8_t *src, uint8_t *dst, int step, int stride,
> + int width, int height)
> +{
> + int i, j;
> +
> + for (j = 0; j < height; j++) {
> + for (i = 0; i < width * step; i += step)
> + *dst++ = src[i];
> +
> + src += stride;
> + }
> +}
> +
> +/* Write data to a plane with left prediction */
> +static void left_predict(uint8_t *src, uint8_t *dst, int step, int stride,
> + int width, int height)
> +{
> + int i, j;
> + uint8_t prev;
> +
> + prev = 0x80; /* Set the initial value */
> + for (j = 0; j < height; j++) {
> + for (i = 0; i < width * step; i += step) {
> + *dst++ = src[i] - prev;
> + prev = src[i]; // set the last pred. value as previous
drop the comment, the code is trivial to understand
> + }
> + src += stride;
> + }
> +}
> +
> +/* Write data to a plane with median prediction */
> +static void median_predict(uint8_t *src, uint8_t *dst, int step, int stride,
> + int width, int height)
> +{
> + int i, j;
> + int A, B, C;
> + uint8_t prev;
> +
> + /* First line uses left neighbour prediction */
> + prev = 0x80; /* Set the initial value */
> + for (i = 0; i < width * step; i += step) {
> + *dst++ = src[i] - prev;
> + prev = src[i]; /* set the last pred. value as previous */
ditto
> + }
> +
> + if (height == 1)
> + return;
> +
> + src += stride;
> +
> + /*
> + * Second line uses top prediction for the first sample,
> + * and median for the rest.
> + */
> + C = src[-stride];
> + *dst++ = src[0] - C;
> + A = src[0];
> + for (i = step; i < width * step; i += step) {
> + B = src[i - stride];
> + *dst++ = src[i] - mid_pred(A, B, (A + B - C) & 0xFF);
> + C = B;
> + A = src[i];
> + }
> +
> + src += stride;
> +
> + /* Rest of the coded part uses median prediction */
> + for (j = 2; j < height; j++) {
> + for (i = 0; i < width * step; i += step) {
> + B = src[i - stride];
> + *dst++ = src[i] - mid_pred(A, B, (A + B - C) & 0xFF);
> + C = B;
> + A = src[i];
> + }
> + src += stride;
> + }
> +}
> +
[...]
> +
> +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 */
> + if (!pkt->data)
> + ret = av_new_packet(pkt, (256 + 4 * c->slices +
> + width * height) * c->planes + 4);
> +
> + if (ret) {
> + av_log(avctx, AV_LOG_ERROR, "Error allocating output packet.\n");
> + return ret;
> + }
> +
> + dst = pkt->data;
> +
> + bytestream2_init_writer(&pb, dst, (256 + 4 * c->slices +
> + width * height) * c->planes + 4);
> +
> + av_fast_malloc(&c->slice_bits, &c->slice_bits_size,
> + width * height + FF_INPUT_BUFFER_PADDING_SIZE);
> +
> + 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 == PIX_FMT_RGBA || avctx->pix_fmt == PIX_FMT_RGB24)
> + mangle_rgb_planes(pic->data[0], c->planes, pic->linesize[0], width,
> + height);
> +
> + /* Deal with the planes */
> + switch (avctx->pix_fmt) {
> + case PIX_FMT_RGB24:
> + case PIX_FMT_RGBA:
> + for (i = 0; i < c->planes; i++) {
> + ret = encode_plane(avctx, pic->data[0] + rgb_order[i],
> + c->slice_buffer, c->planes, pic->linesize[0],
> + width, height, &pb);
> +
> + if (ret) {
> + av_log(avctx, AV_LOG_ERROR, "Error encoding plane %d.\n", i);
> + return ret;
> + }
> + }
> + break;
> + case PIX_FMT_YUV422P:
> + for (i = 0; i < c->planes; i++) {
> + ret = encode_plane(avctx, pic->data[i], c->slice_buffer, 1,
> + pic->linesize[i], width >>!! i, height, &pb);
weird spacing (and I'm quite sure it's not my mail client mangling it)
> +
> + if (ret) {
> + av_log(avctx, AV_LOG_ERROR, "Error encoding plane %d.\n", i);
> + return ret;
> + }
> + }
> + break;
> + case PIX_FMT_YUV420P:
> + for (i = 0; i < c->planes; i++) {
> + ret = encode_plane(avctx, pic->data[i], c->slice_buffer, 1,
> + pic->linesize[i], width >>!! i, height >>!! i,
ditto
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
