Review of HPX2700 Varicam

[Christian Magnussen]

Originally Posted by Jeff Regan

If the BBC didn't like the 2100, I assume this means they didn't like the HDX900 either?

Since they accepted TopGear NorthPole for BBCHD with HDX900 that seems a bit odd. Since TG is one of BBC's high end productions, according to themselves..

And while we are debating numbers, there are numerous show shoot with say Hpx2100. Here in Norway the most popular series at the moment, "Himmelblå"(not sure what the correct translation would be, but it means blue sky(sort of)) is shot with Hpx2100. 1080p25 with DvcproHD. At the most 1,4million watch, or about 1/4 of Norway. It's not film but buy far the best looking series produced in Scandinavia, broadcasted in SD though.

Sometimes the numbers in a lab don't seem to mean to much in the real world...

If were to put numbers behind every desicion quality wise Xdcam wouldn't be accepted as a format for anything else than news.

[Jeff Regan]

Originally Posted by Steve Phillipps

According to the technical guru at the BBC the Varicam, 900 and 2100 all have different images, with the Varicam top of the heap.
The latest price for an HPX2100 I had in the UK was around £21000!
Steve

Steve,

I was in error re: US price for HPX2000, it is $27K USD for the body, not $25K, so add an AVC Intra board and you're at $30K, still close to $10K less than an HPX2700. It is very easy to buy an HPX2000 for $22K new and under $20K for a demo or B-stock unit.

As far as the 2000/2100 having different images vs. an HDX900, this makes no sense, the 2000 is the P2 version of the 900. It does have CAC, which the 900 does not, but comparing the two cameras in DVCPRO HD should yield the same image.

Jeff Regan
Shooting Star Video

[Steve Phillipps]

Check out the BBC test documents here BBC - R&D - Publications - WHP034 and it shows that there is a difference somewhere (I think some mention of optical low pass filters being relevant?) Anyway I seem to recall the upshot being that there was a lot of aliasing on the 2100, unlike the 900?
Steve

[David Heath]

Originally Posted by Steve Phillipps

Anyway I seem to recall the upshot being that there was a lot of aliasing on the 2100, unlike the 900?

The tests on the 900 don't seem to have included zone plate images, so it's not possible to say from them how well or badly the 900 fared. But the 2100 did indeed exhibit bad aliasing, so bad that the conclusion seems to be that no optical low pass filter is fitted. The direct link to the report is at http://www.bbc.co.uk/rd/pubs/whp/whp...ic-HPX2100.pdf .

Originally Posted by Christian Magnussen

Sometimes the numbers in a lab don't seem to mean to much in the real world...

Well, sometimes you need lab numbers to explain otherwise inexplicable happenings in the real world, and a very good example comes from Norway. I've quoted a piece from TVBEurope before about HD pictures of the football - maybe you remember the press comments from the time? I suspect DigiBeta is a misprint for HDCAM.

NRK principal engineer Per Bohler was receiving calls from leading newspapers in Norway asking why the first HDTV pictures from Germany were so poor.

“I had to admit it was poor quality, and at first we couldn’t explain why. The EBU satellite feed was fine, giving us MPEG-2 422 profile at 24 Mbps. We recorded it to DigiBeta, and our transmission output looked good when it left us — but the viewers received disappointing pictures.

“It really astonished me that the pictures from the satellite looked so good, but collapsed so quickly when we compressed them for transmission. It seems that concatenation of different compressions from acquisition, to the EBU and on to us, meant all the headroom in the signal had been lost by the time it reached us, with nothing left for the last encoder to work on,” he said.

And I suspect that's one reason why lab tests, including the BBC zone plate tests, have assumed so much more significance in the last year or so. You can no longer just look at camera pictures and draw full conclusions, attention has to be given to more scientific lab tests - or risk having the same problems as NRK.

It wouldn't matter if broadcasters could just throw bandwidth at the problem. Then the aliases would be of less significance. But bandwidth equals money, and whilst two cameras may look similar on straight recorded pictures, or with light compession, they could show severe differences with normal transmission compression.

Which is why the review of the 2700 I'm waiting for is one with zone plate tests. I would hope that unlike the 2100 it has the optical filter.

[Jeff Regan]

Great info from the BBC. Thanks Steve. I would have to think that the 2000/2100 and 900 would both have the same aliasing issues, however, these were pre-production cameras by the sound of it, so maybe low pass optical filtration was added for production versions?

Jeff Regan
Shooting Star Video

[David Heath]

Originally Posted by Jeff Regan

..........these were pre-production cameras by the sound of it, so maybe low pass optical filtration was added for production versions?

I think it's highly unlikely. Surely any manufacturer submitting a camera to the BBC for formal type evaluation would flag up any areas in which the production models would be better than the tested model?

Even if that had been initially overlooked, I believe these reports are submitted to the manufacturer before publication, and if their intention was to add optical filtering to subsequent units, I find it inconceivable that wouldn't have been picked up on and a statement of intent added to the published review.

[Jeff Regan]

Originally Posted by David Heath

I think it's highly unlikely. Surely any manufacturer submitting a camera to the BBC for formal type evaluation would flag up any areas in which the production models would be better than the tested model?

Even if that had been initially overlooked, I believe these reports are submitted to the manufacturer before publication, and if their intention was to add optical filtering to subsequent units, I find it inconceivable that wouldn't have been picked up on and a statement of intent added to the published review.

David,

Yes, that makes sense. The questions I have are why haven't I noticed aliasing artifacts with my HDX900 and does the HPX2700 have a low pass filter?

Jeff Regan
Shooting Star Video

[David Heath]

Originally Posted by Jeff Regan

Yes, that makes sense. The questions I have are why haven't I noticed aliasing artifacts with my HDX900 ..........

The way I understand the subject is that the impact on the straight out-of-the-camera picture may be barely noticeable. For straightforward viewing, alias artifacts may be largely masked by the actual picture itself. It's when the images become compressed that problems can start, and then will depend on bitrates, compression systems, cascaded codecs (as with the Norweigian problem) etc etc.

As I understand it, static aliases (as in a still photo) aren't too objectionable. With movement, the aliases move in the opposite direction to the shapes that are causing them, and that's what can cause coders confusion, and to waste data trying to compress something that shouldn't be there. It's the fact that they can survive unnoticed through the production process, but have a big effect right on final transmission that makes them so undesirable, and why broadcasters now HAVE to do lab tests, why you can't simply rely on "picture looks good to me". That's a fact of the digital age.

........and does the HPX2700 have a low pass filter?

That's what I'd like to know!

[Jeff Regan]

I asked a Panasonic product manager if the HPX2700 had different sensors than an HPX2000. The answer was no. I neglected to ask if there was low pass filtering for the CCD's. So, I'm still unclear as to why the 2700 would look better than a 2000 or 900 in DVCPRO HD or AVC Intra for the 2000 or 2700? It would have to be processing differences, I would guess.

Jeff Regan
Shooting Star Video

[Shawn Alyasiri]

My third HPX2000 came as an a-stock with the varicam logo on it. For whatever side you're on in this continued discussion, I would think that might tell you something. Wonder how that camera was being shown around...

I'll look forward to going side by side with the 2700. I've held the thought since NAB that they're the same camera, with extra software (cranking and Film Rec), the intra card (which mine has), and the extra HD-SDI.

Good problems I guess - with a 2700, I'll have 4 cams (2 with vari logos), and I expect that they'll match & switch images identically. Just think I'm paying another $12K for math and a logo with the 2700...

[Ellis Kendrick]

Hi Shawn,
Any further impressions on the HPX2700, and how its image stacks up against the HPX2000?

Originally Posted by Shawn Alyasiri

My third HPX2000 came as an a-stock with the varicam logo on it. For whatever side you're on in this continued discussion, I would think that might tell you something. Wonder how that camera was being shown around...

I'll look forward to going side by side with the 2700. I've held the thought since NAB that they're the same camera, with extra software (cranking and Film Rec), the intra card (which mine has), and the extra HD-SDI.

Good problems I guess - with a 2700, I'll have 4 cams (2 with vari logos), and I expect that they'll match & switch images identically. Just think I'm paying another $12K for math and a logo with the 2700...

[Steve Phillipps]

It's a mistake and very misleading to put a Varicam badge on an HPX2000, the Varicam means variable frame rates, which the 2000 just doesn't have.
I think it might be one of these comparisons where on-screen you don't see that much difference, but if you want to measure it the extra resolution to AVC-Intra and the 10 bit vs 8 bit should make a fair difference.
There's a Panasonic day with my local dealer next week where they'll have all their cameras, but not sure if I'll be able to make it. If I do I'll post any info.
Steve