Stupid question about the formats
 

Hey, I realise this is maybe a stupid question, but if I'm right, this new camera shoots in "100 Mbps DVCPRO HD" and "DVCPRO50 and DVCPRO"

I know this maybe is going to sound stupid, but what is the difference between al this formats? I know the difference between SD and HD, but not much more than that, and I hear DVCPRO here constantly, but I still don't know what it is (and the other formats). (Or is Format not the right word?)

So can someone explain me briefly? What's the difference in picture quality, which difference can you see,...?
I thought I saw in the end credits of the movie The Edukators it was filmed on DVC or something too, but I don't know what it was, I do know picture quality in the cinema was very good.

Thanks very much in advance, sorry if it's a stupid question.

DVCPro is the Panasonic version of DV25 -- the video data is recorded using the DV25 codec, same as DV & DVCam, though Panasonic has chosen to have some minor variations on the audio handling (has linear audio tracks too) and records to a MP tape instead of an ME tape. But the video codec and therefore 'quality' is the same as all the other DV25 formats.

DVCPro50 uses two DV codecs in parallel, giving a result that is still SD resolution but has 33% more chroma data and a reduced compression -- 4:2:2 colour sampling, and a compression ratio nearer 3.5:1 than the 5:1 of DV25. This same system is used by the JVC Digital-S format, though it records to a VHS sized cassette.

DVCPro100 HD uses four DV codecs at once to deliver an HD resolution, 4:2:2 chroma sampling and a compression ratio of ... well, I'm not sure of that detail.

GB

DVCPRO:
25 mbps
4:1:1
720x480i 60
DCT 5:1 compression
Tapespeed: 33.82 mm/sec.
Tracks per frame: 10
2 x audiochannels
1 x cue audio track


DVCPRO P:
50 mbps
4:2:0
720x480p 60
DCT 5:1 compression
Tape speed: 67.64mm/s
Tracks per frame: 20
4 x audiochannels
1 x cue audio track


DVCPRO 50:
50 mbps
4:2:2
720x480i 60
DCT 3.3:1 compression
Tape speed: 67.64mm/s
Tracks per frame: 20
4 x audiochannels
1 x cue audio track (<6 khz)


DVCPRO HD:
100 mbps
4:2:2
960x720p 60
1280x1080i 60
DCT 6.7:1 compression
Tape speed: 135.42 mm/s
Tracks per frame: 40
8 x audiochannels
1 x cue audio track (<12khz)

Thanks for the replies

For compression ratios:

@ 1280x1080 at 30p, uncompressed is ~124 MBps
@ 960x720 at 60p, uncompressed is ~124 MBps

Thus the compression ratio is 124 MBps / 100 Mbps ~ 10:1

(assuming 24 bit colour)

<<<-- Originally posted by Steven White
(assuming 24 bit colour) -->>>

But we are not, we are talking 4:2:2 systems...not 4:4:4....

Because the chroma for DVCPro100 HD is reduced to 66% the compression ratio is not measured compared to uncompressed, undownsampled -- it is measured based on the post chroma sampling. Same for DV -- the 5:1 figure is calculated _after_ the 50% chroma downsampling. SD without any chroma downsampling is about 250Mb/s -- or 10 times the DV data rate ... but after chroma downsampling of 50% the DCT compression applies a 5:1 rate, not a 10:1 rate.

HTH

GB

Ah. Thanks for the correction I guess.

Seems like a cheap marketing way to quote less compression though. If throwing away 50% of the colour information does not count as compression to you, feel free to dividing all numbers by a factor of 2.

Well, by that standard the analog formats were 'compressed' too as they discarded colour information -- BetaSP discarded roughly a third, VHS quite a bit more ...

Better to use the terms as understood then reinvent the wheel, I'd say.

Cheers,
GB

Well - I would argue that the best way to guage compression is what's recorded vs. what's coming off the CCD DACs.

For example, w/ pixel-shifting someone was saying the best we could do in theory was 4:2:2 anyway - in which case, any 4:2:2 colourspace with no other compression would be "uncompressed" - in that case DVCPRO HD would be 5:1.

If you use 4:4:4 as a gold standard (which is currently what we all dream for... right?) then DVCPRO HD would be 10:1.

Of course, if you stored 12 or 16 bits per/colour then you could get "better than uncompressed 4:4:4" so I guess it gets too complicated. Add to the fact that pre-recording requires higher precision DACs than are output to tape to allow colour-manipulation in-camera...

It really leaves a lot of room for people to quote whatever they want in specs... doesn't it?

Maybe another stupid question: what does the 4:2:2 etcetera mean? I think with 5:1 you mean that dv-image is 5 times compressed or something if it goes to tape, but what does it mean with 3 numbers?

Thanks,

The terms describe the relationship between luma and chroma sampling.

The four indicates that the master sample frequency is 13.5MHz, and so 720 samples per line, the next two values indicate how often the chroma values are sampled. So a 4:4:4 image would have as many colour as gray scale values per line, a 4:2:2 format samples the chroma half as often as the luma (so in SD video the maximum resolution of the chroma is 360 samples per line), in 4:1:1 video the chroma is sampled a quarter as often as the luma. 4:2:0 actually describes a variation on 4:1:1 -- the ratio of luma to chroma samples is still four to one, but instead of each chroma sample continuing across four pixels wide, instead it is carried across two pixels wide and two pixels tall.

Picture every frame of video like a page of graph paper 720 squares wide by 480 squares tall (NTSC). Fill in each square a shade of gray -- that's the luma component. Now overlay with a transparent sheet of graph paper that is the same size, but is 180 square wide and 480 tall -- colour those squares any colour you want, one colour per square ... overlay on the gray scale page & the two combine to give you the luma and the chroma of a single frame.

If you were working in 4:2:2 space, the chroma overlay page would be 360 squares wide and still 480 tall -- 4:2:0 would use a page that is 360 wide and 240 tall.

If the values have a fourth 4 it means there is an alpha channel -- so 4:2:2:4 is the same as 4:2:2 but with an alpha. Occasionally you'll see the legacy analog format BetaSP described as being like 3:1:1 -- the lower first value signals that the 'sample rate' per line is less than 720 samples per line, the ratio between the values indicates that this format too delivers less chroma than luma in the final result.

HTH

GB

Thank you