implemented Id Quake II CIN support

Originally committed as revision 2339 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Mike Melanson 2003-10-03 05:43:03 +00:00
parent a686caf03d
commit 4120a53a86
10 changed files with 605 additions and 2 deletions

View File

@ -6,6 +6,7 @@ version <next>:
- Cinepak decoder
- Sega FILM (CPK) file demuxer
- Westwood multimedia support (VQA/AUD file demuxer, audio decoder)
- Id Quake II CIN playback support
version 0.4.8:

View File

@ -633,6 +633,8 @@ library:
@tab used in many Sega Saturn console games
@item Westwood Studios VQA/AUD @tab @tab X
@tab Multimedia formats used in Westwood Studios games
@item Id Cinematic (.cin) @tab @tab X
@tab Used in Quake II
@end multitable
@code{X} means that the encoding (resp. decoding) is supported.
@ -692,6 +694,8 @@ following image formats are supported:
@item Cinepak @tab @tab X
@item Microsoft RLE @tab @tab X
@item Microsoft Video-1 @tab @tab X
@item Westwood VQA @tab @tab X
@item Id Cinematic Video @tab @tab X @tab used in Quake II
@end multitable
@code{X} means that the encoding (resp. decoding) is supported.

View File

@ -18,7 +18,7 @@ OBJS= common.o utils.o mem.o allcodecs.o \
fft.o mdct.o mace.o huffyuv.o cyuv.o opts.o raw.o h264.o golomb.o \
vp3.o asv1.o 4xm.o cabac.o ffv1.o ra144.o ra288.o vcr1.o cljr.o \
roqvideo.o dpcm.o interplayvideo.o xan.o rpza.o cinepak.o msrle.o \
msvideo1.o vqavideo.o
msvideo1.o vqavideo.o idcinvideo.o
ifeq ($(AMR_NB),yes)
ifeq ($(AMR_NB_FIXED),yes)

View File

@ -128,6 +128,7 @@ void avcodec_register_all(void)
register_avcodec(&msrle_decoder);
register_avcodec(&msvideo1_decoder);
register_avcodec(&vqa_decoder);
register_avcodec(&idcin_decoder);
#ifdef CONFIG_AC3
register_avcodec(&ac3_decoder);
#endif

View File

@ -79,6 +79,7 @@ enum CodecID {
CODEC_ID_WS_VQA,
CODEC_ID_MSRLE,
CODEC_ID_MSVIDEO1,
CODEC_ID_IDCIN,
/* various pcm "codecs" */
CODEC_ID_PCM_S16LE,
@ -1433,6 +1434,7 @@ extern AVCodec cinepak_decoder;
extern AVCodec msrle_decoder;
extern AVCodec msvideo1_decoder;
extern AVCodec vqa_decoder;
extern AVCodec idcin_decoder;
extern AVCodec ra_144_decoder;
extern AVCodec ra_288_decoder;
extern AVCodec roq_dpcm_decoder;

278
libavcodec/idcinvideo.c Normal file
View File

@ -0,0 +1,278 @@
/*
* Id Quake II CIN Video Decoder
* Copyright (C) 2003 the ffmpeg project
*
* This library 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 of the License, or (at your option) any later version.
*
* This library 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 this library; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
*/
/**
* @file idcinvideo.c
* Id Quake II Cin Video Decoder by Dr. Tim Ferguson
* For more information about the Id CIN format, visit:
* http://www.csse.monash.edu.au/~timf/
*
* This video decoder outputs PAL8 colorspace data.
*
* Id CIN video is purely Huffman-coded, intraframe-only codec. It achieves
* a little more compression by exploiting the fact that adjacent pixels
* tend to be similar.
*
* Note that this decoder could use ffmpeg's optimized VLC facilities
* rather than naive, tree-based Huffman decoding. However, there are 256
* Huffman tables. Plus, the VLC bit coding order is right -> left instead
* or left -> right, so all of the bits would have to be reversed. Further,
* the original Quake II implementation likely used a similar naive
* decoding algorithm and it worked fine on much lower spec machines.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "common.h"
#include "avcodec.h"
#include "dsputil.h"
#define HUFFMAN_TABLE_SIZE 64 * 1024
#define HUF_TOKENS 256
#define PALETTE_COUNT 256
typedef struct
{
int count;
unsigned char used;
int children[2];
} hnode_t;
typedef struct IdcinContext {
AVCodecContext *avctx;
DSPContext dsp;
AVFrame frame;
unsigned char *buf;
int size;
unsigned char palette[PALETTE_COUNT * 4];
hnode_t huff_nodes[256][HUF_TOKENS*2];
int num_huff_nodes[256];
} IdcinContext;
/*
* Find the lowest probability node in a Huffman table, and mark it as
* being assigned to a higher probability.
* Returns the node index of the lowest unused node, or -1 if all nodes
* are used.
*/
static int huff_smallest_node(hnode_t *hnodes, int num_hnodes) {
int i;
int best, best_node;
best = 99999999;
best_node = -1;
for(i = 0; i < num_hnodes; i++) {
if(hnodes[i].used)
continue;
if(!hnodes[i].count)
continue;
if(hnodes[i].count < best) {
best = hnodes[i].count;
best_node = i;
}
}
if(best_node == -1)
return -1;
hnodes[best_node].used = 1;
return best_node;
}
/*
* Build the Huffman tree using the generated/loaded probabilities histogram.
*
* On completion:
* huff_nodes[prev][i < HUF_TOKENS] - are the nodes at the base of the tree.
* huff_nodes[prev][i >= HUF_TOKENS] - are used to construct the tree.
* num_huff_nodes[prev] - contains the index to the root node of the tree.
* That is: huff_nodes[prev][num_huff_nodes[prev]] is the root node.
*/
static void huff_build_tree(IdcinContext *s, int prev) {
hnode_t *node, *hnodes;
int num_hnodes, i;
num_hnodes = HUF_TOKENS;
hnodes = s->huff_nodes[prev];
for(i = 0; i < HUF_TOKENS * 2; i++)
hnodes[i].used = 0;
while (1) {
node = &hnodes[num_hnodes]; /* next free node */
/* pick two lowest counts */
node->children[0] = huff_smallest_node(hnodes, num_hnodes);
if(node->children[0] == -1)
break; /* reached the root node */
node->children[1] = huff_smallest_node(hnodes, num_hnodes);
if(node->children[1] == -1)
break; /* reached the root node */
/* combine nodes probability for new node */
node->count = hnodes[node->children[0]].count +
hnodes[node->children[1]].count;
num_hnodes++;
}
s->num_huff_nodes[prev] = num_hnodes - 1;
}
static int idcin_decode_init(AVCodecContext *avctx)
{
IdcinContext *s = (IdcinContext *)avctx->priv_data;
int i, j, histogram_index = 0;
unsigned char *histograms;
s->avctx = avctx;
avctx->pix_fmt = PIX_FMT_PAL8;
avctx->has_b_frames = 0;
dsputil_init(&s->dsp, avctx);
/* make sure the Huffman tables make it */
if (s->avctx->extradata_size != HUFFMAN_TABLE_SIZE) {
printf(" Id CIN video: expected extradata size of %d\n", HUFFMAN_TABLE_SIZE);
return -1;
}
/* build the 256 Huffman decode trees */
histograms = (unsigned char *)s->avctx->extradata;
for (i = 0; i < 256; i++) {
for(j = 0; j < HUF_TOKENS; j++)
s->huff_nodes[i][j].count = histograms[histogram_index++];
huff_build_tree(s, i);
}
s->frame.data[0] = NULL;
return 0;
}
static void idcin_decode_vlcs(IdcinContext *s)
{
hnode_t *hnodes;
long x, y;
int prev;
unsigned char v = 0;
int bit_pos, node_num, dat_pos;
prev = bit_pos = dat_pos = 0;
for (y = 0; y < (s->frame.linesize[0] * s->avctx->height);
y += s->frame.linesize[0]) {
for (x = y; x < y + s->avctx->width; x++) {
node_num = s->num_huff_nodes[prev];
hnodes = s->huff_nodes[prev];
while(node_num >= HUF_TOKENS) {
if(!bit_pos) {
if(dat_pos > s->size) {
printf("Huffman decode error.\n");
return;
}
bit_pos = 8;
v = s->buf[dat_pos++];
}
node_num = hnodes[node_num].children[v & 0x01];
v = v >> 1;
bit_pos--;
}
s->frame.data[0][x] = node_num;
prev = node_num;
}
}
}
static int idcin_decode_frame(AVCodecContext *avctx,
void *data, int *data_size,
uint8_t *buf, int buf_size)
{
IdcinContext *s = (IdcinContext *)avctx->priv_data;
AVPaletteControl *palette_control =
(AVPaletteControl *)avctx->extradata;
int i;
unsigned int *palette32;
int palette_index = 0;
unsigned char r, g, b;
s->buf = buf;
s->size = buf_size;
if (palette_control->palette_changed) {
palette32 = (unsigned int *)s->palette;
for (i = 0; i < PALETTE_COUNT; i++) {
r = palette_control->palette[palette_index++] * 1;
g = palette_control->palette[palette_index++] * 1;
b = palette_control->palette[palette_index++] * 1;
palette32[i] = (r << 16) | (g << 8) | (b);
}
palette_control->palette_changed = 0;
}
if (s->frame.data[0])
avctx->release_buffer(avctx, &s->frame);
if (avctx->get_buffer(avctx, &s->frame)) {
printf (" Id CIN Video: get_buffer() failed\n");
return -1;
}
idcin_decode_vlcs(s);
/* make the palette available on the way out */
memcpy(s->frame.data[1], s->palette, PALETTE_COUNT * 4);
*data_size = sizeof(AVFrame);
*(AVFrame*)data = s->frame;
/* report that the buffer was completely consumed */
return buf_size;
}
static int idcin_decode_end(AVCodecContext *avctx)
{
IdcinContext *s = (IdcinContext *)avctx->priv_data;
if (s->frame.data[0])
avctx->release_buffer(avctx, &s->frame);
return 0;
}
AVCodec idcin_decoder = {
"idcinvideo",
CODEC_TYPE_VIDEO,
CODEC_ID_IDCIN,
sizeof(IdcinContext),
idcin_decode_init,
NULL,
idcin_decode_end,
idcin_decode_frame,
CODEC_CAP_DR1,
};

View File

@ -15,7 +15,7 @@ PPOBJS=
OBJS+=mpeg.o mpegts.o mpegtsenc.o ffm.o crc.o img.o raw.o rm.o \
avienc.o avidec.o wav.o swf.o au.o gif.o mov.o mpjpeg.o dv.o \
yuv4mpeg.o 4xm.o flvenc.o flvdec.o movenc.o psxstr.o idroq.o ipmovie.o \
nut.o wc3movie.o mp3.o westwood.o segafilm.o
nut.o wc3movie.o mp3.o westwood.o segafilm.o idcin.o
ifeq ($(CONFIG_RISKY),yes)
OBJS+= asf.o

View File

@ -58,6 +58,7 @@ void av_register_all(void)
wc3_init();
westwood_init();
film_init();
idcin_init();
#if defined(AMR_NB) || defined(AMR_NB_FIXED) || defined(AMR_WB)
amr_init();

View File

@ -412,6 +412,9 @@ int westwood_init(void);
/* segafilm.c */
int film_init(void);
/* idcin.c */
int idcin_init(void);
#include "rtp.h"
#include "rtsp.h"

313
libavformat/idcin.c Normal file
View File

@ -0,0 +1,313 @@
/*
* Id Quake II CIN File Demuxer
* Copyright (c) 2003 The ffmpeg Project
*
* This library 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 of the License, or (at your option) any later version.
*
* This library 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 this library; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
/**
* @file idcin.c
* Id Quake II CIN file demuxer by Mike Melanson (melanson@pcisys.net)
* For more information about the Id CIN format, visit:
* http://www.csse.monash.edu.au/~timf/
*
* CIN is a somewhat quirky and ill-defined format. Here are some notes
* for anyone trying to understand the technical details of this format:
*
* The format has no definite file signature. This is problematic for a
* general-purpose media player that wants to automatically detect file
* types. However, a CIN file does start with 5 32-bit numbers that
* specify audio and video parameters. This demuxer gets around the lack
* of file signature by performing sanity checks on those parameters.
* Probabalistically, this is a reasonable solution since the number of
* valid combinations of the 5 parameters is a very small subset of the
* total 160-bit number space.
*
* Refer to the function idcin_probe() for the precise A/V parameters
* that this demuxer allows.
*
* Next, each audio and video frame has a duration of 1/14 sec. If the
* audio sample rate is a multiple of the common frequency 22050 Hz it will
* divide evenly by 14. However, if the sample rate is 11025 Hz:
* 11025 (samples/sec) / 14 (frames/sec) = 787.5 (samples/frame)
* The way the CIN stores audio in this case is by storing 787 sample
* frames in the first audio frame and 788 sample frames in the second
* audio frame. Therefore, the total number of bytes in an audio frame
* is given as:
* audio frame #0: 787 * (bytes/sample) * (# channels) bytes in frame
* audio frame #1: 788 * (bytes/sample) * (# channels) bytes in frame
* audio frame #2: 787 * (bytes/sample) * (# channels) bytes in frame
* audio frame #3: 788 * (bytes/sample) * (# channels) bytes in frame
*
* Finally, not all Id CIN creation tools agree on the resolution of the
* color palette, apparently. Some creation tools specify red, green, and
* blue palette components in terms of 6-bit VGA color DAC values which
* range from 0..63. Other tools specify the RGB components as full 8-bit
* values that range from 0..255. Since there are no markers in the file to
* differentiate between the two variants, this demuxer uses the following
* heuristic:
* - load the 768 palette bytes from disk
* - assume that they will need to be shifted left by 2 bits to
* transform them from 6-bit values to 8-bit values
* - scan through all 768 palette bytes
* - if any bytes exceed 63, do not shift the bytes at all before
* transmitting them to the video decoder
*/
#include "avformat.h"
#define LE_16(x) ((((uint8_t*)(x))[1] << 8) | ((uint8_t*)(x))[0])
#define LE_32(x) ((((uint8_t*)(x))[3] << 24) | \
(((uint8_t*)(x))[2] << 16) | \
(((uint8_t*)(x))[1] << 8) | \
((uint8_t*)(x))[0])
#define HUFFMAN_TABLE_SIZE (64 * 1024)
#define FRAME_PTS_INC (90000 / 14)
typedef struct IdcinDemuxContext {
int video_stream_index;
int audio_stream_index;
int audio_chunk_size1;
int audio_chunk_size2;
/* demux state variables */
int current_audio_chunk;
int next_chunk_is_video;
int audio_present;
int64_t pts;
/* keep reference to extradata but never free it */
void *extradata;
} IdcinDemuxContext;
static int idcin_probe(AVProbeData *p)
{
unsigned int number;
/*
* This is what you could call a "probabilistic" file check: Id CIN
* files don't have a definite file signature. In lieu of such a marker,
* perform sanity checks on the 5 32-bit header fields:
* width, height: greater than 0, less than or equal to 1024
* audio sample rate: greater than or equal to 8000, less than or
* equal to 48000, or 0 for no audio
* audio sample width (bytes/sample): 0 for no audio, or 1 or 2
* audio channels: 0 for no audio, or 1 or 2
*/
/* cannot proceed without 20 bytes */
if (p->buf_size < 20)
return 0;
/* check the video width */
number = LE_32(&p->buf[0]);
if ((number == 0) || (number > 1024))
return 0;
/* check the video height */
number = LE_32(&p->buf[4]);
if ((number == 0) || (number > 1024))
return 0;
/* check the audio sample rate */
number = LE_32(&p->buf[8]);
if ((number != 0) && ((number < 8000) | (number > 48000)))
return 0;
/* check the audio bytes/sample */
number = LE_32(&p->buf[12]);
if (number > 2)
return 0;
/* check the audio channels */
number = LE_32(&p->buf[16]);
if (number > 2)
return 0;
/* return half certainly since this check is a bit sketchy */
return AVPROBE_SCORE_MAX / 2;
}
static int idcin_read_header(AVFormatContext *s,
AVFormatParameters *ap)
{
ByteIOContext *pb = &s->pb;
IdcinDemuxContext *idcin = (IdcinDemuxContext *)s->priv_data;
AVStream *st;
unsigned int width, height;
unsigned int sample_rate, bytes_per_sample, channels;
/* get the 5 header parameters */
width = get_le32(pb);
height = get_le32(pb);
sample_rate = get_le32(pb);
bytes_per_sample = get_le32(pb);
channels = get_le32(pb);
st = av_new_stream(s, 0);
if (!st)
return AVERROR_NOMEM;
idcin->video_stream_index = st->index;
st->codec.codec_type = CODEC_TYPE_VIDEO;
st->codec.codec_id = CODEC_ID_IDCIN;
st->codec.codec_tag = 0; /* no fourcc */
st->codec.width = width;
st->codec.height = height;
/* load up the Huffman tables into extradata */
st->codec.extradata_size = HUFFMAN_TABLE_SIZE;
st->codec.extradata = av_malloc(HUFFMAN_TABLE_SIZE);
if (get_buffer(pb, st->codec.extradata, HUFFMAN_TABLE_SIZE) !=
HUFFMAN_TABLE_SIZE)
return -EIO;
/* save a reference in order to transport the palette */
idcin->extradata = st->codec.extradata;
/* if sample rate is 0, assume no audio */
if (sample_rate) {
idcin->audio_present = 1;
st = av_new_stream(s, 0);
if (!st)
return AVERROR_NOMEM;
idcin->audio_stream_index = st->index;
st->codec.codec_type = CODEC_TYPE_AUDIO;
st->codec.codec_tag = 1;
st->codec.channels = channels;
st->codec.sample_rate = sample_rate;
st->codec.bits_per_sample = bytes_per_sample * 8;
st->codec.bit_rate = sample_rate * bytes_per_sample * 8 * channels;
st->codec.block_align = bytes_per_sample * channels;
if (bytes_per_sample == 1)
st->codec.codec_id = CODEC_ID_PCM_U8;
else
st->codec.codec_id = CODEC_ID_PCM_S16LE;
if (sample_rate % 14 != 0) {
idcin->audio_chunk_size1 = (sample_rate / 14) *
bytes_per_sample * channels;
idcin->audio_chunk_size2 = (sample_rate / 14 + 1) *
bytes_per_sample * channels;
} else {
idcin->audio_chunk_size1 = idcin->audio_chunk_size2 =
(sample_rate / 14) * bytes_per_sample * channels;
}
idcin->current_audio_chunk = 0;
} else
idcin->audio_present = 1;
idcin->next_chunk_is_video = 1;
idcin->pts = 0;
/* set the pts reference (1 pts = 1/90000) */
s->pts_num = 1;
s->pts_den = 90000;
return 0;
}
static int idcin_read_packet(AVFormatContext *s,
AVPacket *pkt)
{
int ret;
unsigned int command;
unsigned int chunk_size;
IdcinDemuxContext *idcin = (IdcinDemuxContext *)s->priv_data;
ByteIOContext *pb = &s->pb;
AVPaletteControl *palette_control = (AVPaletteControl *)idcin->extradata;
int i;
int palette_scale;
if (url_feof(&s->pb))
return -EIO;
if (idcin->next_chunk_is_video) {
command = get_le32(pb);
if (command == 2) {
return -EIO;
} else if (command == 1) {
/* trigger a palette change */
palette_control->palette_changed = 1;
if (get_buffer(pb, palette_control->palette, 768) != 768)
return -EIO;
/* scale the palette as necessary */
palette_scale = 2;
for (i = 0; i < 768; i++)
if (palette_control->palette[i] > 63) {
palette_scale = 0;
break;
}
if (palette_scale)
for (i = 0; i < 768; i++)
palette_control->palette[i] <<= palette_scale;
}
chunk_size = get_le32(pb);
/* skip the number of decoded bytes (always equal to width * height) */
url_fseek(pb, 4, SEEK_CUR);
chunk_size -= 4;
if (av_new_packet(pkt, chunk_size))
ret = -EIO;
pkt->stream_index = idcin->video_stream_index;
pkt->pts = idcin->pts;
ret = get_buffer(pb, pkt->data, chunk_size);
if (ret != chunk_size)
ret = -EIO;
} else {
/* send out the audio chunk */
if (idcin->current_audio_chunk)
chunk_size = idcin->audio_chunk_size2;
else
chunk_size = idcin->audio_chunk_size1;
if (av_new_packet(pkt, chunk_size))
return -EIO;
pkt->stream_index = idcin->audio_stream_index;
pkt->pts = idcin->pts;
ret = get_buffer(&s->pb, pkt->data, chunk_size);
if (ret != chunk_size)
ret = -EIO;
idcin->current_audio_chunk ^= 1;
idcin->pts += FRAME_PTS_INC;
}
if (idcin->audio_present)
idcin->next_chunk_is_video ^= 1;
return ret;
}
static int idcin_read_close(AVFormatContext *s)
{
return 0;
}
static AVInputFormat idcin_iformat = {
"idcin",
"Id CIN format",
sizeof(IdcinDemuxContext),
idcin_probe,
idcin_read_header,
idcin_read_packet,
idcin_read_close,
};
int idcin_init(void)
{
av_register_input_format(&idcin_iformat);
return 0;
}