Älgjakt Sep 1989

This is only one extract from one digitized video film from this hunt.

Note: even if this transcoded (MPEG-2 to MPEG-4 AVC, H.264) material looks bad, it’s no worse than the original. (Original VHS had quality issues to start with.)

All the details that has been noted here is because this while page is more about the digitizing of material than the material itself. In this case. So far.

Used tool was ffmpeg (primarily, did some – not documented here – tests also with cvlc)

Work so far:

• Digitized video from VHS (did this already in 2006), MPEG-2.
• A relatively big effort to learn and test different tools and settings for optimizing materials was done in December 2020.
  • Focus on video
• Left for improvements:
  • Audio!
    • Clean up sound
      • remove noise
      • remove spikes, radio button (?)
      • increase sound of speech
    • Lower bitrate, encoding of audio (it’s all speech, can quite safely lower bitrate without any impact on quality)
    • All this can likely be done by (e.g.) Audacity but need to learn way more on this first.
  • Video – actual cleaning up, restoration, of video / visible materials
    • This is a more advanced area and not sure which applications exists, that are affordable and doesn’t cost an arm and a leg.

VHS (step #01), to first digitized ‘Original “05-05.46min.mpg” (step #2, 2006), extracted part 03:30-to-end (still mpg) (#3), then transcoded to “05-05.46min-03.30-rest.mp4” (#4) – bit rate is 2 080 kb/s (0.9 GB per hour).

1. 1989-09, original source, Kjell Hultfeldth, VHS, PAL
2. 2006-01-19, digitized by Johan, to MPEG-2 (which was the common format at that time, in the era of DVDs)
   1. From VHS, PAL cassette
   2. Using certainly not the best VHS player, but neither the worst, connected to a HP Media PC, w/ digitizing capability
   3. To MPEG-2, Main@Main, 720×480, 4:3, 29.970 fps, NTSC, YUV
3. 2020-12: various tests to try find a suitable middle-way for getting good quality vs. media size, based on source media that definitely wasn’t very good to start with.

Table 1: Tests with full “05-05.46min” – material

1hr* shows corresponding size for one hour of media based on bitrate, only to provide size indications of different formats and settings.

1. The digitized, MPEG-2, file.
2. Simplest transcoding, no special selections. $ ffmpeg -i 05-05.46min.mpg 05-05.46min.mp4
3. 2-pass, bitrate 3000 (video); F=05-05.46min.mpg; time ffmpeg -y -i “$F” -pass 1 -vcodec libx264 -b 3000k -threads 0 -f mp4  /dev/null && ffmpeg -y -i “$F” -pass 2 -vcodec libx264 -b 3000k -threads 0 -f mp4  `basename $F .mpg`.mp4
4. 2-pass, bitrate 2000 (video); F=05-05.46min.mpg; time ffmpeg -y -i “$F” -pass 1 -vcodec libx264 -b 2000k -threads 0 -f mp4  /dev/null && ffmpeg -y -i “$F” -pass 2 -vcodec libx264 -b 2000k -threads 0 -f mp4  `basename $F .mpg`.mp4

Table 2: Content of “05-05.46min” – material

(In Swedish)

Table 3: Tests with partial “05-05.46min-03.30-rest” – material

1hr* shows corresponding size for one hour of media based on bitrate, only to provide size indications of different formats and settings.

1. Extracting video from 03:30 to end. ffmpeg -i 05-05.46min.mpg -ss 0:03:30.000  -c copy 05-05.46min-03.30-rest.mpg
2. Transcoding, 1 pass, crf=30. ffmpeg -i 05-05.46min-03.30-rest.mpg -crf 30 05-05.46min-03.30-rest.mp4
3. Transcoding, 1 pass, crf=30, H.265. ffmpeg -i 05-05.46min-03.30-rest.mpg -c:v libx265 -vtag hvc1 -crf 30 05-05.46min-03.30-rest.h265.mp4
   NOTE: the first test of transcoding into h.265, HEVC, material created file incompatible with (at least) QuickTime on macOS. Need learning more on H.265!