P2 Varicam hpx2700

[steve phillipps]

Anyone seen/tried/tested/demoed the new P2 Varicams yet, especially the 2700?
Steve

[Alan Roberts]

No, I'm waiting for it, as are lots of people I know.

[steve phillipps]

Any thoughts on the 720 model, wondering how it'll look in 1080 mode and when the 720 is upscaled. Been considering the PDW700 and apparently the 720 upgrade will be available soon, so wondering how that will compare to the 720 Varicam with regards to movement portrayal due to the chips and compression schemes.
Steve

[Alan Roberts]

If the 2700 is what I expect it to be, it'll up-convert to 1080 very well (just look at Planet earth to see what I mean). Just remember that up-converting is far easier to well than is down-conversion, particulary so in-camera. AVC-I is consdierably better than DVCPROHD anyway, so it should be a significant step up from the good old Varicam (still one of my all-time favourite cameras).

[steve phillipps]

Alan, do you know if the 2700 is more or less the same as the 2100 excpet that it'll do 1-60 instead of just 50/60, or are there different chips or other architecture?
Steve

[Alan Roberts]

Sorry Steve, I really don't know, not yet. Panasonic had told me, ages ago, that there was to be a P2 replacement for the Varicam, and it has to be the 2700, but I don't yet know any of the details.

If I had to make a guess though, I'd assume that the 2700 is the Varicam with P2 recording, AVC-I and DVCPROHD, all at 720p. That makes sense to me, because the Varicam front end is a wonderful camera (proper 1280x720 2 ccd, without spatial offset). The 2100 owes far more to the HDX900, which was a reasonable attempt at 1080, but with 1280x720 sensors spatially offset both horizontally and vertically, ala HVX200. But, from my measurements of the 2100, the spatial offsets were removed, meaning that although the raster is 1920x1080, the content is 1280x720 but done wonderfully well. And the cost of a HDX900 is about the same as the replacement optical parts of a HDW750, which is why it's popular.

[steve phillipps]

I was on the verge of getting a PDW700 for a couple of jobs I've got (involving a lot of flying birds and other fast moving wildlife) but just started to wonder whether the 720P in the that camera (when the firmware's released shortly), will be as good or better or worse than the Varicam. If the PDW700 will do as good or better than that'd be the one to go for as you've got 2.2mp chips vs 1.1mp. But starting to wonder if there's more to it than meets the eye and in fact the Varicam would be the better choice.
Steve

[SimonMW]

As Alan says the 2100 uses 1280x720 sensors to obtain 1920x1080, although it can also record 720.

The 2700 is 720 only (and is the official Varicam replacement), it can record variable framerates up to 60fps. The more expensive 3700 is a 1080 version of the Varicam using native 1920x1080 sensors. However it can only do VFR up to 30fps.

I don't think that the 2100 has the Film REC gammas like the 2700.

I was on the verge of getting a PDW700 for a couple of jobs I've got (involving a lot of flying birds and other fast moving wildlife) but just started to wonder whether the 720P in the that camera (when the firmware's released shortly), will be as good or better or worse than the Varicam

720p from the 700 might not be as good as 720p from a camera with native 720p sensors. But at least the option is there.

[steve phillipps]

The 2700 will do 1080, in fact it'll more or less do the exact things that the 2100 will do, except it'll also do 1-60 fps, this is why I wondered if they were similar chips etc.
As for 720 from the PDW700, why should it be inferior to that from the Varicam? As it starts with bigger chips is there not the potential for it to actually be better?
Steve

[Alan Roberts]

The PDW700's 720 is an in-camera down-conversion, and that can never be as good as doing it properly on native 720p sensors.

HPX2100 does have 3 good "Filmlike" curves, but not the excellent "FILM REC" curves of the Varicam and others.

You can get my test/settings docs of all these except the as yet untested 2700 from the usual places :)

[SimonMW]

The 2700 will do 1080, in fact it'll more or less do the exact things that the 2100 will do, except it'll also do 1-60 fp

Sorry, I stand corrected, sort of. It will not do 1-60fps at 1080.

[steve phillipps]

Correct. And you do have to wonder what the 1080/25P will look like coming as it does from 1.1mp chips.
Steve

[Alan Roberts]

Look at the performance of Varicam when up-converted, any episode of Planet Earth will do. It looks very good indeed.

[steve phillipps]

I must admit that is what I say to other people when they question 720, I ask them "so you're saying Planet Earth looked a bit ropey then are you?"
But in terms of commissioning and acceptance a lot of folks do seem to be starting to show a preference for 1080 these days, you always just worry about buying a redundant bit of kit.
Steve

[Alan Roberts]

Don't get me wrong, I think 720p has no place in HD, it was a stop-gap solution to a problem that we never had .(Tests done to establish that 720p was a good idea were based on false assumptions about the rate of take up of bigger screens, and on some dubious test material. I was involved in some of those tests just before I retired, and disagreed violently with the conclusions being drawn from them, but that's history now.) But, the Varicam has a special place in all this, because of the huge flexibility of the settings. I spent many happy days finding out exactly what I could get from it, many of them in Bristol with the PE crew, which is why PE looks so amazing. But, you can still tell, if you know what to look for, that many shots don't have full HD resolution. It's a classic case of the content shining through the technology, so that you don't mind the problems.

720p done in many other cameras is far less successful, increasingly so as the size of consumer displays goes up. Bear in mind that the tests done to establish 720p as the EBU-preferred transmission format were done on the assumption the Europe would not sell displays into the home greater than 37" diagonal, and that they'd range in resolution from 853x480 to 1366x768. I protested at the time that this was nonsense, citing the sale of bigger sets in the US, but I was ignored. Now, 5 years on, 42" is the average size sold in the UK this year. The panel I bought this year is available only in sizes from 42 to 50", there isn't a 37" option, and they're all genuinely 1920x1080. I was right. Not surprisingly, many European broadcasters have ignored the EBU advice and gone for 1920x1080, only SVT (Sweden) and some German channels are broadcasting 720p (as far as I'm aware). That will change though :)

I'd never buy any kit labelled "HD" that did only 720p, it's days are numbered. 1080 rules.

[steve phillipps]

Thanks Alan, your opinion always valuable.
So any reason why the PDW700 might not be a good camera to buy for shooting broadcast, notably for wildlife docs (hence the need for 50-60 fps - ideally more, but that's Phantom territory, my current favourite camera by a country mile!).
Steve

[Alan Roberts]

I deliberately don't say whether cameras are approved for broadcast, in the settings docs, for two reasons: first, it isn't my decision; second, decisions can be changed. In the case of XDCAMHD, the decisions seems to be being made by DiscoveryHD, who currently haven't accepted it because the compression "is being overdriven", and I can understand that. I think it's not accepted by the BBC either, for the same reason.

The problem is the data rate. At 50Mb/s, it's a bit low. Consider: it claims to be full 1920x1080, 4:2:2 and 10-bit. So the compression aretfacts would be at about tyhe same levels as they would be for a system of 1440x1080, 4:2:2 and 8-bit were it to record at 50*1440/1920*8/10=30Mb/s. Mind you, the 2 lsbs are probably of less use, and there's gerenally not much going on between 1440and 1080, so maybe the comparison is mofre like 35Mb/s. Nevertheless, it's a bit too marginal for the broadcasters.

Please note that I don't test compression systems, that needs better kit than I have, so I don't make the rules.

[infocus]

The problem is the data rate. At 50Mb/s, it's a bit low. Consider: it claims to be full 1920x1080, 4:2:2 and 10-bit.
1920x1080 and 4:2:2 yes, but I don't think it is 10 bit. If it's MPEG2 it has to be 8bit, doesn't it?

I don't know about this particular camera, but understood that the latest version of what the BBC does and does not consider acceptable for unrestricted acquisition specified that an interframe codec should be no less than 50Mbs - so my assumption is that the PDW700 codec is considered OK, whilst 35Mbs XDCAM HD isn't. See http://www.bbc.co.uk/guidelines/dq/contents/television.shtml#Guidelines

Discovery seem a bit clearer with their Bronze, Silver, Gold categories, and I seem to recall hearing that the PDW700 had made it to Gold. (Unrestricted acquisition.)

[Alan Roberts]

Not sure about the 10-bit, you could be right. But the 50Mb/s cutoff is based on 1440 not 1920. What I'm hearing on the grape vine is that 50's not enough for 1920.

[infocus]

But, from my measurements of the 2100, the spatial offsets were removed, meaning that although the raster is 1920x1080, the content is 1280x720 but done wonderfully well.
I have to admit some confusion, Alan. Taking a look at your settings document for the 2100 - http://www.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP034_ADD24-Panasonic-HPX2100.pdf - I got the feeling that you felt the performance left a lot to be desired. You refer to "serious aliasing", and that "there is no optical spatial filtering in this camera", which I assume to mean it doesn't have an optical low pass filter?

[SimonMW]

The 700 uses 8-bit. The 50Mb/s is also CBR unlike the 1/2" cameras. That datarate is also purely picture and does not include audio.

Mainly the 50Mb/s over the 35Mb/s is down to the 4:2:2 color sampling, so as Alan states the compression artefacts will be the same as 35Mb/s. However the CBR factor does mean that the camera handles issues such as flash strobes etc better than the 1/2" cameras.

My understanding of the disc system was that they were supposed to have made the newer ones a lot faster. But this doesn't seem to be the case. I do know that Sony's line is that their in-camera compressors are so efficient now that there would be no noticeable gain in quality by increasing the datarate.

I feel now that they should start the shift over to solid state and higher datarates, but keep the disc system in terms of devices such as the U1 for backup (for small operators such as myself) now that SS cards are affordable enough to record several hours without issue. Even if the Ikonoskop DII doesn't take off as a camera, I have heard that at least one of the major camera manufacturers is interested in their SS technology.

[Alan Roberts]

Indeed so. I think what they've done is to use the front end from the Varicam and not the HDX900. The Varicam has pixels aligned, the HDX900 has half-pixel offset, giving more apparent resolution but some visible aliasing, enough to cause problems. Being charitable, perhaps the bi-refringent is aimed at the resolution the HDX gives, and not the Varicam, that would let more aliases through, as seen. Or maybe there's no filter at all, which I don't really believe. Either way, I got the impression that the 2100 had been rushed through a bit, the 3000 is a lot better.

[infocus]

Being charitable, perhaps the bi-refringent is aimed at the resolution the HDX gives, and not the Varicam, that would let more aliases through, as seen. Or maybe there's no filter at all, which I don't really believe. Either way, I got the impression that the 2100 had been rushed through a bit, the 3000 is a lot better.
Looking again at your zone plate images, my impression is that the aliasing is strong all the way up to the limits of the chart - if it was a case of the bi-refringet being more suited to HDX900 resolutions then wouldn't it at least be starting to roll off well before the 1920/1080 limits?

But the absence of an optical low pass filter in a camera at this price does seem pretty unbelievable. Did you query the matter with Panasonic? I suppose there's no chance it was a early model sent for test, and not representative of the production run?

Either way, it's something to be on the look out for with the 2700. As far as Steve's question goes - the 2700 or a PDW700 - the PDW700 will obviously give far superior 1080 output. At first sight, the native 1280x720 sensors of the 2700 should give it an advantage in 720 mode, but if the level of aliasing that the 2100 shows are present in the 2700, they may be worse than downconversion compromises in the PDW700.

[Alan Roberts]

The manufacturers always get a copy of the docs before publishing. It's up to them to respond if there's a problem. I report what I find, so.......

The problem with aliases is that they're burnt in at the front end of the signal process, so down-converting to 720 won't make it any better (because all the stuff above 1280px720 that causes the aliasing is still there). Until I see a 2700, I can't really say much more, it'd only be guesswork.

[steve phillipps]

Alan, you say that you were never a big fan of 720, but I suppose the issue was forced to a large extent - as far as wildlife goes at least - by the need for some sort of slomo. I think think most wildlife camerapeople would agree that even 60 fps is not really enough, but it's miles better than 25fps for wildlife, especially things like flying birds, and there was not (and still is not practically speaking) a 1080 cam that does more than 30 fps.
The slomo issue cannot be overstated with wildlife work and when you get jobs with ithe NHU you don't even have to ask what cam you'll be using 'cos it's Varicam every time, almost entirely 'cos you need 60 fps at the very least. And I expect the P2 2700 Varicam to be the logical next step, though I do hear a lot of interest in the PDW700 - we shall have to see.
Steve

[Alan Roberts]

I couldn't agree more. When we were setting up PE, there were financial incentives to go entirely Varicam or entirely HDCAM. I said No, we had to have the best cameras for each job, so the under-water stuff went HDW750 (Peter Scoones already had a housing for digi that would take it) and all the in-air stuff went Varicam. I spent a ling time, with NHU crews, working out the best way to use it all, ending up with 2.5 days in a shed with Mark Linfield to derive the Varicam's "wildlife" settings. Mark subsequently went on to use the Film Rec curve to great effect, but we'd learned of the limitations by then, so we all knew what we were doing.

Many times I've said, to both Sony and Panasonic, that we need a camera that will go up to 12fps because 60 isn't enough. But, it's essential that the resolution doesn't change with shooting speed: the only way Sony could get "off-speed" would be to shoot interlace and slow down 2:1 in post, with a big hit on resolution. I'm still totally convinced that we were right to reject that route.

And, that's why we're all still looking out for the right camera, and I haven't seen it yet.

[steve phillipps]

And it still doesn't look anything like imminent either. You get excited about cameras like the RED that'll now do 120 fps at 2K res (and there are others that sound interesting too), but there's no free lunch and there are always penalties to pay. Same goes with this new Ikonoscop camera that's getting a bit of interest, there will be some sort of trade-off. There are reasons that the expensive broadcast cams are made the way they are, and there are reasons why they cost what they do (mega-corporation profiteering being only 1 part of it).
Look at the Sony EX1/3 - super high spec, really low price - too good to be true? You bet it is, there are some real issues in there.
If only we could have a high-res, reliable camera that'd do 150fps - I'd call it something like Arri HSR2! Still long to have mine back again at times!
Steve

[Alan Roberts]

Yep, spot on. The only part I'm not convinced about is the corporate profiteering bit, I've never really been convinced that it actually does happen. Obviously, it costs quite a lot to do the R&D for new technologies, and it has to be recouped somehow, that explains the initial high prices, or the high prices of kit that has a small market (Genesis, D20, F23, F35 etc). And the other side of the same coin is that any new product that seems too good to be true either doesn't achieve what it aims, or is being funded from somewhere else. It's a harsh world out there.

[infocus]

When we were setting up PE, there were financial incentives to go entirely Varicam or entirely HDCAM. I said No, we had to have the best cameras for each job, so the under-water stuff went HDW750 (Peter Scoones already had a housing for digi that would take it) and all the in-air stuff went Varicam.
If comparisons are between the HPX-2700 and the PDW700, then on the face of it the 700 may be capable of being todays dual purpose camera - it will do true full raster 1080, the 2100 can only upres.

And both can do 60fps 720 - the question then becomes which does it better? I'd have assumed the 2700, as the 700 has to downresolve, the 2700 has native chips. But seeing the zone plate results from the 2100 that must be in doubt - if they've just kept the same front end, the lack of the optical low pass filter may be a bigger drawback than the downconversion, quite apart from it just not being able to compete at 1080.

As far as Planet Earth goes, then obviously stunning, and undoubtably the best that could be done at the time. But a lot of it's initial (technical) appeal may have been in that there was so little other HD material around when it was released, and most of the comparisons were of it versus SD. Coupled with the content, it's little surprise it got the deserved acclaim it did.

But with so much more HD material now around, and with 1080 native screens so much more common, it is becoming more obvious that there's HD and HD. There certainly seems a gap between most broadcast HD and most Blu-Ray disc.

[Alan Roberts]

I'm holding all comments on the 2700 until I've measured one.

But you're right about the impact PE had, we were coming from Blue Planet and mostly film. The extra cleanliness we got with PE was stunning. But, things have moved on and we can't rest on laurels, they wilt.

[Bruce]

Talking of HD and hd. I have just viewed Wild China on Bluray and many of the pictures are as soft as old boots! Very disappointing.

[steve phillipps]

Varicam again.
Bear in mind though that with wildlife work there are a LOT of factors involved in image quality, most notably distance to subject which can introduce a lot of haze. This is one way you can tell if the subject was tame/captive/habituated, the shots look so sharp the cameraperson was obviously about 5 feet from the subject.
Steve

[Alan Roberts]

Yep, Varicam is showing it's age. Just remember that I first saw the origins of that camera (when is was called the 27 and shot only at 60fps) in about 1991 or 2. I was amazed at what we could do in the menus, particularly so with the gamma and knee controls. Whe I was given a 27V to play with, the first Varicam (1~60Hz) I commented that the knee controls made it look like film, and measured the curves. It was the only camera I've ever seen where the knee point could be taken all the way down to zero, black level. The resulting curve looked uncannily like a proper film curve, and became what we now know as Film Rec. I took this as direct evidence that Panasonic listen to the users, not a common situation in this game.

[infocus]

Yep, Varicam is showing it's age.
Made even worse by the 720 variant of the DVCProHD codec being only 960x720, not 1280x720. And leading to the strange situation of some Panasonic camera users, whose cameras have resolutions around the 1280x720 mark, running them in 1080 mode just to get that 1280 horizontal definition!
I took this as direct evidence that Panasonic listen to the users, not a common situation in this game.
Maybe in this situation, but definately not in a lot of others. Maybe the classic is the HVX200, which was announced as prototype long before it actually saw the light of day. During that period there was a clamour for it to DVCProHD *AND* HDV. Use the former when downloading wasn't a problem, use HDV when P2 workflow isn't practical. Totally ignored by Panasonic, who I believe then saw the big majority of sector sales then go to Z1s etc. Whereas I suspect an either/or camera would have cornered the market.

And when P2 first came out (SD only), I know many people suggested a 2/3" hybrid P2 with DV or DVCPro tape deck as a transitional product. Not taken up until the HVX200 - which then couldn't record SD to tape and card together. Doh!

[Alan Roberts]

Panasonic aren't in the HDV consortium, they never signed up to it, so they can't make HDV kit. They saw the consumer market going straight to solid state rather than to tape. I agree that the HVX200 is a rogue, it doesn't neatly fit into any niche that I can see, except as a test bed for the HMC150/151.

[infocus]

Panasonic aren't in the HDV consortium, they never signed up to it, so they can't make HDV kit. They saw the consumer market going straight to solid state rather than to tape.
I think the argument was that they SHOULD have signed up to HDV, and it was a mistake not to. The consumer market may NOW be going towards solid state, but there have been a very large number of HDV cameras sold in the last few years, and that's a market that Panasonic have not had any slice of.

Their argument was "choose the HDV tape workflow, or a high quality codec with DVCProHD". One or the other. What many people asked for was to have the choice within the same package. I think the relative number of Z1s that you see about, compared to HVX200s tells it's own story as to how the majority of people then voted.

[Alan Roberts]

Couldn't agree more, but that's history now. They didn't think HDV worthwhile and went their own way. I presume they saw things differently.