Re: [FFmpeg-devel] [PATCH] MXF - Add jpeg2000 subdescriptor - Sponsored by INA

["Cédric Le Barz" <clebarz@ektacom.com>]

[-- Attachment #1: Type: text/plain, Size: 3293 bytes --]


Le 05/04/2023 à 15:53, Tomas Härdin a écrit :
> ons 2023-04-05 klockan 15:05 +0200 skrev Cédric Le Barz:
>> Le 03/04/2023 à 17:14, Michael Niedermayer a écrit :
>>> On Mon, Apr 03, 2023 at 10:08:25AM +0200, Cédric Le Barz wrote:
>>>> Hi,
>>>>
>>>> I've attached the patch to this mail, in order to solve newlines
>>> insertion
>>>> issue.
>>> Please make sure each patch also updates the fate tests so
>>> make fate
>>> doesnt fail
>> I've attached to this mail the new patch. Fate test issue is fixed.
>>
> Please avoid top posting.
>
> I was actually about to suggest merging these two patches but I see you
> read my mind :)
>
>> @@ -1131,9 +1164,9 @@ static const UID mxf_aes3_descriptor_key      =
>> { 0x06,0x0E,0x2B,0x34,0x02,0x53,
>>   static const UID mxf_cdci_descriptor_key      = {
>> 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0D,0x01,0x01,0x01,0x01,0x01
>> ,0x28,0x00 };
>>   static const UID mxf_rgba_descriptor_key      = {
>> 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0D,0x01,0x01,0x01,0x01,0x01
>> ,0x29,0x00 };
>>   static const UID mxf_generic_sound_descriptor_key = {
>> 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0D,0x01,0x01,0x01,0x01,0x01
>> ,0x42,0x00 };
>> -
> Stray line deletion
>
>> +    mxf_write_local_tag(s, 2, 0x8401);
>> +    avio_wb16(pb, 0x0000);
>> +    mxf_write_local_tag(s, 4, 0x8402);
>> +    avio_wb32(pb, st->codecpar->width);
>> +    mxf_write_local_tag(s, 4, 0x8403);
>> +    avio_wb32(pb, st->codecpar->height);
>> +    mxf_write_local_tag(s, 4, 0x8404);
>> +    avio_wb32(pb, 0);
>> +    mxf_write_local_tag(s, 4, 0x8405);
>> +    avio_wb32(pb, 0);
>> +    mxf_write_local_tag(s, 4, 0x8406);
>> +    avio_wb32(pb, st->codecpar->width);
>> +    mxf_write_local_tag(s, 4, 0x8407);
>> +    avio_wb32(pb, st->codecpar->height);
>> +    mxf_write_local_tag(s, 4, 0x8408);
>> +    avio_wb32(pb, 0);
>> +    mxf_write_local_tag(s, 4, 0x8409);
>> +    avio_wb32(pb, 0);
>> +    mxf_write_local_tag(s, 2, 0x840A);
>> +    avio_wb16(pb, component_count);
> A comment on each of these explaining what they are would be nice.
>
>> +    {
>> +        char _desc [3][3]= {  {0x09,0x01,0x01} , {0x09,0x02,0x01} ,
>> {0x09,0x02,0x01} };
>> +        int comp = 0;
>> +        for ( comp = 0; comp< component_count ;comp++ ) {
>> +            avio_write(pb, _desc[comp%3] , 3);
>> +        }
>> +    }
> Maybe just a style nit but you could move the char desc[] into the loop
> body, int comp to the start of the function and then you can remove the
> extra {} around this. Also you could make desc static const.
>
>> +    {
>> +        char _layout[16] = {  'Y' , '\n', 'U' , '\n', 'V' , '\n',
>> 'F' , 0x02,
>> +                            0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
>> 0x00, 0x00 };
>> +        avio_write(pb, _layout , 16);
>> +    }
> Again there is the issue of RGB(A)
>
> /Tomas
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".


Please consider this new patch taking into account remarks. For the 
moment, I remove the RGB(A) / YUV code part as it is an optional feature 
for the JPEG2000 subdescriptor.

Regards,

Cédric

[-- Attachment #2: 0001-Add-jpeg2000-subdescriptor-2.eml --]
[-- Type: message/rfc822, Size: 14080 bytes --]

From: Cedric Le Barz <clebarz@ektacom.com>
To: ffmpeg-devel@ffmpeg.org
Subject: [PATCH] Add jpeg2000 subdescriptor (2).
Date: Wed, 5 Apr 2023 11:59:18 +0200

Signed-off-by: Cedric Le Barz <clebarz@ektacom.com>
---
 ffmpeg/libavformat/mxf.h    |   1 +
 ffmpeg/libavformat/mxfenc.c | 169 +++++++++++++++++++++++++++++++++++-
 2 files changed, 167 insertions(+), 3 deletions(-)

diff --git a/ffmpeg/libavformat/mxf.h b/ffmpeg/libavformat/mxf.h
index 2561605..7dd1681 100644
--- a/ffmpeg/libavformat/mxf.h
+++ b/ffmpeg/libavformat/mxf.h
@@ -55,6 +55,7 @@ enum MXFMetadataSetType {
     SoundfieldGroupLabelSubDescriptor,
     GroupOfSoundfieldGroupsLabelSubDescriptor,
     FFV1SubDescriptor,
+    JPEG2000SubDescriptor,
 };
 
 enum MXFFrameLayout {
diff --git a/ffmpeg/libavformat/mxfenc.c b/ffmpeg/libavformat/mxfenc.c
index a29d678..7065a7d 100644
--- a/ffmpeg/libavformat/mxfenc.c
+++ b/ffmpeg/libavformat/mxfenc.c
@@ -48,8 +48,10 @@
 #include "libavutil/pixdesc.h"
 #include "libavutil/time_internal.h"
 #include "libavcodec/avcodec.h"
+#include "libavcodec/bytestream.h"
 #include "libavcodec/golomb.h"
 #include "libavcodec/h264.h"
+#include "libavcodec/jpeg2000.h"
 #include "libavcodec/packet_internal.h"
 #include "libavcodec/startcode.h"
 #include "avformat.h"
@@ -102,6 +104,16 @@ typedef struct MXFStreamContext {
     int b_picture_count;     ///< maximum number of consecutive b pictures, used in mpeg-2 descriptor
     int low_delay;           ///< low delay, used in mpeg-2 descriptor
     int avc_intra;
+    uint16_t j2k_cap;        ///< j2k required decoder capabilities
+    uint32_t j2k_xsiz;       ///< j2k widht of the reference grid
+    uint32_t j2k_ysiz;       ///< j2k height of the reference grid
+    uint32_t j2k_x0siz;      ///< j2k horizontal offset from the origin of the reference grid to the left side of the image
+    uint32_t j2k_y0siz;      ///< j2k vertical offset from the origin of the reference grid to the left side of the image
+    uint32_t j2k_xtsiz;      ///< j2k width of one reference tile with respect to the reference grid
+    uint32_t j2k_ytsiz;      ///< j2k height of one reference tile with respect to the reference grid
+    uint32_t j2k_xt0siz;     ///< j2k horizontal offset from the origin of the reference grid to the left side of the first tile
+    uint32_t j2k_yt0siz;     ///< j2k vertical offset from the origin of the reference grid to the left side of the first tile
+    uint8_t  j2k_comp_desc[12]; ///< j2k components descriptor
 } MXFStreamContext;
 
 typedef struct MXFContainerEssenceEntry {
@@ -390,6 +402,20 @@ static const MXFLocalTagPair mxf_local_tag_batch[] = {
     { 0x8302, FF_MXF_MasteringDisplayWhitePointChromaticity },
     { 0x8303, FF_MXF_MasteringDisplayMaximumLuminance },
     { 0x8304, FF_MXF_MasteringDisplayMinimumLuminance },
+    // ff_mxf_jpeg2000_local_tags
+    { 0x8400, {0x06,0x0E,0x2B,0x34,0x01,0x01,0x01,0x09,0x06,0x01,0x01,0x04,0x06,0x10,0x00,0x00}}, /* Sub Descriptors / Opt Ordered array of strong references to sub descriptor sets */
+    { 0x8401, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x01,0x00,0x00,0x00}}, /* An enumerated value that defines the decoder capabilities */
+    { 0x8402, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x02,0x00,0x00,0x00}}, /* Width of the reference grid */
+    { 0x8403, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x03,0x00,0x00,0x00}}, /* Height of the reference grid */
+    { 0x8404, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x04,0x00,0x00,0x00}}, /* Horizontal offset from the origin of the reference grid to the left side of the image area */
+    { 0x8405, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x05,0x00,0x00,0x00}}, /* Vertical offset from the origin of the reference grid to the left side of the image area */
+    { 0x8406, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x06,0x00,0x00,0x00}}, /* Width of one reference tile with respect to the reference grid */
+    { 0x8407, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x07,0x00,0x00,0x00}}, /* Height of one reference tile with respect to the reference grid */
+    { 0x8408, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x08,0x00,0x00,0x00}}, /* Horizontal offset from the origin of the reference grid to the left side of the first tile */
+    { 0x8409, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x09,0x00,0x00,0x00}}, /* Vertical offset from the origin of the reference grid to the left side of the first tile */
+    { 0x840A, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x0A,0x00,0x00,0x00}}, /* The number of components in the picture */
+    { 0x840B, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x0B,0x00,0x00,0x00}}, /* Array of picture components where each component comprises 3 bytes named Ssizi, XRSizi, YRSizi.  The array of 3-byte groups is preceded by the array header comprising a 4-byte value of the number of components followed by a 4-byte value of 3. */
+    { 0x840C, {0x06,0x0e,0x2b,0x34,0x01,0x01,0x01,0x0a,0x04,0x01,0x06,0x03,0x0E,0x00,0x00,0x00}}, /* The nature and order of the image components in the compressed domain as carried in the J2C codestream. */
 };
 
 #define MXF_NUM_TAGS FF_ARRAY_ELEMS(mxf_local_tag_batch)
@@ -526,7 +552,7 @@ static void mxf_write_primer_pack(AVFormatContext *s)
     MXFContext *mxf = s->priv_data;
     AVIOContext *pb = s->pb;
     int local_tag_number = MXF_NUM_TAGS, i;
-    int will_have_avc_tags = 0, will_have_mastering_tags = 0;
+    int will_have_avc_tags = 0, will_have_mastering_tags = 0, will_have_jpeg2000_tags = 0;
 
     for (i = 0; i < s->nb_streams; i++) {
         MXFStreamContext *sc = s->streams[i]->priv_data;
@@ -536,6 +562,9 @@ static void mxf_write_primer_pack(AVFormatContext *s)
         if (av_stream_get_side_data(s->streams[i], AV_PKT_DATA_MASTERING_DISPLAY_METADATA, NULL)) {
             will_have_mastering_tags = 1;
         }
+        if (s->streams[i]->codecpar->codec_id == AV_CODEC_ID_JPEG2000){
+            will_have_jpeg2000_tags = 1;
+        }
     }
 
     if (!mxf->store_user_comments) {
@@ -558,6 +587,22 @@ static void mxf_write_primer_pack(AVFormatContext *s)
         mxf_mark_tag_unused(mxf, 0x8304);
     }
 
+    if (!will_have_jpeg2000_tags) {
+        mxf_mark_tag_unused(mxf, 0x8400);
+        mxf_mark_tag_unused(mxf, 0x8401);
+        mxf_mark_tag_unused(mxf, 0x8402);
+        mxf_mark_tag_unused(mxf, 0x8403);
+        mxf_mark_tag_unused(mxf, 0x8404);
+        mxf_mark_tag_unused(mxf, 0x8405);
+        mxf_mark_tag_unused(mxf, 0x8406);
+        mxf_mark_tag_unused(mxf, 0x8407);
+        mxf_mark_tag_unused(mxf, 0x8408);
+        mxf_mark_tag_unused(mxf, 0x8409);
+        mxf_mark_tag_unused(mxf, 0x840A);
+        mxf_mark_tag_unused(mxf, 0x840B);
+        mxf_mark_tag_unused(mxf, 0x840C);
+    }
+
     for (i = 0; i < MXF_NUM_TAGS; i++) {
         if (mxf->unused_tags[i]) {
             local_tag_number--;
@@ -1095,8 +1140,8 @@ static const UID mxf_wav_descriptor_key       = { 0x06,0x0E,0x2B,0x34,0x02,0x53,
 static const UID mxf_aes3_descriptor_key      = { 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x47,0x00 };
 static const UID mxf_cdci_descriptor_key      = { 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0D,0x01,0x01,0x01,0x01,0x01,0x28,0x00 };
 static const UID mxf_generic_sound_descriptor_key = { 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0D,0x01,0x01,0x01,0x01,0x01,0x42,0x00 };
-
 static const UID mxf_avc_subdescriptor_key = { 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0d,0x01,0x01,0x01,0x01,0x01,0x6E,0x00 };
+static const UID mxf_jpeg2000_subdescriptor_key = { 0x06,0x0E,0x2B,0x34,0x02,0x53,0x01,0x01,0x0D,0x01,0x01,0x01,0x01,0x01,0x5A,0x00};
 
 static inline uint16_t rescale_mastering_chroma(AVRational q)
 {
@@ -1260,7 +1305,6 @@ static int64_t mxf_write_cdci_common(AVFormatContext *s, AVStream *st, const UID
         f1 *= 2;
     }
 
-
     mxf_write_local_tag(s, 16, 0x320D);
     avio_wb32(pb, 2);
     avio_wb32(pb, 4);
@@ -1365,6 +1409,65 @@ static void mxf_write_avc_subdesc(AVFormatContext *s, AVStream *st)
     mxf_update_klv_size(s->pb, pos);
 }
 
+static void mxf_write_jpeg2000_subdesc(AVFormatContext *s, AVStream *st)
+{
+    MXFStreamContext *sc = st->priv_data;
+    AVIOContext *pb = s->pb;
+    int64_t pos;
+    int component_count = av_pix_fmt_count_planes(st->codecpar->format);
+    int comp = 0;
+
+    /* JPEG2000 subdescriptor key */
+    avio_write(pb, mxf_jpeg2000_subdescriptor_key, 16);
+    klv_encode_ber4_length(pb, 0);
+    pos = avio_tell(pb);
+
+    mxf_write_local_tag(s, 16, 0x3C0A);
+    mxf_write_uuid(pb, JPEG2000SubDescriptor, 0);
+
+    /* Value defining the decoder capabilities */
+    mxf_write_local_tag(s, 2, 0x8401);
+    avio_wb16(pb, sc->j2k_cap);
+    /* Width of the JPEG2000 reference grid */
+    mxf_write_local_tag(s, 4, 0x8402);
+    avio_wb32(pb, st->codecpar->width);
+    /* Height of the JPEG2000 reference grid */
+    mxf_write_local_tag(s, 4, 0x8403);
+    avio_wb32(pb, st->codecpar->height);
+    /* Horizontal offset from the reference grid origin to the left side of the image area */
+    mxf_write_local_tag(s, 4, 0x8404);
+    avio_wb32(pb, sc->j2k_x0siz);
+    /* Vertical offset from the reference grid origin to the left side of the image area */
+    mxf_write_local_tag(s, 4, 0x8405);
+    avio_wb32(pb, sc->j2k_y0siz);
+    /* Width of one reference tile with respect to the reference grid */
+    mxf_write_local_tag(s, 4, 0x8406);
+    avio_wb32(pb, sc->j2k_xtsiz);
+    /* Height of one reference tile with respect to the reference grid */
+    mxf_write_local_tag(s, 4, 0x8407);
+    avio_wb32(pb, sc->j2k_ytsiz);
+    /* Horizontal offset from the origin of the reference grid to the left side of the first tile */
+    mxf_write_local_tag(s, 4, 0x8408);
+    avio_wb32(pb, sc->j2k_xt0siz);
+    /* Vertical offset from the origin of the reference grid to the left side of the first tile */
+    mxf_write_local_tag(s, 4, 0x8409);
+    avio_wb32(pb, sc->j2k_yt0siz);
+    /* Image components number */
+    mxf_write_local_tag(s, 2, 0x840A);
+    avio_wb16(pb, component_count);
+   /* Array of picture components where each component comprises 3 bytes named Ssiz(i) (Pixel bitdepth - 1), XRSiz(i) (Horizontal sampling), YRSiz(i) (Vertical sampling).
+     The array of 3-byte groups is preceded by the array header comprising a 4-byte value of the number of components
+     followed by a 4-byte value of 3. */
+    mxf_write_local_tag(s, 8 + 3*component_count, 0x840B);
+    avio_wb32(pb, component_count);
+    avio_wb32(pb, 3);
+    for ( comp = 0; comp < component_count; comp++ ) {
+        avio_write(pb, &sc->j2k_comp_desc[3*comp] , 3);
+    }
+
+    mxf_update_klv_size(pb, pos);
+}
+
 static void mxf_write_cdci_desc(AVFormatContext *s, AVStream *st)
 {
     int64_t pos = mxf_write_cdci_common(s, st, mxf_cdci_descriptor_key);
@@ -1373,6 +1476,9 @@ static void mxf_write_cdci_desc(AVFormatContext *s, AVStream *st)
     if (st->codecpar->codec_id == AV_CODEC_ID_H264) {
         mxf_write_avc_subdesc(s, st);
     }
+    if (st->codecpar->codec_id == AV_CODEC_ID_JPEG2000) {
+         mxf_write_jpeg2000_subdesc(s, st);
+    }
 }
 
 static void mxf_write_h264_desc(AVFormatContext *s, AVStream *st)
@@ -2113,6 +2219,58 @@ static int mxf_parse_dnxhd_frame(AVFormatContext *s, AVStream *st, AVPacket *pkt
     return 1;
 }
 
+static int mxf_parse_jpeg2000_frame(AVFormatContext *s, AVStream *st, AVPacket *pkt)
+{
+    MXFContext *mxf = s->priv_data;
+    MXFStreamContext *sc = st->priv_data;
+    int component_count = av_pix_fmt_count_planes(st->codecpar->format);
+    GetByteContext g;
+    uint32_t j2k_ncomponents;
+    int comp;
+
+    if (mxf->header_written)
+        return 1;
+
+    bytestream2_init(&g,pkt->data,pkt->size);
+
+    while (bytestream2_get_bytes_left(&g) >= 3 && bytestream2_peek_be16(&g) != JPEG2000_SOC)
+        bytestream2_skip(&g, 1);
+
+    if (bytestream2_get_be16u(&g) != JPEG2000_SOC) {
+        av_log(s, AV_LOG_ERROR, "SOC marker not present\n");
+        return 0;
+    }
+
+    /* Extract usefull size infromation from the SIZ marker */
+    if (bytestream2_get_be16u(&g) != JPEG2000_SIZ) {
+        av_log(s, AV_LOG_ERROR, "SIZ marker not present\n");
+        return 0;
+    }
+    bytestream2_skip(&g, 2); // Skip Lsiz
+    sc->j2k_cap = bytestream2_get_be16u(&g);
+    sc->j2k_xsiz = bytestream2_get_be32u(&g);
+    sc->j2k_ysiz = bytestream2_get_be32u(&g);
+    sc->j2k_x0siz = bytestream2_get_be32u(&g);
+    sc->j2k_y0siz = bytestream2_get_be32u(&g);
+    sc->j2k_xtsiz = bytestream2_get_be32u(&g);
+    sc->j2k_ytsiz = bytestream2_get_be32u(&g);
+    sc->j2k_xt0siz = bytestream2_get_be32u(&g);
+    sc->j2k_yt0siz = bytestream2_get_be32u(&g);
+    j2k_ncomponents = bytestream2_get_be16u(&g);
+    if (j2k_ncomponents != component_count) {
+        av_log(s, AV_LOG_WARNING, "Incoherence about components image number.\n");
+    }
+    for (comp = 0; comp < j2k_ncomponents; comp++) {
+        sc->j2k_comp_desc[comp*j2k_ncomponents] = bytestream2_get_byteu(&g);   // Bitdepth for each component
+        sc->j2k_comp_desc[comp*j2k_ncomponents+1] = bytestream2_get_byteu(&g); // Horizontal sampling for each component
+        sc->j2k_comp_desc[comp*j2k_ncomponents+2] = bytestream2_get_byteu(&g); // Vertical sampling for each component
+    }
+
+    sc->frame_size = pkt->size;
+
+    return 1;
+}
+
 static const struct {
     const UID container_ul;
     const UID codec_ul;
@@ -2958,6 +3116,11 @@ static int mxf_write_packet(AVFormatContext *s, AVPacket *pkt)
             av_log(s, AV_LOG_ERROR, "could not get h264 profile\n");
             return -1;
         }
+    } else if (st->codecpar->codec_id == AV_CODEC_ID_JPEG2000) {
+        if (!mxf_parse_jpeg2000_frame(s, st, pkt)) {
+            av_log(s, AV_LOG_ERROR, "could not get jpeg2000 profile\n");
+            return -1;
+        }
     }
 
     if (mxf->cbr_index) {
-- 
2.34.1


[-- Attachment #3: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".