Figuring out MINI-DV Format

[Vlady217]

Dear friends,
I need a tip about mini dv format. I'm doing a little research about a very old video clip.
The footage was shot on Sony DCR HC90. The camera was purchased and used in Russia (PAL).
The camera's online manual says it shoots 320/240. But my footage is 720/576.
Does it mean that the footage was edited and then exported as 720/576?

Thanx

[smrpix]

320 appears to refer to memory stick recording. If your source is a PAL DV tape it should be 720x576

[Vlady217]

320 appears to refer to memory stick recording. If your source is a PAL DV tape it should be 720x576

Thank you!

1. Does it mean that the camera's sensor is 720/576?
2. Here's something I'm trying to figure out. (Please see the picture attached, ignore the white area)
There are some colorful pixellated areas in the footage (in the red squares). I know this camera was used for a long time before this footage was taken.
Is it possible that these colorful bands of pixels were caused by an overused sensor? If not what could cause them? Could they appear later, during the export of the footage from the pc?
[Attachment 56143 - Click to enlarge]

[Cornucopia]

The sensor is likely 3200x2400 or some such UHD number, but that only gets fully used when taking photographs. When in movie mode (or taking photos to mem stick while in movie mode) it is constrained to standard DV maximums (720x480 - NTSC, 720x576 - PAL). While there is a little known sub-720 variant of the standard, I agree with smrpix and think that if it is referring to 320 it is likely a pic/movie made to the mem stick (while in movie mode). Note that movies made to mem stick are MPEG-based not DV-based. Tape is movie-only & DV-only.
When a sensor gets constrained like that, it usually is binned (skipping wells) or has its sensor wells summed (combining neighbors) to arrive at the usual 720x576, etc.
Binning would be more accurate for those sensor points, and be faster to operate, but summing wouldn't be that much slower and it would also improve on the sensitivity/SNR. Potato/Potahto.

Those color smears would be consistent with there being a data read error, especially in the color channel(s) where the cam is attempting to do error concealment (usually by guessing from surroundings). Since PAL is 4:2:0, the area to conceal for chroma would correspondingly be 4x the size of the luma. If the luma needed to conceal that much, you would know it was doing it because it would be obvious. With chroma, it is harder for human eyes to tell.

Scott

[smrpix]

1. Does it mean that the camera's sensor is 720/576?

From your manual, effective pixels still: 3 050 000, effective pixels movie: 2 050 000 -- so considerably more than 720x576 pre-interpolation.

[Vlady217]

The sensor is likely 3200x2400 or some such UHD number, but that only gets fully used when taking photographs. When in movie mode (or taking photos to mem stick while in movie mode) it is constrained to standard DV maximums (720x480 - NTSC, 720x576 - PAL). While there is a little known sub-720 variant of the standard, I agree with smrpix and think that if it is referring to 320 it is likely a pic/movie made to the mem stick (while in movie mode). Note that movies made to mem stick are MPEG-based not DV-based. Tape is movie-only & DV-only.
When a sensor gets constrained like that, it usually is binned (skipping wells) or has its sensor wells summed (combining neighbors) to arrive at the usual 720x576, etc.
Binning would be more accurate for those sensor points, and be faster to operate, but summing wouldn't be that much slower and it would also improve on the sensitivity/SNR. Potato/Potahto.

Those color smears would be consistent with there being a data read error, especially in the color channel(s) where the cam is attempting to do error concealment (usually by guessing from surroundings). Since PAL is 4:2:0, the area to conceal for chroma would correspondingly be 4x the size of the luma. If the luma needed to conceal that much, you would know it was doing it because it would be obvious. With chroma, it is harder for human eyes to tell.

Scott

Scott, THANK YOU!
So if understand you correctly -
These color smears are sensor specific - meaning IF these exact smears appear on another footage it would suggest that both videos were shot on the same camera. Is that so?
Is there anything else I should pay my attention to?

[Cornucopia]

It could be sensor specific or tape specific (most likely, IMO) or (playback) camera specific.

Scott

[Vlady217]

Got it.

THANK YOU!

[dellsam34]

The sensor is likely 3200x2400 or some such UHD number,Scott

Man I wish I had a camcorder in the 90's with that sensor. Yah, DV camcorders averaged 2 mega pixels back in the day which was good enough to take jpeg stills and a SD video footage such as DV.

[Cornucopia]

They may have boasted that many sensor sites, but I can guarantee they didn't actually provide that much clarity - ever. Mediocre optics, small sensor's poor sensitivity (thus much noise), early debeyer algorithms, and early compression options all led to blah images, even in SD.


Scott

[jagabo]

That looks to me to be a DV recording of a VHS tape via a composite cable -- not from the camera's CCD.

[aedipuss]

the hc90 was a decent minidv cam. originally shot video should be sharp and clear at 720/480 ntsc or 720/576 pal if captured over firewire from a tape.

[Cornucopia]

That looks to me to be a DV recording of a VHS tape via a composite cable -- not from the camera's CCD.

That's also a possibility, visually.


Scott