pkt_duration, pkt_duration_time missing for first 20 frames

[frito-ray]

(Nice work by the way!) There's something a bit off with time.mp4, the first 20 frames are missing pkt_duration and pkt_duration_time. I'm not sure how various decoders use pkt_duration but I know ffmpeg will calculate it.

The following output was generated using ffprobe -of json -v debug -show_frames -show_format -show_streams -show_log 56 -unit $1 > $1.json 2> $1.mp4-errors.json

frame 17 - missing pkt_duration

 {
            "media_type": "video",
            "stream_index": 0,
            "key_frame": 0,
            "pkt_pts": 22000,
            "pkt_pts_time": "0.917584 s",
            "pkt_dts": 22000,
            "pkt_dts_time": "0.917584 s",
            "best_effort_timestamp": 22000,
            "best_effort_timestamp_time": "0.917584 s",
            "pkt_pos": "34056",
            "pkt_size": "2380 byte",
            "width": 1280,
            "height": 720,
            "pix_fmt": "yuv420p",
            "pict_type": "P",
            "coded_picture_number": 17,
            "display_picture_number": 0,
            "interlaced_frame": 0,
            "top_field_first": 0,
            "repeat_pict": 0,
            "chroma_location": "left",
            "logs": [
                {
                    "context": "h264",
                    "level": 48,
                    "category": 6,
                    "message": "[h264 @ 0x7fa8c780ac00] nal_unit_type: 1(Coded slice of a non-IDR picture), nal_ref_idc: 2"
                }
            ]
        }

frame 24 - has pkt_duration

{
            "media_type": "video",
            "stream_index": 0,
            "key_frame": 0,
            "pkt_pts": 25000,
            "pkt_pts_time": "1.042709 s",
            "pkt_dts": 25000,
            "pkt_dts_time": "1.042709 s",
            "best_effort_timestamp": 25000,
            "best_effort_timestamp_time": "1.042709 s",
            "pkt_duration": 1000,
            "pkt_duration_time": "0.041708 s",
            "pkt_pos": "44314",
            "pkt_size": "1377 byte",
            "width": 1280,
            "height": 720,
            "pix_fmt": "yuv420p",
            "pict_type": "B",
            "coded_picture_number": 24,
            "display_picture_number": 0,
            "interlaced_frame": 0,
            "top_field_first": 0,
            "repeat_pict": 0,
            "chroma_location": "left",
            "logs": [
                {
                    "context": "h264",
                    "level": 48,
                    "category": 6,
                    "message": "[h264 @ 0x7fa8c780ac00] nal_unit_type: 1(Coded slice of a non-IDR picture), nal_ref_idc: 2"
                }
            ]
        }

It appears to be encoded at 23.976fps:

"streams": [
        {
            "index": 0,
            "codec_name": "h264",
            "codec_long_name": "H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10",
            "profile": "High",
            "codec_type": "video",
            "codec_time_base": "1001/48000",
            "codec_tag_string": "avc1",
            "codec_tag": "0x31637661",
            "width": 1280,
            "height": 720,
            "coded_width": 1280,
            "coded_height": 720,
            "closed_captions": 0,
            "has_b_frames": 2,
            "pix_fmt": "yuv420p",
            "level": 31,
            "chroma_location": "left",
            "refs": 4,
            "is_avc": "true",
            "nal_length_size": "4",
            "r_frame_rate": "2997/125",
            "avg_frame_rate": "2997/125",
            "time_base": "1/23976",
            "start_pts": 2000,
            "start_time": "0.083417 s",
            "duration_ts": 480000,
            "duration": "20.020020 s",
            "bit_rate": "306544 bit/s",
            "bits_per_raw_sample": "8",
            "nb_read_frames": "480",
            "disposition": {
                "default": 1,
                "dub": 0,
                "original": 0,
                "comment": 0,
                "lyrics": 0,
                "karaoke": 0,
                "forced": 0,
                "hearing_impaired": 0,
                "visual_impaired": 0,
                "clean_effects": 0,
                "attached_pic": 0,
                "timed_thumbnails": 0
            },
            "tags": {
                "language": "und",
                "handler_name": "Bento4 Video Handler",
                "encoder": "AVC Coding"
            }
        }
    ],
    "format": {
        "filename": "time.mp4",
        "nb_streams": 1,
        "nb_programs": 0,
        "format_name": "mov,mp4,m4a,3gp,3g2,mj2",
        "format_long_name": "QuickTime / MOV",
        "start_time": "0.083417 s",
        "duration": "20.020020 s",
        "size": "773045 byte",
        "bit_rate": "308908 bit/s",
        "probe_score": 100,
        "tags": {
            "major_brand": "mp42",
            "minor_version": "1",
            "compatible_brands": "isommp42avc1iso5"
        }
    }

Can you share how you generated test.mp4?

I'm working on a project where frame accuracy is paramount and of course are seeing some of the same issues. We are evaluating a commercial frame-accurate player as a potential workaround. They claim to provide an accurate video.currentTime...