New fringe codecs: WC3/Xan video, Xan DPCM, DK3 & DK4 ADPCM

Originally committed as revision 2217 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavcodec/Makefile

@@ -18,7 +18,7 @@ OBJS= common.o utils.o mem.o allcodecs.o \
       ratecontrol.o adpcm.o eval.o dv.o error_resilience.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
+      roqvideo.o dpcm.o interplayvideo.o xan.o
 
 ifeq ($(AMR_NB),yes)
 ifeq ($(AMR_NB_FIXED),yes)

libavcodec/adpcm.c

 /*
  * ADPCM codecs
- * Copyright (c) 2001 Fabrice Bellard.
+ * Copyright (c) 2001-2003 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
@@ -22,11 +22,13 @@
  * @file adpcm.c
  * ADPCM codecs.
  * First version by Francois Revol revol@free.fr
+ * Fringe ADPCM codecs (e.g., DK3 and DK4) 
+ *   by Mike Melanson (melanson@pcisys.net)
  *
  * Features and limitations:
  *
  * Reference documents:
- * http://www.pcisys.net/~melanson/codecs/adpcm.txt
+ * http://www.pcisys.net/~melanson/codecs/simpleaudio.html
  * http://www.geocities.com/SiliconValley/8682/aud3.txt
  * http://openquicktime.sourceforge.net/plugins.htm
  * XAnim sources (xa_codec.c) http://www.rasnaimaging.com/people/lapus/download.html
@@ -293,14 +295,10 @@ static inline short adpcm_ima_expand_nibble(ADPCMChannelStatus *c, char nibble)
 
     sign = nibble & 8;
     delta = nibble & 7;
-#if 0
-    diff = step >> 3;
-    if (delta & 4) diff += step;
-    if (delta & 2) diff += step >> 1;
-    if (delta & 1) diff += step >> 2;
-#else
-    diff = ((2 * delta + 1) * step) >> 3; // no jumps
-#endif
+    /* perform direct multiplication instead of series of jumps proposed by
+     * the reference ADPCM implementation since modern CPUs can do the mults
+     * quickly enough */
+    diff = ((2 * delta + 1) * step) >> 3;
     predictor = c->predictor;
     if (sign) predictor -= diff;
     else predictor += diff;
@@ -355,6 +353,21 @@ static inline short adpcm_ms_expand_nibble(ADPCMChannelStatus *c, char nibble)
     return (short)predictor;
 }
 
+/* DK3 ADPCM support macro */
+#define DK3_GET_NEXT_NIBBLE() \
+    if (decode_top_nibble_next) \
+    { \
+        nibble = (last_byte >> 4) & 0x0F; \
+        decode_top_nibble_next = 0; \
+    } \
+    else \
+    { \
+        last_byte = *src++; \
+        if (src >= buf + buf_size) break; \
+        nibble = last_byte & 0x0F; \
+        decode_top_nibble_next = 1; \
+    }
+
 static int adpcm_decode_frame(AVCodecContext *avctx,
 			    void *data, int *data_size,
 			    uint8_t *buf, int buf_size)
@@ -367,6 +380,12 @@ static int adpcm_decode_frame(AVCodecContext *avctx,
     uint8_t *src;
     int st; /* stereo */
 
+    /* DK3 ADPCM accounting variables */
+    unsigned char last_byte = 0;
+    unsigned char nibble;
+    int decode_top_nibble_next = 0;
+    int diff_channel;
+
     samples = data;
     src = buf;
 
@@ -551,6 +570,94 @@ static int adpcm_decode_frame(AVCodecContext *avctx,
             src ++;
         }
         break;
+    case CODEC_ID_ADPCM_IMA_DK4:
+        if (buf_size > BLKSIZE) {
+            if (avctx->block_align != 0)
+                buf_size = avctx->block_align;
+            else
+                buf_size = BLKSIZE;
+        }
+        c->status[0].predictor = (src[0] | (src[1] << 8));
+        c->status[0].step_index = src[2];
+        src += 4;
+        if(c->status[0].predictor & 0x8000)
+            c->status[0].predictor -= 0x10000;
+        *samples++ = c->status[0].predictor;
+        if (st) {
+            c->status[1].predictor = (src[0] | (src[1] << 8));
+            c->status[1].step_index = src[2];
+            src += 4;
+            if(c->status[1].predictor & 0x8000)
+                c->status[1].predictor -= 0x10000;
+            *samples++ = c->status[1].predictor;
+        }
+        while (src < buf + buf_size) {
+
+            /* take care of the top nibble (always left or mono channel) */
+            *samples++ = adpcm_ima_expand_nibble(&c->status[0], 
+                (src[0] >> 4) & 0x0F);
+
+            /* take care of the bottom nibble, which is right sample for
+             * stereo, or another mono sample */
+            if (st)
+                *samples++ = adpcm_ima_expand_nibble(&c->status[1], 
+                    src[0] & 0x0F);
+            else
+                *samples++ = adpcm_ima_expand_nibble(&c->status[0], 
+                    src[0] & 0x0F);
+
+            src++;
+        }
+        break;
+    case CODEC_ID_ADPCM_IMA_DK3:
+        if (buf_size > BLKSIZE) {
+            if (avctx->block_align != 0)
+                buf_size = avctx->block_align;
+            else
+                buf_size = BLKSIZE;
+        }
+        c->status[0].predictor = (src[10] | (src[11] << 8));
+        c->status[1].predictor = (src[12] | (src[13] << 8));
+        c->status[0].step_index = src[14];
+        c->status[1].step_index = src[15];
+        /* sign extend the predictors */
+        if(c->status[0].predictor & 0x8000)
+            c->status[0].predictor -= 0x10000;
+        if(c->status[1].predictor & 0x8000)
+            c->status[1].predictor -= 0x10000;
+        src += 16;
+        diff_channel = c->status[1].predictor;
+
+        /* the DK3_GET_NEXT_NIBBLE macro issues the break statement when
+         * the buffer is consumed */
+        while (1) {
+
+            /* for this algorithm, c->status[0] is the sum channel and
+             * c->status[1] is the diff channel */
+
+            /* process the first predictor of the sum channel */
+            DK3_GET_NEXT_NIBBLE();
+            adpcm_ima_expand_nibble(&c->status[0], nibble);
+
+            /* process the diff channel predictor */
+            DK3_GET_NEXT_NIBBLE();
+            adpcm_ima_expand_nibble(&c->status[1], nibble);
+
+            /* process the first pair of stereo PCM samples */
+            diff_channel = (diff_channel + c->status[1].predictor) / 2;
+            *samples++ = c->status[0].predictor + c->status[1].predictor;
+            *samples++ = c->status[0].predictor - c->status[1].predictor;
+
+            /* process the second predictor of the sum channel */
+            DK3_GET_NEXT_NIBBLE();
+            adpcm_ima_expand_nibble(&c->status[0], nibble);
+
+            /* process the second pair of stereo PCM samples */
+            diff_channel = (diff_channel + c->status[1].predictor) / 2;
+            *samples++ = c->status[0].predictor + c->status[1].predictor;
+            *samples++ = c->status[0].predictor - c->status[1].predictor;
+        }
+        break;
     default:
         *data_size = 0;
         return -1;
@@ -583,8 +690,9 @@ AVCodec name ## _decoder = {                    \
 
 ADPCM_CODEC(CODEC_ID_ADPCM_IMA_QT, adpcm_ima_qt);
 ADPCM_CODEC(CODEC_ID_ADPCM_IMA_WAV, adpcm_ima_wav);
+ADPCM_CODEC(CODEC_ID_ADPCM_IMA_DK3, adpcm_ima_dk3);
+ADPCM_CODEC(CODEC_ID_ADPCM_IMA_DK4, adpcm_ima_dk4);
 ADPCM_CODEC(CODEC_ID_ADPCM_MS, adpcm_ms);
 ADPCM_CODEC(CODEC_ID_ADPCM_4XM, adpcm_4xm);
 
 #undef ADPCM_CODEC
-

libavcodec/allcodecs.c

@@ -123,6 +123,7 @@ void avcodec_register_all(void)
     register_avcodec(&mdec_decoder);
     register_avcodec(&roq_decoder);
     register_avcodec(&interplay_video_decoder);
+    register_avcodec(&xan_wc3_decoder);
 #ifdef CONFIG_AC3
     register_avcodec(&ac3_decoder);
 #endif
@@ -130,6 +131,7 @@ void avcodec_register_all(void)
     register_avcodec(&ra_288_decoder);
     register_avcodec(&roq_dpcm_decoder);
     register_avcodec(&interplay_dpcm_decoder);
+    register_avcodec(&xan_dpcm_decoder);
 #endif /* CONFIG_DECODERS */
 
 #ifdef AMR_NB
@@ -154,6 +156,8 @@ PCM_CODEC(CODEC_ID_PCM_MULAW, pcm_mulaw);
     /* adpcm codecs */
 PCM_CODEC(CODEC_ID_ADPCM_IMA_QT, adpcm_ima_qt);
 PCM_CODEC(CODEC_ID_ADPCM_IMA_WAV, adpcm_ima_wav);
+PCM_CODEC(CODEC_ID_ADPCM_IMA_DK3, adpcm_ima_dk3);
+PCM_CODEC(CODEC_ID_ADPCM_IMA_DK4, adpcm_ima_dk4);
 PCM_CODEC(CODEC_ID_ADPCM_MS, adpcm_ms);
 PCM_CODEC(CODEC_ID_ADPCM_4XM, adpcm_4xm);
 

libavcodec/avcodec.h

@@ -68,6 +68,8 @@ enum CodecID {
     CODEC_ID_MDEC,
     CODEC_ID_ROQ,
     CODEC_ID_INTERPLAY_VIDEO,
+    CODEC_ID_XAN_WC3,
+    CODEC_ID_XAN_WC4,
 
     /* various pcm "codecs" */
     CODEC_ID_PCM_S16LE,
@@ -82,6 +84,8 @@ enum CodecID {
     /* various adpcm codecs */
     CODEC_ID_ADPCM_IMA_QT,
     CODEC_ID_ADPCM_IMA_WAV,
+    CODEC_ID_ADPCM_IMA_DK3,
+    CODEC_ID_ADPCM_IMA_DK4,
     CODEC_ID_ADPCM_MS,
     CODEC_ID_ADPCM_4XM,
 
@@ -94,6 +98,7 @@ enum CodecID {
     /* various DPCM codecs */
     CODEC_ID_ROQ_DPCM,
     CODEC_ID_INTERPLAY_DPCM,
+    CODEC_ID_XAN_DPCM,
 };
 #define CODEC_ID_MPEGVIDEO CODEC_ID_MPEG1VIDEO
 
@@ -1356,10 +1361,12 @@ extern AVCodec fourxm_decoder;
 extern AVCodec mdec_decoder;
 extern AVCodec roq_decoder;
 extern AVCodec interplay_video_decoder;
+extern AVCodec xan_wc3_decoder;
 extern AVCodec ra_144_decoder;
 extern AVCodec ra_288_decoder;
 extern AVCodec roq_dpcm_decoder;
 extern AVCodec interplay_dpcm_decoder;
+extern AVCodec xan_dpcm_decoder;
 
 /* pcm codecs */
 #define PCM_CODEC(id, name) \
@@ -1379,6 +1386,8 @@ PCM_CODEC(CODEC_ID_PCM_MULAW, pcm_mulaw);
 
 PCM_CODEC(CODEC_ID_ADPCM_IMA_QT, adpcm_ima_qt);
 PCM_CODEC(CODEC_ID_ADPCM_IMA_WAV, adpcm_ima_wav);
+PCM_CODEC(CODEC_ID_ADPCM_IMA_DK3, adpcm_ima_dk3);
+PCM_CODEC(CODEC_ID_ADPCM_IMA_DK4, adpcm_ima_dk4);
 PCM_CODEC(CODEC_ID_ADPCM_MS, adpcm_ms);
 PCM_CODEC(CODEC_ID_ADPCM_4XM, adpcm_4xm);
 

libavcodec/dpcm.c

@@ -21,8 +21,17 @@
  * @file: dpcm.c
  * Assorted DPCM (differential pulse code modulation) audio codecs
  * by Mike Melanson (melanson@pcisys.net)
+ * Xan DPCM decoder by Mario Brito (mbrito@student.dei.uc.pt)
  * for more information on the specific data formats, visit:
  *   http://www.pcisys.net/~melanson/codecs/simpleaudio.html
+ *
+ * Note about using the Xan DPCM decoder: Xan DPCM is used in AVI files
+ * found in the Wing Commander IV computer game. These AVI files contain
+ * WAVEFORMAT headers which report the audio format as 0x01: raw PCM.
+ * Clearly incorrect. To detect Xan DPCM, you will probably have to
+ * special-case your AVI demuxer to use Xan DPCM if the file uses 'Xxan'
+ * (Xan video) for its video codec. Alternately, such AVI files also contain
+ * the fourcc 'Axan' in the 'auds' chunk of the AVI header.
  */
 
 #include "avcodec.h"
@@ -115,6 +124,9 @@ static int dpcm_decode_frame(AVCodecContext *avctx,
     int channel_number = 0;
     short *output_samples = data;
     int sequence_number;
+    int shift[2];
+    unsigned char byte;
+    short diff;
 
     switch(avctx->codec->id) {
 
@@ -171,6 +183,40 @@ static int dpcm_decode_frame(AVCodecContext *avctx,
             s->last_delta[i] = predictor[i];
 
         break;
+
+    case CODEC_ID_XAN_DPCM:
+        in = 0;
+        shift[0] = shift[1] = 4;
+        predictor[0] = LE_16(&buf[in]);
+        in += 2;
+        SE_16BIT(predictor[0]);
+        if (s->channels == 2) {
+            predictor[1] = LE_16(&buf[in]);
+            in += 2;
+            SE_16BIT(predictor[1]);
+        }
+
+        while (in < buf_size) {
+            byte = buf[in++];
+            diff = (byte & 0xFC) << 8;
+            if ((byte & 0x03) == 3)
+                shift[channel_number]++;
+            else
+                shift[channel_number] -= (2 * (byte & 3));
+            /* saturate the shifter to a lower limit of 0 */
+            if (shift[channel_number] < 0)
+                shift[channel_number] = 0;
+
+            diff >>= shift[channel_number];
+            predictor[channel_number] += diff;
+
+            SATURATE_S16(predictor[channel_number]);
+            output_samples[out++] = predictor[channel_number];
+
+            /* toggle channel */
+            channel_number ^= s->channels - 1;
+        }
+        break;
     }
 
     *data_size = out * sizeof(short);
@@ -198,3 +244,14 @@ AVCodec interplay_dpcm_decoder = {
     NULL,
     dpcm_decode_frame,
 };
+
+AVCodec xan_dpcm_decoder = {
+    "xan_dpcm",
+    CODEC_TYPE_AUDIO,
+    CODEC_ID_XAN_DPCM,
+    sizeof(DPCMContext),
+    dpcm_decode_init,
+    NULL,
+    NULL,
+    dpcm_decode_frame,
+};