idcin.c File Reference

id Quake II CIN file demuxer by Mike Melanson (melan.nosp@m.son@.nosp@m.pcisy.nosp@m.s.ne.nosp@m.t) For more information about the id CIN format, visit: http://www.csse.monash.edu.au/~timf/ More...

#include "libavutil/channel_layout.h"
#include "libavutil/imgutils.h"
#include "libavutil/intreadwrite.h"
#include "avformat.h"
#include "internal.h"

Classes
struct	IdcinDemuxContext

Macros
#define	HUFFMAN_TABLE_SIZE   (64 * 1024)
#define	IDCIN_FPS   14

Typedefs
typedef struct IdcinDemuxContext	IdcinDemuxContext

Variables
AVInputFormat	ff_idcin_demuxer

Detailed Description

id Quake II CIN file demuxer by Mike Melanson (melan.nosp@m.son@.nosp@m.pcisy.nosp@m.s.ne.nosp@m.t) 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. Probabilistically, 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

Variable Documentation

ff_idcin_demuxer

AVInputFormat ff_idcin_demuxer

Initial value:

= {
    .name           = "idcin",
    .long_name      = NULL_IF_CONFIG_SMALL("id Cinematic"),
    .priv_data_size = sizeof(IdcinDemuxContext),
    .read_probe     = idcin_probe,
    .read_header    = idcin_read_header,
    .read_packet    = idcin_read_packet,
    .read_seek      = idcin_read_seek,
    .flags          = AVFMT_NO_BYTE_SEEK,
}