Quality and the Pulldown from HV20's 24p

S

sean90291

Veteran
A guy posted this comment over at Cineform's forum:

Apparently, the HV20 always insists on introducing a pulldown to the 24p for a 30fps playback. No matter what software you use to capture 24p from the HV20's tape, you must do a pullup to regain the 24p. But there is inherent artifacting due to the MPEG-2 tape conversion to 4:2:0, resulting in chroma bleed and the fields can not be completely eliminated by any capture or post pullup. There are ways of further processing the image to get rid of the artifacts but they only result in loss of vertical resolution. This all seems to defeat the purpose of an HDV 24p palmcorder unless you are always hooked to a computer and capture directly to disk from the CMOS thru the HDMI. Did I miss anything before I return the camera?

Curious what you guys think. Does the process of pulldown removal to get 24p cause this "chroma bleed" or other loss of image quality?
 
I'm hardly a post-prod guy, but that guy doesn't know what he's talking about. The HV20 shoots 24 full-progressive frames per second (in that mode, of course), which then get encoded into 60 fields per second in a specific cadence of original frames and duplicates. All you need to do is throw away the duplicate frames and you're back to a 24p (actually 23.97, but who's counting) timeline. You can even do it by hand, if you have unlimited time and patience, but the big problem at the moment is basically no NLE software can yet handle the process entirely on its own; there are either third-party programs and/or specific, non-default options to be selected when doing the capture.

HV20.com is amassing a pretty good collection of known-good 24p workflows for various NLEs:

http://www.hv20.com/forumdisplay.php?f=32
 
Thanks. I did note that he had got the frame rate wrong, which made me suspicious.

And your answer raises another question for me. Is the 24p 60i footage from the HV20 interlaced then? I do see duplicated frames...but I do NOT see any interlaced frames. I don't see a single shred of interlacing at all. So I'm a bit confused. I thought it was 24fps converted to an interlaced 60i, which could be converted back to 24p using one of the software methods (please someone write a SIMPLE HV20 pulldown application). So, I'm just a bit confused: is the HV20's HDV footage (shot in CineMode) interlaced or not?
 
Well, you've stated that a bit confusingly: "cine" mode can be used whether you're shooting 24p or 60i. But you basically have it right: the HV20's 24p is just that, "real" 24 frame progressive, if you record in 24p mode, capture and remove pulldown (pain in the rump though it currently is) you'll end up with non-interlaced 24-frames-per-second.
 
The HV20 definitely uses a 2:3 pulldown system, and yes there are interlaced frames. This is different from the XLH1/A1/G1's "24F" mode, which uses no interlacing and no pulldown at all. All HV20 footage is recorded in a 60i data stream, and 24p footage gets 2:3 pulldown inserted to imbed it into the 60i data stream.

The issue with chroma bleed is perhaps mislabeled, but something that may actually be occurring. It has to do with the way interlaced HDV records chroma; interlaced 4:2:0 gets spread across multiple fields. It's entirely possible that reversing pulldown (and discarding pulldown fields) may screw up the chroma distribution. I haven't experimented with it to figure out whether it's an issue or not; it's definitely not an issue if you don't strip out the pulldown, but if you do, it might be an issue.
 
Barry_Green said:
The issue with chroma bleed is perhaps mislabeled, but something that may actually be occurring. It has to do with the way interlaced HDV records chroma; interlaced 4:2:0 gets spread across multiple fields. It's entirely possible that reversing pulldown (and discarding pulldown fields) may screw up the chroma distribution. I haven't experimented with it to figure out whether it's an issue or not; it's definitely not an issue if you don't strip out the pulldown, but if you do, it might be an issue.
Click to expand...

Does anyone actually have any proof of this?

Has anyone seen HV20 24P footage with this "chroma bleed" issue??

I'm really concerned about what is getting tossed around in this forum without any real proof to back it up. Rumours are rampant.

I'm not singling anyone out, it's just that it seems that I'm reading a lot of speculation here.

Can anyone back this up?
 
The original post that I saw mention of the issue is at Cineform's forum (search for "HV20"). Apparently someone named David Newman who works for Cineform raised the issue on an Adobe forum first. There are a couple of links to sample grabs posted there too, but they don't have any info or commentary along with them:

http://www.box.net/shared/static/rgvfqtzu9r.jpg
http://www.box.net/shared/static/15ydiu43co.jpg

I brought this up in DVXuser because it's being discussed elsewhere, and I want to get to the bottom of it. It might turn out to be a non-issue (or such a minor issue not worth caring about). But as Barry Green suggests, the problem is theoretically possible. And other people are posting that they see it as an issue. We're only here to try and get to the bottom of it, not spread rumours. I'd be happy to hear it's a non-issue.

EDIT: I wandered over to Adobe's forums and found what I think may be the original mention of the issue:

" CineForm Aspect HD does support automatic extraction of 24p from HV20's 60i stream. However, Ridlen is correct that the 60i encoding of 24p is not completely reversible, whereas the in 24F in the XH-A1 and XL-H1 is it. The reason the lack of completely reversibility is the 4:2:0 60i encoding which has only chroma value for pixels over two scan lines -- these scan lines may contain data from different frames, cause chroma cross-talk. The luminance is perfectly correct. Luma is encoded at 60i, and chroma at 30p, you can extract 24p from luma but not fully from chroma. If the pulldown is extracted correctly, the chroma cross-talk will only appear on every fourth frame. As our eyes are far more sensitive to luma, it is very hard to see this in motion. The only time this can be an issue is when keying, for that I recommend using HMDI and shooting live via the Black Magic Intensity card (using the new 1.5 drivers.) All CineForm tools support real-time pulldown removal using that card from the HV20. When you do a live capture the 24p in 60i is encoded as 4:2:2, with has separate chroma values for each scanlines, now the pulldown is completely reversable.

David Newman
CTO, CineForm "

And going on what David Newman (probably a very knowledgeable guy) has to say, the only issue for the naked human eye may be when keying. Otherwise, we probably won't ever see it. Perhaps that's case closed.
 
Last edited:
I agree that almost any problem is theoretically possible. It would just be nice to see evidence of a suggested problem.

Thanks for the link.

Is this a Cineform issue when extracting 24p?

I guess why I'm asking is because I've seen plenty of great looking 24p HV20 footage. Maybe I'm just not good enough to spot this issue.

Mike
 
I use the HV20 with the XH-A1 for events. I shoot at 24 with both cameras. After capturing the HV20 footage, I do reverse telecine and make it into a progressive 23.98
file to edit seamlessly with the XH-A1 footage. I haven't noticed ANY problems on either HD or downconverting to SD. It looks fantastic, and to be honest, much better than I initially expected.
I used a VX2000 as my B-camera in the SD days. The HV20 image beats the pants off the VX2000 image, unless in really low light situations.
 
I don't do any chroma keying but my question is....if you plan on chroma keying would it not be best done in 60i anyways? Just like whn you want a slow motion effect? If so then even if this problem does exist (theoretically) it would stil be a non-issue.

I have shot many many footage over the last few months with my HV20 and never seen this problem. Could it just be a problem with Cineform/HV20? I don't use Cineform but when I did use the demo I still did not see this problem.
 
It's not a CineForm issue. It's a 4:2:0 plus 2:3 pulldown issue. CineForm is top-notch software.

Like the guy from CineForm said, it's not an issue that we see in moving pictures. But it might crop up in keying. He's also probably talking about pro-level detail here. If you're keying your short film for the internet, no one might care. If you're keying for a national TV commercial or something, that could well be different.
 
Posting at the same time (thanks Sean)

It is not a CineForm problem -- please guys, you should know better than that. Unfortunately this is one of those cases when I gave too much information for most users to ingest. CineForm can completely reverse the pulldown live form the HDMI feed, with zero chroma issues, it is perfect -- try it for yourself. However, if you capture from tape, over Firewire or HMDI the chroma's pulldown can't be fully reversed -- but 99% of you will never see the issue if you are not keying. So in both sitatuations CineForm is capturing over HMDI, in one case (live) it is prefect, in the other (tape) everything 4th frame has a 50% chroma cross-talk with the next. So this not a CineForm screw-up, this is the result of MPEG2 4:2:0 encoding.

4:2:0 encoding of 60i, produces a chroma signal of 30p (not 60i which requires 4:2:2 color sampling.)


luma field 0 1 2 3 4 5 6 7 8 9 0 1 2 3
in 24p frames 01 234 56 789 01

chroma field (4:2:0) 0 0 1 1 2 2 3 3 4 4 5 5 6 6
in 24p frame 00 11(2) [23] (3)44 55

For three fields to construct a chroma "frame" 112 and 344 we can ignore the (2) and (3) to get full chroma -- no cross talk -- the CineForm pulldown handles these perfectly. However the middle chroma frame is contructed from two fields from different chroma source frames, an therefore choma crosstalk. This can't be fully reversed.

But like I said, most of you will not see this, it is very mild unless you are keying. For keying, capture live via HDMI and the issue completely go away.
 
Last edited:
David Newman said:
But like I said, most of you will not see this, it is very mild unless you are keying. For keying, capture live via HDMI and the issue completely go away.
Click to expand...

Thanks for that.

I use Cineform HDV to capture from the HV20 with Premiere Pro 1.5
I think it came with it?

I may need to purchase an upgrade or something to get it to capture 24p.

Mike
 
Kyser said:
Thanks for that.

I use Cineform HDV to capture from the HV20 with Premiere Pro 1.5
I think it came with it?

I may need to purchase an upgrade or something to get it to capture 24p.

Mike
Click to expand...

Yes for 24p support you need either Aspect HD ($499), or for a 1.5.1 like experience (i.e. not as fast) you can use NEO HDV ($249.) Everything has a free 15-day trial, download from www.cineform.com.
 
Is there any way to use Neo HDV with Avid? I read somewhere that CineForm might offer a QuickTime codec output. But the proprietary codec won't work with Avid is that right? (Avid, a dying breed.)
 
David Newman said:
Posting at the same time (thanks Sean)

It is not a CineForm problem -- please guys, you should know better than that. Unfortunately this is one of those cases when I gave too much information for most users to ingest. CineForm can completely reverse the pulldown live form the HDMI feed, with zero chroma issues, it is perfect -- try it for yourself. However, if you capture from tape, over Firewire or HMDI the chroma's pulldown can't be fully reversed -- but 99% of you will never see the issue if you are not keying. So in both sitatuations CineForm is capturing over HMDI, in one case (live) it is prefect, in the other (tape) everything 4th frame has a 50% chroma cross-talk with the next. So this not a CineForm screw-up, this is the result of MPEG2 4:2:0 encoding.

4:2:0 encoding of 60i, produces a chroma signal of 30p (not 60i which requires 4:2:2 color sampling.)


luma field 0 1 2 3 4 5 6 7 8 9 0 1 2 3
in 24p frames 01 234 56 789 01

chroma field (4:2:0) 0 0 1 1 2 2 3 3 4 4 5 5 6 6
in 24p frame 00 11(2) [23] (3)44 55

For three fields to construct a chroma "frame" 112 and 344 we can ignore the (2) and (3) to get full chroma -- no cross talk -- the CineForm pulldown handles these perfectly. However the middle chroma frame is contructed from two fields from different chroma source frames, an therefore choma crosstalk. This can't be fully reversed.

But like I said, most of you will not see this, it is very mild unless you are keying. For keying, capture live via HDMI and the issue completely go away.
Click to expand...
Sorry David, was not trying to imply that it is or was a Cineform problem....just reflecting on the original question " Does the process of pulldown removal to get 24p cause this "chroma bleed" or other loss of image quality?"
I've used the Cineform demo before and I believe it probably gave me the best picture out of all the methods I have tried so far (though all were great results). My earlier point was to state that I have not seen this issue at all...no matter which process I used to extract the 24p footage.
 
sean90291 said:
Is there any way to use Neo HDV with Avid? I read somewhere that CineForm might offer a QuickTime codec output. But the proprietary codec won't work with Avid is that right? (Avid, a dying breed.)
Click to expand...


Yes, CineForm now has an MOV capture and export option, easing the porting of CineForm media to the other two A's (Avid and Apple). Avid will not play third codecs, but I believe you can convert third party MOVs to DNxHD.
 
You must log in or register to reply here.
Back
Top