SD or HDV for DVD delivery.....

[Gerald Webb]

I think this is a no brainer but....
Its Eisteddfod time of year again for me, three weekends, full days shooting for hundreds of DVDs.
Last year I shot HDV to the Mrc1k, dropped the files in Vegas and burnt the disc from the timeline.
This took my Progressive HDV files and down sized to LFF Pal mpg2.
It didnt look too bad.
Had a thought yesterday, that I may as well shoot SD since it ends up SD anyway, and Ill fit more files on the CF cards, and avoid any loss in quality through down converting.
Can anyone explain why 1440x1080 and 720x576 are both 25Mbps?
Does this mean the PAL SD files are much higher quality per pixel?
So back to the original question-
Is there any reason whatsoever to shoot HDV if its going on a DVD?

[Leslie Wand]

i find down-rezzed hdv gives a much 'sharper', cleaner image than sd on dvd

add to that - in editing on a sd tl you can zoom in, pan, crop the hdv down to sd and you then have unlimited options with your hdv footage.

ymmv

[Gerald Webb]

Thanks Leslie,
Ill do a test between HDV and SD and see how I go. The panning would be handy I guess.
cheers

[Jeff Pulera]

Hi Gerald,

The DV format uses very mild compression, I believe 5:1, and each frame is compressed individually, known as intra-frame encoding. Each compressed frame by itself contains all the data to reconstruct the full frame again. If you get a drop-out on the miniDV playback, it is usually just 1 or 2 frames, not affecting other frames.

When HDV came out, they needed to figure out how to cram 5x as many pixels onto the same miniDV cassette. So they downsampled the 1920x1080 HD video to 1440x1080 anamorphic HD (1.33 pixel aspect) for starts, then used 4:2:0 color, tossing out some color data. Finally, they went to Long-GOP MPEG-2 compression. Unlike DV, HDV uses "inter-frame" compression, called "Long-GOP". GOP means Group of Pictures. Not sure for PAL, but for NTSC, HDV uses a GOP of 15 frames.

Basically, there are just a couple of FULL frames recorded each second, and the rest of the compressed frames have only partial data to rebuild the frame, and must look at frames ahead and behind and combine parts from several frames to make a complete frame. This takes more processing power for playback since several frames have to be processed and combined to display one frame. With fast action or complex scenes with movement, the HDV format can fall apart, with blockiness and artifacting.

Note that if you get a dropout with HDV playback, it typically lasts half a second (15 frames for NTSC) since a GOP is 15 frames and a loss of data from any 1 frame affects all those in the GOP, so it takes out the whole batch!

As for downconverting, 1080i can be difficult to convert to DVD since (with NTSC) the field order flips, and between that and the downscaling, doesn't always yield the best results. Probably easier to convert 1080p. But if you are doing high-volume direct-to-DVD type events like competitions, probably does make sense to just shoot as miniDV and not be converting, should look very good.

Thanks

[Gerald Webb]

Thanks for that Jeff, it explains a lot.
I did a few quick tests and there doesn't seem to be a great deal of difference in the finished DVD between original HDV or DV. I was shooting Progressive HDV though, so the field flip issue was probably negated.
It almost seems a shame that for high volume turnover like this the Z5 cant just capture in mpg2.
Straight from card to DVD author would be handy....

[Tom Hardwick]

The Z5 uses three 1440 x 1080 chips, so if you switch it into the DV mode the camera is doing the down-conversion between chips and tape, doing it in real time and doing it using a pretty basic conversion pcb. A computer has far more grunt and can usually do a far better job, so if I were you I'd shoot the footage at the highest possible quality and make DVDs off the end of the HDV timeline edit.

I've tested this with my Z5. Shooting DV (ie downconverting between chips and tape) and editing that in Edius 6 produces DVDs off the end of the timeline that don't look as good as shooting HDV, editing that in Edius and pumping out DVDs at the end of it.

tom.

[Gerald Webb]

Hmmm, there is more to this than I thought.
Native sized HD image with weak codec.
Or,
poorly in cam down converted image with much better codec.
I've tried both and out of Vegas it's hard to see any quality difference on the DVD via a 50" plasma. Guess it just comes down to fastest workflow.
thanks for the input, it's much appreciated.