How many early adopters vs. tried&true buyers?

[Barlow Elton]

Originally Posted by Barry Green

Really. Weird. Tim Smith said it was a straight 24p recording. I'll have to look at the manual.

If it's recorded as 60i, why wouldn't it play back in the Sony?

I think the way it's written to tape isn't standard HDV, but the H1 can play it out as a standard 60i transport stream. This will be an issue, for sure, until there is a universal deck on the market. Doesn't standard H1 1080 60i play back properly from the Sony?

[Pete Bauer]

Barry,

Good discussion...but now I'm confused, too! The text on page 56 of the manual about HD 24F and the accompanying diagram seem like a bit of a disconnect to me. I THINK the text and the caption for the diagram mean to say that the HDV recorded on tape is 60i with 2:3 pulldown, taken from a 24 fps image coming out of the camera head. But the diagram shows the tape actually being 24 fps -- I suspect that's an error? What's your take?

Martin R.,

I haven't really had the chance to push the codec to try to induce GOP breakdown artifacts in an HDV2 recording (something I'm interested in doing because of the controversy where lines seem to be drawn between users and non-users of HDV cameras) -- so it may or may not be an issue. But considering there is as yet no evidence that it is a practical issue with HDV2, I think it is at best premature to say "DVCPRO easier & cost effective to edit and capture with no compression problems" thereby implying that compression is not only a problem for the XL H1's HDV2, but that DVCProHD cameras don't have compression issues. HDV and DVCProHD are both compressed formats and depending on a number of factors and what the editor wants out of it, either one may have practical limitations. Especially when we are discussing this in relation to brand new, and unreleased cameras, opinions on what the quality of the final images will be is really just speculation.

FWIW, editing 1440x1080 on a 24 fps timeline is as realtime and easy for me with using PPro 1.5.1 and Aspect HD 3.4 as is miniDV; I get a true 23.976fps timeline in high definition from the camera, just as easily as SD is done with an XL2 or DVX100. I'm picky about artifacts, and in the early going, NO complaints at all. I'll report what I find, good or bad, and hopefully others will, too.

[Barry Green]

Originally Posted by Barlow Elton

Doesn't standard H1 1080 60i play back properly from the Sony?

I'm sure regular 1080/60i does play in the Sony, but the 24F and 30F footage doesn't. Which leads me to conclude that it isn't 60i on the tape, but actually stored as 24 progressive frames (or 30 progressive).

[Barry Green]

Originally Posted by Pete Bauer

I THINK the text and the caption for the diagram mean to say that the HDV recorded on tape is 60i with 2:3 pulldown, taken from a 24 fps image coming out of the camera head. But the diagram shows the tape actually being 24 fps -- I suspect that's an error? What's your take?

Haven't seen the manual. Is it available for download somewhere?

I'd suspect that the analog outputs are being converted to 60i with pulldown, but that the firewire and tape outputs are 24p or 30p. Would such an interpretation fit with what the manual shows?

[Pete Bauer]

Ah, Barry Wan, too hard you work. The forest for the trees you miss! ;-)

The link to the manual is in a sticky thread at the top of the forum: http://www.dvinfo.net/conf/showthread.php?t=55123

Also, if you haven't already, I welcome you to take a look at my evolving web page (new XL H1, yippee!) and comment at THIS OTHER THREAD ... in addition to checking back in on this one, of course.

[Barry Green]

Okay, downloaded the manual.

Well, that's clear as mud. What is it saying? One the one had it says it records in 24 fps progressive mode, and on playback the signal is converted to 60i with pulldown. That makes sense. It also says that the output from the firewire port will be 24fps -- that also makes sense. And it says that the HD-SDI output will be converted to 3:2 pulldown. That also makes sense, and it all seems to be implying that the signal on tape is, indeed, 24 frames per second without pulldown.

However, on the chart on the right, it shows it on tape as discrete frames, but it labels it "60i". Why? If it weren't for that "60i" label, I'd say the manual is clearly showing that 24F is recorded as 24 discrete progressive frames, with no adjustment being made for interlace anywhere along the recording path. On playback, it does get converted to 60i by introducing 2:3 pulldown.

But here's where it gets even less clear: on the 30F diagram, it shows the 30 full frames being split into fields on tape. Well, if they're fields on tape, why can't a Sony play it back? Doesn't make much sense. The description paragraph says it records 30 frames, not 60fps fields... I'd have to go with the notion that it is, indeed, recording 30 frames. And that's why the Sony can't play it back.

But from reviewing this page, I'll stand by my original assertion which is that the 24F recording mode does not, in fact, use pulldown but records as 24 distinct frames. It's the only explanation that makes sense in light of the two things we know about 24f: 1) Tim Smith said it's 24 discrete frames, not 60i; and 2) the Sony can't play it, which it should be able to if it was actually a pulled-down 60i.

Note: that only holds for HD mode. In DV mode, the signals on tape always have pulldown introduced and are always recorded as 60i.

[Steven White]

No need for confusion - though David Newman could explain this better than me... and can certainly correct anything I mess up.

In HDV mode 24F is recorded using progressive MPEG-2 encoding. This maximizes the compression efficiency and reduces artefacts. Definitely a good thing when the software companies get it supported.

24F is played back on the analog and HD-SDI outputs with a 2:3 pull-down to maintain compatibility with the 1080i60 and HD-SDI standards.

In DV MODE 24F can be recorded to miniDV via either the 2:3 or 2:3:3:2 pull-down methods.

-Steve

[David Newman]

Barry,
Why worry, it works, if you are ready, here is the confusing bit. It is 24 frames progressively encoded (the picture header has the progressive flag set, not so in Sony CineFrame), with the field repeat flags set to apply pull-down over 60i. So the repeat flags are for the decoder to output 60i in 3:2 pulldown, encode has NOT applied pulldown to the frame data. So the trick is not to let the decoder apply the pull-down and just use the raw 24p encoded data. And it works great, i.e. buy Aspect HD today! ;)

[Nate Weaver]

It's the same scheme used in JVCs 24p, just on a field basis instead of frames.

Depending on how you look at it, one could say the stream has pulldown applied. Or you could say only 24fps are encoded and the stream is "native" 24p. Both statements would be correct.

I have this right, David?

[David Newman]

Yes, you have it.

[Barlow Elton]

Originally Posted by Barry Green

But from reviewing this page, I'll stand by my original assertion which is that the 24F recording mode does not, in fact, use pulldown but records as 24 distinct frames. It's the only explanation that makes sense in light of the two things we know about 24f: 1) Tim Smith said it's 24 discrete frames, not 60i; and 2) the Sony can't play it, which it should be able to if it was actually a pulled-down 60i.

Note: that only holds for HD mode. In DV mode, the signals on tape always have pulldown introduced and are always recorded as 60i.

Wouldn't the fact that you lose some sensitivity when switching to 24F also imply actual progressive recording? One thing I know with the XL1 was you didn't lose any sensitivity when switching to frame mode, which makes sense because it was a futzing of interlace fields. What I don't understand is how--if the chips are interlace--it loses that sensitivity if it's not switched into true progessive scanning when in 24F mode. It must actually scan 24 frames before anything else.

Weird.

[Barry Green]

Well, now that's interesting -- I haven't had a chance to use the XLH1 much at all, other than a brief time at ResFest. So I didn't get a chance to compare sensitivity.

If it's scanning a single field off the chip, similar to CineFrame 25, then it would lose the benefit (and blurriness) of row-pair summation, so that might explain the sensitivity difference, while still not making it a progressive chip. And it's not scanned progressively, Canon has said that. It's an interlace scan.

I don't know, really -- gotta get one and put it to the test to see if we can figure out what it's doing. Sort of academic though, because in practice I thought it looked very, very nice.

[Barry Green]

Okay, found something interesting:

Coursedesign posted this link: http://www.geosynchrony.com/scratchpad.htm

In there, an XLH1 owner has shot res charts using 24F and 60i, with almost identical framing, identical focus, etc.

Examining the charts, it looks like the 60i res maxxes out at about 700 lines, and the 24F looks like it's more like 600. On a different chart, I measured Sony CineFrame 25 at 575 lines. Both would seem to be within the margin of error that one will get from using a chart like this, especially considering minor framing differences that would occur, so I'd formally speculate that 24F is indeed one raw field of video, at 540 pixels high, and then the Digic II is synthesizing the second field, interpolating some data between the lines of the captured field. It would appear that 24F is the same thing as Sony's CineFrame 25, except clocking the CCD at 48hz rather than 50hz like the Sony.

Note, this is nothing like Sony's CineFrame 24!

With that speculation in place, it would seem like 24F is not "just like progressive scan", as the observed res should actually go higher on a true progressive scan capture as opposed to an interlaced capture (assuming, of course, that the lens and chips could resolve that higher detail!) However, since it's carried within HDV compression, it seems like 24F should produce very nice results, given that the somewhat softer detail, combined with the lower data rate and the progressive encoding, should lead to significantly fewer HDV-induced artifacts. And in practice, 24F did look very good.

[David Newman]

From my own experiments I find the 60i over 800 lines and the 24F mode around 650, this is not a single field interpolated up. The Canon 24F resolves more than Sony's CF24/CF25 modes. Although Canon has never stated how they do this, I believe it might be something like what I speculated back in September on http://cineform.blogspot.com

Canon could generate a good resolution pseudo progressive image if they ran the green CCD with reversed field dominance (output a bottom field when red and blue CCD output a top field) this way the same pixel shifting technology that makes Sony's 960 res CCD output near 1440, could help the three 540 fields (red-blue and pixel shifted green) could achieve a progressive vertical resolution approaching 810 (which is about the maximum a normal interlace picture can achieve any way.)

There is some evidence of this, even on Pete's web page posting. The 24Fand 30F have green and red/magneta tints on areas that the line frequency approaches the field resolution, I believe this is aliasing with the green sensor in some form of pixel offset with the blue&red sensor. This is maybe way Canon 24F is output such an impressive image. 24F is very nice.

[Michael Pappas]

Am I reading this right!

So the Canon XLH1 24f mode with it's higher res ccd, supposedly beefier codec engine and better optics is only resolving 25 lines more than a Z1U in 25f mode. Is that right?


Yet everything I have seen from the XLH1 that I shot at Birns then put up on their hd projection system says there is a lot more resolving going on.

Do we have any res chart numbers on the HVX200 yet?

Originally Posted by Barry Green

Okay, found something interesting:

Coursedesign posted this link: http://www.geosynchrony.com/scratchpad.htm

In there, an XLH1 owner has shot res charts using 24F and 60i, with almost identical framing, identical focus, etc.

Examining the charts, it looks like the 60i res maxxes out at about 700 lines, and the 24F looks like it's more like 600. On a different chart, I measured Sony CineFrame 25 at 575 lines. Both would seem to be within the margin of error that one will get from using a chart like this, especially considering minor framing differences that would occur, so I'd formally speculate that 24F is indeed one raw field of video, at 540 pixels high, and then the Digic II is synthesizing the second field, interpolating some data between the lines of the captured field. It would appear that 24F is the same thing as Sony's CineFrame 25, except clocking the CCD at 48hz rather than 50hz like the Sony.

Note, this is nothing like Sony's CineFrame 24!

With that speculation in place, it would seem like 24F is not "just like progressive scan", as the observed res should actually go higher on a true progressive scan capture as opposed to an interlaced capture (assuming, of course, that the lens and chips could resolve that higher detail!) However, since it's carried within HDV compression, it seems like 24F should produce very nice results, given that the somewhat softer detail, combined with the lower data rate and the progressive encoding, should lead to significantly fewer HDV-induced artifacts. And in practice, 24F did look very good.