This video decoder outputs PAL8 colorspace data. Interacting with this decoder is a little involved. During initialization, the demuxer must transmit the 65536-byte Huffman table(s) to the decoder via extradata. Then, whenever a palette change is encountered while demuxing the file, the demuxer must use the same extradata space to transmit an AVPaletteControl structure.
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.
Definition in file idcinvideo.c.
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "common.h"
#include "avcodec.h"
#include "dsputil.h"
Go to the source code of this file.
Classes | |
struct | hnode_t |
struct | IdcinContext |
Defines | |
#define | HUF_TOKENS 256 |
#define | HUFFMAN_TABLE_SIZE 64 * 1024 |
#define | PALETTE_COUNT 256 |
Functions | |
static void | huff_build_tree (IdcinContext *s, int prev) |
static int | huff_smallest_node (hnode_t *hnodes, int num_hnodes) |
static int | idcin_decode_end (AVCodecContext *avctx) |
static int | idcin_decode_frame (AVCodecContext *avctx, void *data, int *data_size, uint8_t *buf, int buf_size) |
static int | idcin_decode_init (AVCodecContext *avctx) |
static void | idcin_decode_vlcs (IdcinContext *s) |
Variables | |
AVCodec | idcin_decoder |