Is there a point in using QuickTime Player's 'maximum' quality for streaming Hi8 analog tapes?

cjp987 wrote:

With 4:3 NTSC scale 720x480 to 648x480 and then crop to the final 640x480.

I understand you are trying to crop the artifact at the bottom of the image

No, that relates to converting rectangular DV pixels to square pixels which computer monitors use. If that is not done, then circles will not be circles but ovals in the output.

Ignoring the artifact issue, I thought going from 720x480 to 640x480 was how the scaling (rectangular pixels to square) was normally done.

Just scaling to 640x480 is the often used quick and dirty way of doing that. The aspect ratio error goes unnoticed unless looked for.

Another way to reach the same goal with 4:3 NTSC 720x480 DV is to crop 4+5 pixels from both sides to 711x480 and then scale to 640x480.

Read all the gory technical details in the link above. Just a snippet from that page about NTSC which is even more weird than in PAL:

"525/59.94 systems have a line length of 63+5/9 (63.555...) µs, of which 52+59/90 (52.6555...) µs is the "active" part that contains actual image information. (The rest is reserved for horizontal blanking.)

52+59/90 µs × 13.5 MHz = 710.85 samples (pixels) per scanline.

In the vertical direction, there are 484 complete scanlines and 2 half lines. As above, all of them get digitized and half lines will be treated as if their missing other half belonged to the active picture, giving a total of 486 scanlines.

Thus, the active image area at 13.5 MHz sampling is 710.85×486 pixels. This is the actual area that forms the 4:3 (or anamorphic 16:9) frame.

However, we cannot use partial pixels in any practical video work. Therefore, the number 710.85 needs to be rounded up to 711, and we get a 711×486 pixel frame instead.

711 samples equals to 52+2/3 (52.666...) µs at 13.5 MHz, so the rounded-to-the-nearest-pixel active area is a little bit wider than it ideally ought to be. Fortunately, the difference of 0.0111... µs is (for all practical purposes) insignificant, and well within the tolerances of NTSC-M specifications. [snip]"

cjp987 wrote:

I played around with ffmpeg.

If you are fluent in using ffmpeg then you might want to test the following command for NTSC 4:3 .dv input to .mp4 output. It converts to square pixels, bob deinterlaces to double frame rate, outputs 10-bit H.265, sets color profiles and adds timecode. It uses the default quality settings but edit to something like '-crf 18 -preset slow' options for somewhat better quality with slower encoding and larger output. The command could be modified as a batch so it processes all .dv clips in a folder.

ffmpeg -i input.dv -vf bwdif=1,scale=648:480,crop=640:480:4:0,setsar=sar=1/1,setparams=range=limited:color_primaries=smpte170m:color_trc=bt709:colorspace=smpte170m -c:v libx265 -crf 28 -preset medium -pix_fmt yuv420p10le -profile:v main10 -timecode '00:00:00;00' -tag:v hvc1 -c:a aac -b:a 128k output.mp4

cjp987 wrote:

I could really use a video editor that allow me to edit the DV clip (namely, remove sections of video from the clip), and then re-export as DV without decoding/encoding the entire clip (but possibly doing so for a few seconds around the edit). ffmpeg can accomplish that, but you need to specify the time code of where to split the clip, which I imagine will take some trial an error before getting it just right. And then of course you will need to split again at the end of the section of the clip you want to edit out. So then you have 3 clips and you throw away the middle one, and then concat the two remaining clips. Rather laborious, especially if there are multiple edits you want to make.

MPEG Streamclip can edit .dv and losslessly export back to .dv. Just set in/out points via i/o keyboard shortcuts.

It is a great pity the 32-bit MPEG Streamclip and the old QuickTime Player 7 Pro can no longer be used in current 64-bit macOS (I boot to Mojave or use El Capitan or Snow Leopard Server via a virtual machine for that sometimes). As a best workaround in Sequoia I use Avidemux to losslessly cut or trim H.264/265 original drone movies for the archive because it readily displays I-frames where to do the cuts with such movies with GOPs.

cjp987 wrote:

It looks like by default my camera recorded in 12-bit:

So it 16-bit, but 32000 Hz???

s16le seems to indicate that one channel of 32 kHz stereo audio is stored in a 16-bit container in the DV stream but there is room for another such channel because only 32 kHz was used. 48 kHz would have occupied all that space with better quality.

(le = little-endian (Intel CPU default most commonly used) while be would be big endian (Motorola processor which old Macs used) -- refers to Jonathan Swift, who in Gulliver's Travels (1726) used them to describe the opposing positions of two factions in the nation of Lilliput. The Big-Endians, who broke their boiled eggs at the big end, rebelled against the king, who demanded that his subjects break their eggs at the little end.)

The factory default 12-bit audio option in DV/D8 camcorders is "historical" now and was not very useful even then.

Use 16-bit single stereo (2 tracks) unless you intend to perfom in-camera "overdubs" (assuming your camera supports this - many do not) in which case 12-bit means there's "room" in the signal for an extra stereo pair for a voiceover or music.

Modern editing software has more or less obviated the need to record 12-bit audio when shooting video but some apps might produce audio glitches or audio sync issues with 12-bit 32 kHz audio. You can convert to 48 kHz to but you can't restore 16-bit 48 KHz audio quality.

LPCM/WAV bitrate is 1536kb/s ONLY when the sampling rate is 48KHz and the quantization is 16bit (16bits x 48000Hz x 2, the 2 is for 2-channel). A 12bit, 32KHz, 2-ch WAV file will have a bitrate of 768kb/s (12 x 32000 x 2). These specs were chosen for DV so that two of the latter can fit in the space occupied by one of the former (choosing 12bit setting in the audio in a DV/D8 camcorder affords TWO 2-ch pairs, as against only one for the 16bit setting).

cjp987 wrote:

Regarding audio bit-rate, I'm not too clear what you are suggesting I do for my situation. Using ffmpeg with -an at least lets me work with the .dv files (concat them without the audio error), although I'm still not clear on why -an works, or possible negative side affects of using it.

If you think there is something I can do with ffmpeg when concatenating .dv files to improve the audio situation, please let me know which options I should use. As mentioned earlier, -ar doesn't seem to have any impact. Also let me know if any special handling of the audio will be needed when eventually converting to other formats like h.264.

As I said before, I'd use "cat" or MPEG Streamclip for concatenating .dv files instead using ffmpeg which seems to drop frames when doing so in at least when I now briefly tested it.

If you decide to use MPEG Streamclip for concatenating (or splitting) .dv clips, you might let it re-sample 32 kHz audio to 48 kHz while doing so. But that is slightly lossy and might not matter much unless there are audio sync issues in the possible editing and exporting to .mp4 phase. You finally export that DV's PCM audio as AAC codec with about the default 128 kb/s (with either 32 or 48 kHz sampling rate) as .mp4 audio.

cjp987 wrote:

Just to be clear on the ability of MPEG Streamclip when dealing with DV, you mentioned using it to concat, but I was wondering if it could be used to edit. Specifically, to remove a section of a DV clip, and then re-export as DV (with no loss except at the edit point). I can do this with ffmpeg, but it's very clumsy to do from the command line instead of a GUI where I would just click on the start and end points that I want to remove.

In MPEG Streamclip you can set edit points via the Edit menu or by just clicking I (in) and O (out) keyboard shortcuts (In point is included in the selection, the Out point is not included). Then Cut off footage or Paste it elsewhere or even to another movie. Then via the File menu losslessly export as .dv (or some lossy codec as deinterlaced etc). Or open many .dv clips and export them as one .dv file. Or open a large .dv and export it as 2 GB chunks.

You can move the playhead via the mouse, via JKL keyboard shortcuts, via up/down keys (to the GOP I frames, but .dv is frame accurate so every frame is the keyframe) and left/right arrow keys.

You can print MPEG Streamclip User's Guide.pdf from its Help menu and check entries for MAKING A SELECTION, EDITING THE STREAM, SAVING THE STREAM etc for details.