JVC GR-PD1 review in March CV

The recording system is HDV (MPEG, short GoP, about 19Mb/s) for the 1280x720p and 720x576p formats, but standard DV (25Mb/s) for 720x576i. What you've spotted is something I've been banging on about for a few years now, that the pixel numbers in the camera don't bear direct relatinshgips to the recorded formats or the expected resolutions. The difference between 720 and 659 is a reflection on the squareness or otherwise of the pixels, and the sensor does not have to have square pixels whatever anyone says, that's a matter for the camera designer. It is quite possible to source pictures for a 1280x720 recording format by using fewer than 1280x720 pixels, prviding the right filtering is used, since you can never get 640x360 cycles/width or height because thaose are the alias-free limits of the system and filters are always used to prevent image frequencies exiting the format at unacceptable levels. So 659 may well be enough, given how good the lens is, only lab tests will ever show this.

You're right about the colour pixel structure, JVC have used a negative filtering system that is more sensitive tha positive filtering, but still not as sensitive as 3-ccd via dichroics. The filters still absorb unwanted light which is therefore lost rather than redirected as in a 3-ccd camera. But these differences are now getting rather small. The main difference in this camera is that the negative filtering demands matrixing for the extracxtion of RGB from the sensor. Even for green. Each of R G and B is extracted as a set of sum/differences for a 2x2 block of pixels, I have a note from JVC explaining it. What this means is that the resolution of the device is only half what you'd expect for alias-free detail. So, for the 1280 horizontal pixels, you get up to 320cycles/picture width resolution without aliases. From 320 to 640 the detail contains, but may not be dominated by aliases. The clever bit is that by doing it this way there isn't a resolution-less frequency band separating them, the aliasing range blends with the clean range resulting in aliases that are less objectionable than you get from RGB sensors.

The PD1 is a Europeanised version of the 59.94 version that was launched over a year ago to much trumpeting in the US, and it is still the only camera aiming at HD in the cinsumer market. It isn't perfect. It isn't acceptable as a source of HD pictures for professional programme-making. BUT it will get used a lot until better ones appear, simply because it's small and cheap.

If you're at the Video Forun this year, I'll be happy to go into details of all this, I can't do it here because it needs too much maths and diagrams to get it right.

Thanks Alan - I think I'd guessed that the image sensor must have non-square pixels! My main point is that the difference between the (US) HD1 and the (PAL territories) GD1 is the former records 720p, the latter only 576p - though I believe some schools of thought take 576p to be HD.

If you go to the original link I posted, the diagram and formulae relating to the pixels indicate that for LUMINANCE the pixels are matrixed in horizontal pairs, so I believe that for luminance the full resolution of the chip is maintained in the vertical direction. For chroma, yes, the derivation is from 2x2 blocks with consequent loss of chroma detail. I wonder whether R, G, or B are ever derived in their own right - or are the blocks of 4 are used to derive the colour difference signals directly? Anyway, am I right in therefore thinking that the luminance resolution is equivalent to a 3 CCD camera with 640x659 pixels for each of R,G,B?

I am registered for video forum, but may have to work away from London for the first two weeks in Feb. I'd like to get there if only to see the latest in HD!

I don't know enough about the 50Hz version to comment further, but I'm detting round to it

On the JVC UK site (www.jvc.co.uk), there's a slightly better explained spec -

Multi-format recording and playback in MPEG2:
16:9 625/50P
16:9 625/25P
4:3 625/25P
and in conventional miniDV:
3:3 625/50I

There's also a blurb sheet called "JVC progressive World" that gives it in less technical terms.

So, it seems that all recordings are 576 lines (i.e the active lines of 625-line), but that the various formats use different areas of the (single) ccd, with appropriate filtering to drive the output formats. That all sounds sensible to me.

Interestingly, nowhere do they claim the camera to be "HDTV", only to be "Hi-Def". That may sound like nit-picking, but HDTV has specific meanings defined in SMPTE and ITU documents. The 59.94Hz version (HD10) claims to be HDTV in that it outputs in the 1280x720P format (but only at 29.97P) and will upconvert any of its formats to 1920x1080/29.97Hz P, albeit with no increase in image sharpness.

So, it looks like JVC have pulled their horns in a bit, and are selling the PD1 without really claiming that it makes HDTV, only that it makes "Hi-Def" pictures, i.e. pictures on 576-line standards that have higher image sharpness than can normally be achieved using conventional techniques.

Sounds ok to me, but I can't imagine anyone really wanting one yet in the UK.

[This message has been edited by Alan Roberts at work (edited 27 January 2004).]

I think they make an even bigger play about it's "progressive" nature than "HiRes" Alan, and as I originally posted: ".....it is the progressive mode that seems to give the camera it’s name – the US equivalent looks like the GR-HD1 (H for HD)..." - in the 50Hz case P for progressive!

I take quotes about "pixel counts" with a bit of a pinch of salt anyway, a camera could have twice as many pixels again, but with a poor lens it would still be a poor camera! I'd prefer to see more resolution chart results in tests. Interesting to compare the Australian literature ( http://www.jvc-australia.com/!ProductFeatures/pics/GR-PD1_PDF.pdf ) with the UK - JVC seem to think Aus readers are more capable of being swayed by tech details!

Regarding what you say about meanings of HDTV, I think I remember reading that currently in Australia all broadcasters are required by their government to transmit so many hours of HD per week - but they've all chosen differing ways of doing it. One (think offhand it's SBS) I'm sure meets it's quota with 576p!

Finally, an interesting top story about NAB, Pinnacle and Sony at http://www.definitionmagazine.com/ and I somehow think we're going to hear a lot, lot more about HDV in the pages of CV in the rest of this year......

Agreed that P means HD for some people. 576P in Oz just about scrapes in, and even 480P in the US. I take your point about pixels and lenses, but my own tests on the HD10 show that the lens is rather good, certainly it isn't the limiting factor for resolution, and I'm certain that the housing and lens are identical for the two models, only small changes in processing demarcate them.

I think the reason why the Oz and UK sites have different levels of info is diown to the expected market size. Australians, like Americans, are starting to buy "HD" kit because they've got a transmitted service, so the canmeras can be expected to sell there. But with no immediate prospect of HD in Europe (remember, all you read is about the arguments over formats, not about whether and how to transmit it) for then next 5 years or so, HD bits are really not in the consumer domain. Broadcasters and indie producers are using HD quite a lot across Europe, but only where there's a HD co-producer who'll buy the result, there';s very little HD production going on that has no HD outlet, with one excpetion that I'm immensely proud of but can't talk about. Yet.

The comment about lens quality was intended as a general observation - I've no direct experience with this camera. Would prefer reviews to do res chart tests, rather than quote pixels numbers though.

Heard some more of interest on the Euro HD front, but will post that on the HDTV thread I started in "Chatter".

Regarding the “Hybrid Complementary-Primary colour filter system” that the PD1 uses, I'm amazed I hadn't heard of it before a couple of days ago. Is is used by anybody other than JVC? Any stills cameras?

Yes, Canon have published that they have a slightly different version with magenta instead of green. The idea's not new, but it hasn't really been getting into cameras until about the last 3 years, because of the processing overhead.

I'd be muich happier with test cards showing what happens, I'll bring some along to the Video Forum and will be happy to talk about what they show and how to use them (you can print your own if you know what you're doing, and I have software that will draw very many test charts for exactly this purpose).

[This message has been edited by Alan Roberts at work (edited 29 January 2004).]

Just so you don't think we are overlooking this thread - James Morris, the reviewer, has been out of the country for a number of weeks but is back now and will respond once he's managed to get himself settled back in.

Bob C

I like your thought that:
"I take quotes about "pixel counts" with a bit of a pinch of salt anyway, a camera could have twice as many pixels again, but with a poor lens it would still be a poor camera!"

How I agree with that. But the public has been 'educated' to think that the more mega-pixels, the better the image. Oh, how easily we're baited.

I liked your thoughts on the GR-PD1 review and would have liked to have read similar thoughts on my critical appraisal of Sony's PDX10 in the Feb issue. Remember though that CV is primarily a computer video magazine, and camcorders are but one small link in the chain that ends with a DVD. Because of this Bob (rightfully) limits the word count that can be devoted to this particular front end.

tom.

Thanks for your comments regarding my review. I actually had a physical copy of the brochure you link to on JVC Australia when I wrote it and I think you’ve interpreted what it says slightly differently to me.

First, the captured resolution. It’s not at all clear from this brochure (or, indeed, anything I’ve been told by JVC’s own UK representatives) what the actual resolution captured to tape really is in HI-RES mode. It’s definitely down-sampled to 720 x 576 when the footage is transferred to a computer using the supplied software. I can see how you’d say that it is in fact 576 lines from the table on page 4 – it calls the format 625/25, which would imply 576 lines. But on the last page of the document, the specs state that DV is at 839 x 576 whereas HI-RES is 1,280 x 659. In my opinion, that’s extremely confusing. DV is a standard and must be 720 x 576, but the HI-RES mode isn’t and only JVC knows what’s really going on and they didn’t give me any more information than was contained in that brochure or the manual (which has virtually no technical info).

On to colour fidelity. This actually is a problem for all digital cameras, including high-end SLRs, as they use a similar RGB filtering system to single-chip camcorders and have a colour resolution that’s one third the monochrome resolution. A good interpolation algorithm reduces the effect, and the overall resolution is so much higher than DV it’s less of a problem anyway. For a camcorder, with just 720 x 576 to play with, you’re talking a pretty low res for each colour channel. This is why professional camcorders have three chips. Larger CCDs do increase sensitivity and prevent bleeding between pixels, but they don’t get round the fact that the RGB filtering system used reduces colour resolution by a third. JVC’s Hybrid system, however, uses an array of eight pixels to produce the RGB signal by a complex algorithm of green, cyan, yellow and clear pixels, taken in three stages of four grouped pixels. This is shown on page 7 of the brochure. I didn’t say all this in the review because it was long enough already and would have required a lot of background on how CCDs work, which I didn’t think was appropriate for a CV review – more for an in-depth feature. Maybe I was wrong about that, but as Tom states the emphasis was meant to be on how the PD1 works in the computer editing environment, so I focused on that rather than the intimate details of how the CCD works. I think what I wrote was sufficient to explain why JVC can argue that the single-chip PD1 has colour fidelity effectively as good as 3-chip models.

Yes, you got me on the 50 fields/frames thing, but that was just a typo. I meant fields, I wrote frames. Sorry. Otherwise, though, I’m not sure what I really got “plain wrong” as I had the same document as you and merely stated what it says in that document, albeit interpreted differently. JVC are actually quite pleased with the article, as no other review yet in the UK has got to grips with what’s special about the PD1 in such detail, even the camcorder mags.

Thanks for the HDV link – that info will come in handy for future articles.

Finally, I must apologise for not owning a progressive scan widescreen TV. Sadly, JVC didn’t lend me one for the review and I really can’t afford to buy one, so I had to use what I have available. To be honest, I don’t even have enough space in my house for another large TV. Reviewing kit is not a perfect activity, and you never have all the facilities you need to do a fully comprehensive scientific job. As the primary market for this camcorder, in the absence of HD broadcasting, is DVD creation, I considered the quality of DVD output to be the primary factor to consider, rather than output to a progressive scan widescreen TV with component input which almost nobody has. And DVD output quality is truly excellent from this camcorder.

James Morris

[This message has been edited by cyberwest (edited 29 January 2004).]

What first struck me when I first came across the HD10, is that the colour-pixel layout and processing means that it's a Constant-Luminance camera, and possibly the first since the gand old EMI2001 of the 1960s. Since the luma signal is derived from linear mixing of primaries and secondaries, the signal is genuinely luminance and not luma. If they've gamma-corrected that and then doen all the processing in YRB or YUV, then it's much more of a breakthroguh than they're claiming.

I agree with your comments about its purpose in the UK, I can't see a real purpose for the Hi-Res modes until we have displays for it. the notion of using it as a DVD-feeder is as good as you'llget for it until then. And I agree that the documentation is confusing. You're right that all the standards it can conform to are published, and that 659 etc don't fit any of them; that's where the confusion lies.

I hope to get a bit more sense out of this in the next few days.

Having started the discussion, I'm gratified by the detail of the recent response!

Firstly, to Tom Hardwick, I had to fish out Feb's issue to refresh my memory, and now remember that I was impressed by it. I don't know that camera, but found your comments about the limited aperture range and use of ND very interesting, and highlight the reasons for using full size cameras rather than prosumer. Equally, objective comments, such as low light performance comparisons with the VX2000, are useful rather than comments along the lines of "it was very good in low light" etc.

I was also pleased to see your graph paper filming tests, which answered what would otherwise have been a question in my mind (as a TRV 900 owner!). Reading it again, I think a typo may have crept in: the numbers given don't correspond with the size of the boxes, and using your figures I think the most likely number to be wrong is the 16:9 vertical, and suspect it should be 648. Nicer still would be res charts, but to be of real use, they'd have to be standardised across all reviews (Bob?) but in these computer days would be a lot easier to do than in the days of film.

I also liked the DVvDVCAM comments box, and was pleased you didn't say that DVCAM was higher picture quality than DV. (I have read that, lots of times.....) Just for the record, a key advantage for professional use of DVCAM is that the tracks are wide enough to allow accurate INSERT editing, in a way that's not reliable with DV. That's of course only relevant if you edit tape to tape on DVCAM, and then only on the recorder in the suite. I also seem to remember that DV only has unlocked audio in the NTSC spec, in PAL DV is locked as well, but I'm not sure about that.

Regarding James's reply, my apologies for banging on about this, but I do believe that a fundamental problem that been around for years in the industry is confusion between the "camera" and the "recorder", bad enough in U-matic days, but now the two are in one box.... Now, is one just being pedantic, or does the confusion really matter? Well, I believe the latter, and one illustration goes back to when camcorders were going digital. (For recording!) Small DV cameras arrived amidst a professional world of Betacam, and I believe that the RECORDER on even the humblest camcorder RECORDS in higher quality than Beta SP. But is it a better picture? No, but thats down to lens, chip, signal processing etc - the CAMERA rather than the RECORDER.

What the above meant was that many good (and influential) cameramen associated DV/DVCAM with the image quality of the small cameras, and were minded to tar all DVCAM cameras with the same brush. I personally have compared a DSR-500 with an early DigiBeta (700) on a broadcast spec monitor, and the DSR-500 not only looked better, but was about 2 stops more sensitive to light! So, does this mean "DVCAM is superior to DigiBeta" - NO, because it was the CAMERAS that were being compared, and had the comparison been done with a modern DigiBeta, I suspect the conclusions would have been different.

But down to detail - regarding resolution, we really need a definitve answer from JVC technical, but I believe that only 720x576 are captured to tape. Yes, the documentation is unclear, but all the more reason for NOT actually stating that "1280x659 pixels...are captured to tape".

In your reply you say that ...."with just 720 x 576 to play with, you’re talking a pretty low res for each colour channel. This is why professional camcorders have three chips".

I think you're confusing two separate issues - resolution and sensitivity. Pro CAMERAS (not camcorders! :) ) have three chips for sensitivity reasons, not resolution. For easy numbers, imagine an imaging system with a resolution of just 4x4 pixels. Lenses limit one to square format of (say) 4cmx4cm. In a monochrome system therefore each pixel is 1cm square, and a certain amount of light falls on each one. We now want to move to colour, and have two choices. Either replace each pixel by three, (filtered R,G,B), or split the light to three unfiltered sensors, each one the same size as the original. In the latter case, no light is lost through filter absortion, and each is the same size as before, in the former each is only .3sq cm and light is lost due to absorption. But in each case, the luminance resolution is the same.

I take the point regarding the length of the review, but have to say that in spite of an engineering background I had no idea how the system worked from the CCD section your review, whereas I understood quickly from the page 7 diagram and associated formulae. And that was shorter than the space you devoted to the topic. The key point is the use of W/G/Cy/Ye rather than R/G/B and no matter how limited your space I would have thought that was the most important fact to present.

Regarding quality monitoring, it's not necessary to have a large monitor to assess the sharpness. One suggestion would be to film a test chart, import frame grabs into Photoshop, and magnify the relevant detail sections to see at what point detail is lost. Something along these lines seems standard practice in magazines such as Amateur Photographer. If you read the HDTV topic I started in "Chatter", then as Alan Roberts puts it, HD is not relevant for standard screen sizes, and I suggest that the screen size you use would mask quite large differences between pieces of equipment, even when transferred to SD DVD. Differences which CV readers would like to be aware of.

I must also disagree with ...."the primary market for this camcorder, in the absence of HD broadcasting, is DVD creation". It seems likely that HD will make it's presence felt in the UK firstly through pre-recorded material, such as Blu-Ray. Since many cameras are brought to record family memories for the future, this camera, or similar products, must have a big appeal in a "future-proofing" sense, definately because of 16:9, but the HD aspect must also be relevant.

I had been hoping that a definitive answer regarding what actually does exist on tape might have arrived from JVC by now....

To be honest, I think on reflection that you're right about the HI-RES resolution recorded to tape being 720 x 576. But that's just a gut feeling - I still think it's unclear in the brochure. I don't have the review in front of me at the moment, but I think I just said "captured" rather than "captured to tape", so even I have left it unclear regarding the camera / recorder dichotomy - but I'm merely reflecting JVC's own lack of clarity. I just don't have a definitive answer.

As for the sensitivity/colour resolution issue, I believe we're talking at cross purposes slightly. All of the digital still cameras I've come across from low end to semi-pro quote the monochrome resolution as the resolution, which is the total of all red, green and blue filtered pixels. The colour resolution (which they never quote) is therefore one third of the monochrome res. This is the same for single-chip camcorders, which use a similar RGB filtering system. In neither case is one pixel replaced by three - one pixel is interpolated to make three. The prosumer camcorders generally use a prism and three CCDs with a pixel count enough for a full 720 x 576, so the colour and monochrome res are the same. I take your point about the sensitivity benefit of not having an RGB filter in the way, but you should also take into the account the fact that high resolution CCDs are more expensive. This is the real reason why the JVC is the first camcorder at the prosumer level to use a larger, higher res CCD to increase colour resolution rather than 3 CCDs. It's now cost effective to do so. I expect JVC won't be the only company doing it in the near future. Using a physically larger CCD has always meant a better image for the obvious reason that more light will fall on each pixel, making the camera more sensitive. I think that's what you already said.

I'm not sure I agree that page 7 of the brochure is better than what I wrote. While you may have understood it easily, I'm pretty sure many CV readers would have no idea what it was talking about, which is why I tried to simplify it to the important end result. How the colour filtering system used with CCDs works is not very common knowledge for the average user. The fact that W/G/Cy/Ye is used by the PD1 instead of R/G/B would have been pointless technical detail for many people, who really want to know what the end result of that is - which is what I wrote. I fear this is another example of video engineers getting lost in the technical detail rather than realising what it actually means in terms of quality output.

I really wanted to do some frame grab type tests with the PD1 but there's actually no software available that reads the raw MPEG2-TS files and does frame grabbing. It has to be converted to standard MPEG-2 first which inserts a layer of error. That's also why there's no full-res screengrab in the review - I had to burn a DVD and screengrab off that to get the one I used. Wider software support for MPEG2-TS will make this possible in the future, though.

I understand what you're saying about BlueRay and HD, but by the time we have this in the UK the PD1 will be obsolete. The PD1 may encourage this market to start, but anyone thinking about buying one now will probably not be using it when we have HD broadcast and pre-recorded discs. So, again, I focused on what you'd use the PD1 for right now, as "future-proofing" more often than not turns out to be a white elephant with digital technology. It becomes obsolete far too quickly.

This will have to be fairly brief, but:

You actually wrote "....the PD1 captures video at a resolution of 1,280x659 pixels..... This is recorded to DV tape in MPEG-2 format at...." and later "widescreen being recorded at 941x485 pixels, and standard being 839x576". Subsequently you refer to a "regular DV mode at 50 frames" (!) "per second, but the resolution is still 839x576 pixels". The clear implication from the review is that these are the resolutions recorded to tape, which I don't believe to be the case. Yes, JVC haven't written clearly, but all the more reason for not stating it definitively in a review!

Referring to the right hand two columns on p28, dealing with the workings of the chip, and calling myself a "canny reader" I did indeed spot that "Hi-Res ....uses fewer than 844,000 of the 1.18 million of the chip". Rightly discounting that all the shortfall is not down to management functions, you go on to say that "The answer is in JVC's Hybrid Complementary-Primary colour filter system...." Errr, I'll think you'll find if you do the maths that the shortfall is simply due to the chip having 4:3 dimensions, and the Hi-Res mode being the center 16:9 part of it - and nothing to do with what you claim!

Was this right hand column on p28 what you referred to when you posted "another example of video engineers getting lost in the technical detail rather than realising what it actually means in terms of quality output."? Considering some of the detail in the review, I don't think what's in the JVC brochure is too technical for CV readers, given the specialist nature of the magazine! Perhaps others may wish to visit the link and give their opinion?

There has been software around for a while now to convert MPEG-TS to MPEG-PS - mainly useful for recording DTT to DVD. I'm not quite sure what you mean by "standard" MPEG-2 - MP@ML?

I could say more, but have to go, and will be away next week. Hopefully by the end of then we will have the definitive answer from JVC on what resolution it really does record!

Okay, so you agree I didn't say "captured to tape" - you misquoted me there. You're also doing me an injustice with your other examples, as I clearly state on p29 of the review "Despite the PD1's 839 x 576 recording resolution, DV is still captured at 720 x 576 like any other camcorder". And please stop banging on about the fact that the review accidentally said frames when it meant fields for standard DV. I've already addressed that - you don't need to keep mentioning it. When I sent the review in to CV that sentence said fields and it was edited to say frames. It's a typo. Leave it alone.

I agree that the math would imply that 1,280 x 659 is 16:9 of a 1,280 x 960 pixel CCD - which is in fact the stills resolution. So maybe that's a bit of a non-sequitor on my part. However, I still think we haven't got to the heart of how the Hybrid Complementary-Primary colour filter system manages to get 0.75 of the chroma pixels compared to luma, where for conventional single CCDs it's more like 0.33. Just using a differently coloured filter and a different interpolation algorithm doesn't explain it (and neither does p7 of the brochure either).

We'll have to agree to differ on what level of technical detail CV readers want. If that kind of detail is required, then I'll be happy to put it in my next review. However, I've written for CV since it launched and the bias has always been towards what the technology allows you to do, and the technical detail must complement that rather than be an end in itself.

Finally - yes, I meant MP@ML by saying "standard MPEG-2". While there is software to convert MPEG2-TS to MPEG2-PS readily available - that's primarily what the software supplied with the PD1 does - I haven't found anything that will screen grab a frame from the MPEG2-TS file, which is what you need to test the PD1's output quality in any kind of scientific way using the Photoshop method you describe.

[This message has been edited by cyberwest (edited 30 January 2004).]

I don't think I did misquote - you clearly at the top of P27 say "This is recorded to DV tape in MPEG-2 format ..." and no reader can be left in doubt that "This" can refer to anything other than your previous sentence - referring to the PD1 capturing video at a resolution of 1,280x659.

Regarding p29, you do indeed state that it is imported into the editing system as standard DV, but compound the earlier errors with the first phrase - "Despite the PD1's 839 x 576 recording resolution" - 839 x 576 simply is not the "recording resolution".

I think p7 of the brochure explains the "Hybrid Complementary-Primary colour filter system" quite well, if needing a bit of interpretation. Just give a bit of thought to what the formulae actually tell you. Apart from the admitted non-sequitor, I'm not sure about your following paragraph either. The formulae seem to indicate that whilst luminance can be derived from pairs of pixels (either W/G or Cy/Ye), chrominance requires all four of the above - and therefore chroma resolution is half that of luminance, rather than the three quarters you state.

I'll agree in principle with the last sentence of your last post, but feel that whilst the technical detail must indeed complement (what the technology allows you to do), the technical detail should either be accurate, or left out altogether. I'm afraid that the level of fundamental inaccuracies here hinder rather than complement the other aspects of the review, and feel that a correction should be printed next issue.

Infocus - thank you for your thoughts on my review of the Sony PDX10p. It's the first constructive criticism I've received and I welcome it. In a way though the PDX10's fiddling with three automatic ND filters and the use of a nasty (nasty in optical engineering terms only) two bladed diaphragm should not matter two hoots to the end user. Sony, after all, is at pains to hide that from any sort of technical specification I could find anywhere. I've seen many tests of this cam (and the TRV950) that say what excellent pictures it gives, testers being oblivious to the fact that Sony blank off a vast range of apertures to ensure this is so. Ergo, should CV's testers ease up on the 'how it works' and delve more into the 'what it gives us'?

I'm glad you liked my DV CAM appraisal. I too have been so disappointed to read that DV LP gives 'grainier, fuzzier pictures', and even CV tests of camcorders have been very lax (err, inaccurate) in this respect.

I agree that resolution charts are the way to go, and it works very well on the web. (Broadband recommended for the big files if they're to be any use). But nowadays you can't test camcorders at f8 - the MX500, TRV950, MVX3i etc with their auto NDs won't allow it.

Testing camcorders will always be a very subjective thing, and finding the balance in a computer magazine is difficult. A review on a website can be as long and convoluted as you like, and beginners to experts can weed from it what they will. Bob obviously doesn't want readers to sigh 'Oh, another page on this Canon', yet he does want reviews to be a lot more poke-in-the-eye than Camcorder User publishes, say.

I liked the review of the GR-PD1 but didn't envy James Morris his task. Any 'more advanced' technology brings with it pros and cons, and I can't help thinking that for all its wizardry JVC could've made a better camcorder at that price if they'd stuck to the DV specification, and not gone out down a quaysi semi-high def route, don't you?

As to thoughts on colour resolution, I'd like to say this.
With a 3 CCD camera, there's a prism at the back of the lens that splits the light into red, green, and blue components, which then go to three CCDs. With a 1 CCD camera, the light hits a single CCD with red, green, and blue (or sometimes magenta, cyan, and yellow) filters in front of the CCD. This means that for every pixel on the 1-CCD camera, there are colours that are invisible to it. If a tiny spot of blue, say, happens to fall on a red-sensitive area of the CCD in the 1-CCD camera, then it will be lost from the image. But in a 3-CCD camera there are no such spots – colour sensitivity is uniform across the whole field.

This isn't quite as bad as it sounds, though. With very few exceptions, colour resolution is much less than black and white resolution in all video systems. In DV, colour has only 1/4 the resolution of luminance (brightness). But you still lose something. More seriously, for a 1-chip camera to have the same luminance resolution, it has to have at least three times the number of pixels as a 3-ccd camera, which is still relatively unusual. But at the same time, it's worth noting that images shot with 1-chip cameras can look amazingly good. It isn't bad versus good, it's good versus even better. And to date I haven’t heard of a single digital still camera – regardless of cost – the uses three chips in preference to one.

If we look beyond consumer formats, a 3-chip analogue Betacam camera will outperform a 1-chip DV camera and most 3-chip ones as well, so even well lit digital isn't necessarily best. Standard definition digital video is limited to 720 pixels horizontally and a maximum theoretical horizontal resolution of about 530 lines, and cameras such as the TRV900 are pretty much at the theoretical limit, which is pretty impressive. Lighting is of course hugely important, and even an amateur camera with excellent lighting can often produce results that look better than a pro camera's output with poor lighting.

NTSC DV uses 4:1:1 colour, which simply means that colour (chrominance) is sampled horizontally at 1/4 the rate of brightness (luminance), giving a colour resolution of 180 by 480. PAL DV uses 4:2:0 colour, which means that colour resolution is halved in both directions, down to 360 by 288. (This sounds better than it works in practice, and some PAL video systems get better results with 4:1:1.)

tom.

Infocus, 839 x 576 *is* the DV recording resolution for the PD1. It's downsampled to 720 x 576 before being recorded to tape, but it's still the resolution used by the CCD when the camcorder's in DV mode. As you said in an earlier post, there is a confusion between camera and tape system, and in my next review of this type I will try to make it clearer what I mean - which in this case was the CCD's pixel resolution.

However, I did believe the HI-RES mode was recorded to tape at 1,280 x 659 when I wrote the review, although now I'm not so sure. I will admit I failed to register the implication of the little table with 625p written next to that mode, and everything else in that brochure implies HI-RES is recorded at 1,280 x 659. We'll still have to see what the JVC techies have to say on the matter to be sure, though.

As for the colour filtering system, maybe if you're a TV cameraman by profession you'll understand the Hybrid system as described on P7. But most of CV's readers are hobbyists and semi-pros, so I suspect that p7 would be gibberish to most of them. Oh, and the 3/4 chroma res I mentioned is direct from JVC's brochure (p7) so your calculation that chroma should be half rather than three quarters shows that p7 really isn't an adequate explanation of how the system works - your calculation doesn't match what JVC states.

On a final note, I would like to underline that I've read PD1 reviews in a number of magazines, and none of them addressed adequately how the PD1 works with computer editing systems. This part of my review is spot on. The technicalities as to what CCD resolutions are being used can explain certain quality issues, but when it comes down to it, it's the end results that matter. The quality the system is capable of when used with computer editing systems is the most important thing. I got all of that correct, and as that's what a magazine called COMPUTER Video is all about I really don't think my review deserves the negativity you've expressed.

I didn’t really want to post again so soon (and after this definitely won’t be able to for a week), but:

Tom: most of what you say I agree with, and whilst I understand what you’re saying with your spot of blue argument, I think that if the spot was that small it’s loss wouldn’t matter – but interesting point! I couldn’t agree more with your comment about lighting, though of course good lighting with a pro camera is best of all…..! In practice, pro camera lenses and larger chips make it easier to reduce depth of field, so that for an interview (say) it’s easier to throw the background out of focus than is possible with small cameras.

You refer to the PD-1, and ask if “for all its wizardry JVC could've made a better camcorder at that price if they'd stuck to the DV specification, and not gone out down a quaysi semi-high def route, don't you?” Difficult, and as with many TV matters nowadays, we have to remember that the 50 Hz world is considered secondary to the 60Hz world. What I think we’re seeing is a derivative of the HD-1 (may be called 10??) and whereas that really can be considered revolutionary, (first to use HDV, unless anyone knows different?) and first camcorder to record a true HD format to tape, it seems a lot of the “revolution” has been lost in the transition to the 50Hz world. It is interesting to note that your review states the PDX-10 to have 3 1,152x864 pixel chips – so I believe a theoretical (!) maximum resolution of 576 horiz line pairs which could be seen to be better than the PD-1. (Whilst that has 1,280 horiz pixels, a luminance signal requires two for derivation, so hence we’re talking about 320 line pairs. In the first reply to my original post Alan explains it all (and explains how some detail does continue beyond 320 lp) and I recommend everybody following this topic to read again what he said on the 26th January.

But these are only sensor figures, (NOT recorded resolution!) and with so many variables the only sensible way to review resolution is with test charts. And don’t do the comparisons on a TV, even an expensive TV, import stills via computer as I’ve suggested before.

James, I’m sorry you see my posts as negativity – I assure you I’m trying to be positive – I’m trying to stop a lot of false “facts” becoming accepted as common wisdom for reasons I explained earlier. In my first post I queried several key points about the camera, and expected either to be definitively corrected, or told “sorry, yes, that was a mistake” the true answer is….. Everything I’ve subsequently read has increased my belief in my original comments, and you appear less interested in getting to the bottom of matters than that your review has been criticised.

Regarding your last post, the first sentence is itself an oxymoron – 839x576 is NOT “the DV recording resolution….” It’s the no of pixels downsampled to derive the 720x576 DV recording resolution, and they are two totally different things.

Regarding the 3/4 chroma res you mentioned, I don’t think JVC ever did state that. I think you may have misinterpreted the statement they do make about ratios of chroma pixels to luma pixels (1:1 vs 0.75:1) which is not the same thing at all - I can only advise you again to look carefully at what the formulae say to see how relative resolutions come about.

Okay - it just seemed negative to me. I think we're just arguing over words now which won't get us very far. What you think I meant to say wasn't actually what I meant to say - I know DV is always recorded to tape at 720 x 576 - that comes from the CCIR 602 standard it was derived from, I believe. If what I wrote is easily misinterpreted to mean that 839 x 576 is actually recorded on the tape in DV mode then maybe I should take a look at the way I phrase these things. The MPEG-2 modes are a different matter, as I susect MPEG-2 could theoretically be recorded at 839 x 576, then downsampled when the footage is transferred to the computer - the camcorder does downsample on the fly for some of its formats when outputting over analog connections.

I'm afraid I still don't get what you're saying about the colour system. What I don't understand is that if a three-chip camcorder has three CCDs with 420K pixels, one for each colour (ie around 1.2Mpixels in total), and thererfore has 1:1 chroma to luma, how does a CCD that samples using a 16:9 grid of 844K pixels manage to a ratio of 1:0.75? It would need to have 633K chroma pixels, which the diagram on p7 doesn't imply (to me at least) the Hybrid system in the PD1 has. The Bayer system uses every pixel for luma but alternates RGB for chroma, so each colour is one third the resolution of the luma signal. The Hybrid system seems to use every pixel to make the luma signal, only four out of a block of eight for each chroma, shifting down a row for R, G and B. As you said in another post, that would imply half the chroma resolution to luma. But in the box JVC states that the ratio of chroma to luma in 3-chip is 1:1 and for the PD1 it's 0.75:1.

It's just occurred to me that we seem to be working with a different definition of the word "resolution". For me, the number of pixels IS the resolution. That's how it's stated in digital cameras. Surely, 720 x 576 is the resolution of DV? That's pixels, right? I just don't see what you mean by chroma resolution being different to the number of pixels.

To be honest, the more I try to understand how the Hybrid system works, the less confident I feel that I have a handle on it. Contrary to what I suspect you think of me, I'm not a stupid guy and I have quite a bit of experience writing on video subjects. I suspect that the reason you understand it from the JVC brochure (assuming that you actually do) is because you're a professional cameraman. If I didn't get it from JVC's brochure, I would argue that most of CV's readers won't have either. So, please, since I seem to have got it wrong, and you know how it works from the brochure, please explain how the Hybrid colour system works so I do get it.

Haven't time to post properly now, was just checking e-mails before going away, but will try to expand when I get back. But in brief - you are now asking the right questions!! And no,it's not really true to say that no of pixels is the resolution in quite such definitive terms. Talk of "no of pixels" can give indications of resolution, but needs qualifications, and "pixel" can be used in different contexts. None of that's very well written or complete, but I'm in a hurry! It's also complicated by non-square pixel issues.

Thats why I keep on stressing the importance of test charts, since they are the best way to get an overall objective view - not only regarding chip, processing, and recording, but lens as well.

I suspect your last comment that "MPEG-2 could theoretically be recorded at 839 x 576....." is probably true, but that in this case it's unlikely.

Chaps, I haven't the time to read all of this thread again, pressure of work...

Think of this camera as being a Hi-Res source for DV-Res recording. Its oversampling the image, providing the compressor with an image at higher resolution that it needs for tape. the compressor then has a releativekly easy job of filtering it to the recorded resolution. Broadcast cameras have been doing that for yonks, typically over 1000 pixels now for SD output. What's different about this camera is that it oversamples vertically as well. That helps with vertical resolution in the same way.

Since the "aperture" of the camera is defined by the Sinc function (Sin(x)/x) it has a droop-off in amplitude response of 4dB at half the sampling frequency (the content-frequency limit of the system) and this frequency is defined by the sensor. Since it goes up to 839, that means it's 4dB down at 839/2=419.5 cycles/picwidth whereas a normal camera would be 4dB down at 720/2=360. So this camera holds its frequency response much better within the band that gets recorded, so it needs less aperture correction (or detail enhancement), so the pictures shouild look less ringy. the same goes for the vertical numbers.

It's a good way of using a big ccd to get good pictures at lower resolution, the big boys have been doing it for many years, now the consumer gets top do it as well.

So, er, Alan - do you understand how the PD1 achieves 0.75:1 chroma to luma? Is that because of the resolution downsampling from 1,280 x 659 to 720 x 576?

No, it's because of the way the subtractive filtering works. They aren't claiming full luma resolution, they can't because they have to do running signed summation of adjacent pixels along the line. Chroma does the same but does it vertically as well. I can't see how they get 0.75:1 (or more traditionally 4:3:3), but it depends on where you define the edge of band; the top half of the frequency ranges of luma and chroma are each aliased because of this signed summation, so it really depends on where exactly you decide that the aliases are masking the real detail, they've decided it's at 4:3:3. Apparently.

Thanks for that - what you say is in keeping with how it's described in the brochure. But I am definitely beginning to think there's a healthy dose of marketing bollocks in that brochure - surprise surprise. Still, the output really is excellent. I didn't directly compare it to a VX2000 or XL1S, but I've tested both those using similar methods to the ones I used for the PD1 and it certainly holds its own pretty well in DV, and widescreen DVD output is of course exemplary. I think Infocus is right, though, that the HD1 is more revolutionary for the US market than the PD1 is for the UK market, because it doesn't actually have a true HD mode.

Having just returned, James, I’ll have a go at a fuller explanation so, as you said on the 1st Feb, that “you do get it”! Firstly, your comments about resolution: you talk about definitions of “resolution” and say “For me, the number of pixels IS the resolution.” Well, imagine two cameras with the same no of pixels each, everything identical apart from the lens, A with a good lens, B with a bad – B will have a lower resolving power than A! To me, “resolution” is the relative ability of a camera to distinguish fine detail, and all that is meant by “no of pixels” in isolation is, well, no of pixels! As for your comment “That's how it's stated in digital cameras” – yes, it’s an easy figure for a brochure writer to quote, and in the stills world all I can say is that not all 4 megapixel cameras produce the same pictures! Hence my keeping on about test charts in most previous posts – they do truly indicate resolution.

Regarding the 0.75:1 chroma matter, I think this is a red herring and you’re inferring something that JVC don’t actually say – what I think they mean is that 3 out of 4 pixels deal with colour information (the fourth is clear, or white), and that is not the same as saying that chroma resolution is .75 that of luminance, but more of that later. Whilst the statements perfectly true, it doesn’t really tell you very much useful!

You said that “the Bayer system uses every pixel for luma but alternates RGB for chroma, so each colour is one third the resolution of the luma signal” – whilst every pixel is indeed used for luma, you need at least three to derive it, so I don’t think for that system colour would be one third that of luminance. Incidentally, Bayer filtration doesn’t consist of equal nos of R,G,B pixels, but rather a matrix where half are green, and a quarter each of R and B.

The best way I can elaborate is to suggest thinking of three types of pixels – actual pixels, “derived pixels” (which may only be implied), and recorded pixels. The best illustration of the principle I’ve seen is at http://electronics.howstuffworks.com/digital-camera5.htm and the animation showing “demosaicing algorithms” illustrates the difference between “actual” and “derived” pixels quite well, as well as being one of the neatest bits of animation I’ve seen on a web page! You see the “demosaicing algorithms matrix” (of 3 pixels square) step along to derive each R,G,B pixel from 9 actual pixels – turning an x by y matrix of single colour values (say 8 bits each) to an x by y matrix of three colour values (3 x 8 bits each). Important to note that because each IS derived from 9 actual pixels, the derived resolution will not be the same as from a 3 CCD equivalent of 3 chips each of x by y.

In a still camera this derived matrix could then be captured directly, but it could also be captured in a smaller form – say downsampled to one quarter the number of pixels, and therefore the “recorded pixels” would be a different number again. See what I mean about three different definitions – the no of “recorded pixels” need bear no releationship to the no of actual, and, as is the case with the PD1, even the ratio of vertical to horizontal may differ between recorded and actual! (It has to because of different aspect ratios, and the non square pixels of DV.)

A big difference normally between still and video cameras, is that the former tend to capture R,G,B for each pixel, the latter normally defines it’s resolution by the number of luminance (Y) pixels (recorded), and then records chroma difference information with a quarter as many each of U and V samples as of Y. (At least for DV video.)

If we take the JVC definition of Y as R+2G+B, then it can be seen that each four pixel block in the Bayer system uniquely defines a (derived) luminance pixel by summation, and U and V can be similarly derived. Turning to the Hybrid system, then for each comparable 4 pixel block TWO unique Y pixels can be derived – for the top pair, white (R+G+B) and G add to form Y, and for the lower pair yellow (R+G) and cyan (G+B) also add to give R+2G+B – or Y! To form a colour pixel, adding various combinations out of the four will give any desired derivation. So what we see is that four ACTUAL pixels can lead to two unique DERIVED Y pixels, and one each of U,V DERIVED chroma pixels - hence a luminance resolution twice that of chrominance.

What’s more interesting is to compare the amounts of light falling on the pixels in the two cases. To take four pixels in the Bayer case, the total for the square is R+G+B+G, or R+2G+B. In JVC’s case, it’s R+G+B (white), G (green), R+G (yellow) and G+B (cyan) – a total for the four pixel square of 2R+4G+2B! Twice as much light falls on a four pixel block overall as in the case of the Bayer matrix, hence the sensitivity would be expected to be much better, and note how green is not filtered off ANY pixel! Hence my comment many posts ago about 3 CD’s normally being preferred for sensitivity reasons rather than definition or colour idelity.

I’m aware that all the above may not be very rigorous, and confess that I’m not sure about some of the finer detail myself, but I hope this does help with the general picture. It was a much easier visualisation when the recorder and camera were two separate physical units, and it is useful to theoretically think of a camcorder like that.

You finish off with “Contrary to what I suspect you think of me, I'm not a stupid guy” – I never said you were – and wouldn’t be going to all this effort if there didn’t seem to be any point! The whole problem is that a lot of people have written misleading things, lots of “false facts” are in circulation, and if I can help get the true story across, then as a writer you may be able to help put the situation right. Especially important if we do see many more HDV or high definition consumer items, and my feeling is we will, even though there seems to be deafening silence from most manufacturers at the moment.

I don’t really want to write any more about this, but am happy for those in charge of the list to give you my e-mail address if you want to contact me direct. A 15 minute chat with a pencil and paper and you’d wonder what the problems were all about….. I’ve just found out I’m working in Yorkshire all next week, otherwise I’d have been at Videoforum.

I've been going through last weeks mail, and found the Videoforum stuff - catching my eye is the leaflet about the JVC JY-HD10 - HD, as in High Definition, not Progressive scan?! I'm posting a separate thread about it in "Chatter", but all I've seen to date indicates that that model is a 60Hz world model only - so have JVC brought out a 720p25 version?

Whilst searching about it, I came across http://videosystems.primediabusiness.com/ar/video_dv_dead/ , a review of the above camera, which has relevance to this thread. Looking at the paragraph "An Inexpensive High-Def CCD?" I was relieved to find that he too refers to "The 2:1 ratio of luma samples to chroma samples !

I was initially confused by the second paragraph of that section with talk of "480×720, 4:3 window " etc - I think it would be better worded "4:3 window which is used for the 480×720 recording". Other bits of the review hint at the answer to something which has puzzled me (and I'm surprised hasn't been commented on before).

In 4:3 stills mode, the camera uses the full 1280x873 pixels for it's image, in 4:3 DV or p50 mode it only uses 839x576. Since in these latter modes I hope we are now agreed that it only records 720x576 to tape, this may not be a big issue in resolution terms, BUT, and from a cameramans perspective it's a really big BUT, it makes a lot of difference to the angle of view of the zoom range. And a big issue with prosumer cameras compared to proper pro cameras is their wide angle capability.

The last page of JVC's document (on the PD-1) even gives comparable 35mm equivalent figures - max wide angle of 37.1mm equiv for full sensor versus 56.5mm equiv for DV! In 35mm terms, the difference between a moderate w/a and a standard lens!

So why have JVC done this? The only reason I can think of, hinted at in the American review, is that the digital signal processor just could not handle the data rates coming off the full chip. A better chip would have cost more, so the angle of view has been compromised in these modes. From a cameramans point of view - a great shame. I'd be interested if any one could throw further light on this.

And for anyone going to Videoforum next week, a question for the JVC stand?!

Maybe I'm misreading your post, but this 'wider wide-angle in stills mode' has been happening for some time now. The first Sony PC cams that sported mega-pixel chips put 720x576 to tape in DV mode, but when the full area of the chip was used for stills to Memory Stick, the wide-angle coverage increased. And most noticeably, too - although the EIS was abandoned of course as there was now nowhere to float the image.

In the DV mode the tape was being fed info from only the central portion of the lens of course, so that made movies look even better. The TRV950 has a 16:9 wide-screen mode that actually reduces the wide-angle coverage, whereas the PDX10 has what appears to be a 0.8x wide-angle converter added when its widescreen mode is switched in, and this has to do with the full width of the chips being employed.

tom.

No, I think you read me correctly - I assume your second paragraph refers to lens performance being better at centre than edges?

Better still though to use all the area of the chip for video and downsample to 720x576 for recording? AND that would give better w/a coverage?! Am I missing something obvious, or is it as previously said - the camera can't process the data fast enough.

It does seem that the video side of the camera (in regard to lens coverage)is being compromised for the stills ability. For the given size of image sensor, why not just have fewer, larger, pixels - and improve sensitivity as well. Personally I'd trade the stills ability for that and just buy a separate stills camera!!!

As a dedicated JVC person ( I have only ever used JVC) I have been very interested in all the debate on the PD1.

There is no way I wish to comment on what has been said as my knowledge on the subjects do not go that deep. But, I would just like to say this.

The Camera is obviously aimed at people like me. I have the GR-DV2000 and have used it for the past two years and been very happy with it. So much so that I have seen no reason, or advantage to go for the 3000 or 4000. Friends of mine have these Cameras and although they have a larger CCD, 1.33 Meg as against my 0.8 Meg, I do not notice any difference in the produced en' result.
In fact I used my camera for many months without realising that I did not have the Progressive Scan feature turned on. But on or off, again no difference in en'results can be seen. This also applies to the TV the tapes/DVD's are being watched on,ie If it is Progressive Scann or not.

No my eyesight is not bad and this is also the opinion of all the people that have seen the Tapes/DVD's produced. So my question is quite simply this.

What is there about this Camera that should make me want to get it? or put another way, what will this Camera do for me that I cannot do at present? I ask this because for some time I felt as though I should upgrade to a better Camera.( I almost got a second hand GY-DV5000). So when I knew the PD1 was coming, I thought that this may be the one to have.

When reviews are presented like the one in CV Magazine, they are not only of use to the Professional people engaged in such matters, but also to people like me who are trying to establish just what the Camera does and is it right for them. What has been debated here on highly technical matters, although of interest, does not help people like me to come to any decision as if to get it or not. Having said that, I feel Gary has put accross many points that have been helpful in that manner.

So, what has the Camera got that would make me want it, because it will do things my present Camera cannot. My main concern being that what I produce ( Video Tapes and DVD's about Athletic Coaching and Training ) Most of which are played on normal TV's as VHS Tape or the DVD's that are now out numbering the tapes by demand. Your comments on this are most welcome and I am sure will be helpful.

As for the technical debates that have gone on the only one that I would like to comment on is the subject of Resolution in Pixels and its relation to quality. No I quite agree it cannot be used solely to define quality. It is more an indication to the size of a Canvas that a picture can be put on. Just how that picture will look will depend on the tools and ability of the artis. In this case the Lens etc and the ability of the Camera operator in using them.
Resolution is pushed much to much in the stills camera world and certainly a high resolution camera does not always mean, better quality. ( Sorry about adding my comment, I just felt like contributing something )

Looking forward to any advice, all the best, Ron.

Hello!
My name is Sam Sharpe; I am based in the UK. I have recently purchased JVC’s GR-PD1 Camera and have been slightly disappointed to find out that it down converts Hi-Res footage to 576P @25fps, I am wondering if Aspect HD will support capture of footage in HI-Res mode at 1280 x 659 pixels as this is stated as the recording resolution of Hi-RES footage in the manual.

I know the US version the GR-HD1 supports 720P and I am assuming that as this is the same camera series that the UK version should record MPEG2 data at the resolution 1280 X 659 onto the DV tape which is then down converted to transfer to the supplied capture utility, Would Aspect HD solve the down conversion problem? i.e. can it capture the MPEG2 footage to an 1280X659 MPEG2 stream into Premiere Pro? Or even Re-size this 1280 X 659 file to 1280 X 720P and then supply this MPEG2 file to Premiere?

I really want to get these issues sorted as I feel like I have purchased a crippled product knowing that the same Camera over the pond can shoot 720P HDV @30fps, Maybe I should have tried to get the HD1 but import duties here in the UK add greatly to the cost. It’s a very strange thing for JVC to have done as I work primarily in 3D and with compositing the issue of HD frame rates comparing to PAL doesn’t apply, as to my knowledge 720P HD is a standard which doesn’t come in PAL/NTSC/SECAM variations, I believe it’s a constant 30fps.

I would really appreciate some advice from a knowledgeable source.

Thanks
A concerned GR-PD1 owner

PS anyone heard of firmware updates to switch a PD1 into a HD1? prob wishful thinking!

           
Ron,

You wrote:
“What is there about this Camera that should make me want to get it? or put another way, what will this Camera do for me that I cannot do at present?”

1. Unlike most “prosumer” cameras, it appears to be “true wide-screen”

2. It should yield sharper (better spatial frequency response) standard-definition (720 pixels x 576 pixels) pictures than a conventional “prosumer” camera would.

Alan

[This message has been edited by Alan McKeown (edited 11 February 2004).]

Firstly, to Sam Sharpe, although I don’t have the manual for the camera, I strongly suspect that it DOES NOT say that “1280 x 659 pixels… is stated as the recording resolution of Hi-RES footage …..” – which is the whole point of this thread! (Read again my first post and Alan’s reply.) What I suspect it DOES say is something along the lines that 1280 x 659 pixels on the chip are used to derive the image for the Hi-RES mode – it’s then downsampled and recorded to tape as 720x576. If you look very carefully, I think the manual will elsewhere talk about recording at 625p (should more accurately be 576p). My understanding is that 1280x659 never gets recorded to tape as such.

If it really does talk of a “recording resolution” of 1280 x 659 in the manual, and you are very unhappy, I would think this would be sufficient grounds for JVC being legally obliged to take the camera back and refund your money. I regret though, that you have probably fallen into the same trap that James did before writing the review – but I would be interested in the precise wording in the manual.

Secondly, no, 720p is not just a 30Hz standard – 720p at 25 or 50 Hz are perfectly OK. And yes, I fully agree with you that “It’s a very strange thing for JVC to have done”! You may be interested in what Alan Roberts recently wrote in the strand “Videoforum HD camcorder competition” in “Chatter”.

Your talk of “a crippled product” is perhaps a trifle strong – from all I’ve heard the results in the Hi-Res mode are very good, albeit standard definition, and it is a true 16:9 mode – but I understand why you feel as you do. I suspect your post may be the tip of an iceberg with many more buyers out there feeling short changed. My earlier posts may have seemed pedantic, but reading your letter makes me even more convinced that this is a matter which needs to be publicly clarified, and far from just being a matter of wording is of significant importance to you and others like you. If you’re across the thread, Bob, will we see that clarification in CV?

Have you been in touch with JVC themselves? I'd be interested to know what their response is.

Ron, what you ask is interesting, and I fully agree with Alan McKeown’s comments. Regarding your primary use (Athletic Coaching and Training) I would also add that the progressive scan feature may be useful, and 50p should give improved slow motion if you need it. I mentioned “future-proofing” earlier, by which I meant features such as 16:9, true HD etc rather than issues such as HDV format. (In the way that my first domestic camcorder was S-VHS rather than VHS – whilst VHS in all flavours is disappearing, the superior quality of S-VHS is relevant in the transfer to DVD – and whatever will supersede that!) In cameras such as this and future HD ones, currently the quality issues will probably tend to be masked by the display up to about 32in, but screens are getting bigger and better, and soon previously unseen differences will become noticeable.

Technical issues are relevant, but as a cameraman so also are the “ergonomics” of a camera – such as whether it is possible not just to manually adjust iris, but be able to do it without wobbling the camera! It still amazes me that a proper full size camera is easier to operate than a prosumer one – just grab a control and turn it, rather than fiddling with a menu – though I won’t worry about that on holiday for the sake of weight…….!!! :) Equally important for day to day work are issues such as lens coverage, connection and control of sound sources, etc which is why for most work most cameramen will choose a pro camera over a prosumer one any time.

In direct answer to your question though, I would be tempted to wait and see. I think a lot more HDV models will appear soon, hopefully true HDTV rather than just “Hi-Res”, and who knows, maybe you’ll even be tempted away from JVC. :)

I've been out of the office for a few days, but I've got more info.

I can confirm that it does always record 576 lines to tape. I can confirm that it does not necessarily use all the ccd in all modes, it goes for the easiest approach, that of cropping it to pixel map the image to the format, so the apparent lens angle changes as more or less of the ccd is used. Only in the "full PD" mode does it use all the ccd, but it uses an intyerpolator to get 576p from the ccd. It's pretty much as we'd all worked out.

I can also confirm that there will be more, many more, cameras in this niche, very soon. I can't say who from, or what the details are, I still have to observe confidences. NAB is in early April, you can probably expect to hear some announcements there.

In other words, and unlike the American version, not quite the "Hi-Res revolution" that we were led to believe!!

I think my question now, Alan, must be WHY? Using as much of the chip as possible could only improve quality, but from a cameramans perspective the most important aspect is how badly the angle of view at the wide end is compromised. That's one of the areas where prosumer cameras really suffer anyway compared to big cameras, and what JVC do more than negates the effect of a wide angle converter even!!

Which I think we'd pretty well guessed! I'm surprised more hasn't come out about it all, since I for one think the implications are huge, not least as regards the use of such cameras for low budget film production - but perhaps a lot of people are having to observe confidences? The big question is whether the forthcoming models for PAL countries will, unlike JVC's offering, be true HD.

The reason WHY is simple, because it's cheaper to pixel-map from the sensor to the output format than to do the interpolation. It's always better not to interpolate than to do it cheaply or badly, so maybe it was a sensible compromise. I don't know, I've not seen or used a PD1.

I couldn't possibly comment :D

Coming home after a hectic week or so working abroad, what should I find in my inbox than a link towards an announcement by Sony of, guess what, an HDV camcorder!! See http://www.sony-europe.com/PageView.do?site=odw_en_EU&page=PressReleaseDetail&section=en_EU_Press&pressrelease=1079021485605 for the official Sony link. A photograph of the prototype is also available (together with more information and good comment) at http://www.camcorderinfo.com/display.php3?display=sony-hdv-prototype-camcorder-03_17_04&ref=email .

Interestingly, first announcement is from Sony-Europe, so the question I earlier posed…..

has been answered. Very satisfactorily It also appears to be 1080i, (the HDV 1440 horiz res. as opposed to the square pixel 1920 figure) though one must now wonder whether it will also support 720p recording. It also seems to be 3CCD, so will hopefully have better sensitivity than the JVC cameras – the big question now being size and resolution of the chips, and will they be native 16:9?

The blurb at camcorderinfo.com contains a quote that “…..This is the biggest news in the camcorder industry since MiniDV came out.” – which I for one would agree with, and think it worth starting a new strand in “A/V hardware”. Potentially this camera may indeed justify the “Hi-Res Revolution” banner headline in a way the PD-1 never should have. The same link contains rumour about a forthcoming Canon offering.

Before finally leaving the subject of the PD-1, regarding it's lens coverage at w/a:

But surely it already DOES interpolate, from 1280x659 to 720x576 in Hi-Res mode, which is what most of this topic is about!! So why not interpolate the whole area of the chip in 4x3 DV mode? For lens angle reasons alone, let alone quality. The more I hear and think about this camera, the less impressed I become.

Sony will not supply cameras at 720P, it's company policy. They are launching palm-corder HD at CeBit and NAB. Yiu can safely assume that all HD camera will be genui***6:9, there are no 4:3 HD standards at all.

And the PD1 doesnm't seem to interpolate unless it has to, that's why the apparent lens angle changes with image format (hint: we've already talked several times about this, both in CV and here). It's a cost-saving trick.

Thanks Alan, I didn't know that, interesting.

I understand that as regards the recordings made - but the JVC HD-1 in the States makes 16:9 HDV recordings sourced off a single 4:3 high resolution chip, using only the centre portion. My speculation was intended to be if the Sony camera would have true 16:9 sensor chips.

It doesn't really matter what shape the sensor is, what's important is whether the camera reads pixels neatly or has to interpolate to produce the image. If a camera is to be swithcable between 4:3 and 16:9 formats, something has to give. In the current crop of high-street cameras, most use 4:3 sensors but use only 3/4 of the lines to produce wide-screen and have to interpolate the 432 lines to get back to 576. That's to keep the lens angles the same. The "HD" cameras tend to use genuine 16:9 sensors and use only the centre 3/4 of width to get the 4:3 image, thereby changing the lens angle but not needing nasty interpolation (interpolation can be perfect if done properly but it needs big filters and costs money, I can go into details if you want).

One alternative is to shoot everything using all the 16:9 sensor, and to produce the 4:3 or SD output using a down-converter on the output signal. It works very well, the Sony HDW750 does this, as does the 900 with the Miranda back, but we're now talking of cameras costing £40k and well upwards.

Another alternative is to produce SD and HD simultaneously in the camera, using seoparate processing channels, the Thomson (ex. BTS/Philips) LDK6000 does this extremely well (but has other problems), but we're now talking of £70 and upwards for studio kit.

Don't worry about how much of the sensor's used, it's what the pictures look like that matters, the manufacturer's job is to get that right by whatever means best suit the product.

Took me a while to work that one out, but I assume you mean produce an SD 4:3 signal from the HD output, as opposed to the raw output from the CCD? As far as lens angles go, it presumably makes no difference - the 4:3 output still represents the middle 75% of the chip?

Well, yes, I agree in principle, but "what the pictures look like", can also refer to lens coverage issues. No good having sharp images if to get an acceptable wide angle of view you then need a w/a convertor... softening the image nicely!!

I've just read Tom Hardwick's review of the Canon MVX3i in May's CV, and was amused to see that it has MORE pixels than the "Hi-Res" PD-1!! Looking at Tom's graph paper diagram, unlike the latter this much cheaper camera does seem able to interpolate a large number of CCD pixels down to DV resolution, and presumably avoid the lens angle compromises made by JVC in the PD-1.

Tom - if you are reading this, I liked the MVX3i review and think you did a very good job. I believe it would have been genuinely useful if I was in the market for that camera. There's a lot I could say, but will keep it to two things.

As you may gather, I like your graph paper tests, and see you gave 35mm still camera equivalent values for angle of view. Excellant! Personally, I'd give them much more prominence rather than tucking them away in the specs panel.

Secondly, I appreciate your mentioning it's low light performance quantitatively ("lags the VX2000 by two and a half stops"). Which perhaps helps prove something I've said in the past - 3 CCD's aren't necessary for resolution or colour fidelity, but do help with sensitivity, all else being equal. Add together single chip, high resolution, and small chip size, and it's no surprise sensitivity suffers, even it's fast lens can't help that much!

I don't know how you'd do it with all the variables, but in tests like this it would be good to get a meaningful comparitive figure in lux. ("0 lux"!!!??) I'm glad you added "claimed"!

You can certainly get pictures with zero lux illumination, but it's a trick. Silicon is sensitive to IR as well as visible light, so if your scene is warm but not lit, a normal camera will make a picture from it just from the heat. However, decent cameras have IR filters to prevent it because IR from a visible scene pollutes the red channel giving odd colorimetry. You can check for IR sensitivity by pointing the remote control from you tv into the lens and pressing a button on it, if the camera shows the led flashing, it's responding to IR that you can't see. The camera shouldn't see anything that you can't, so IR will pollute red and make for odd colour response.

:confused:Hello all: this is from the land of confusion. The JVC GR-PD1 Hi-Res 1280 x 659 (25p) mode is converted and recorded at 720 x 576 (25p) in MPEG-2-DV-Format (PAL). Is there anybody out there who can tell me why an MPEG-2-encoder is unsed for this standard definition DV? I wonder if it is 4:2:2 or something like that?

Seltsam

As far as I'm aware, the camera uses standard DV encoding for 720x576 interlaced at 50 fields/sec, just like any other DV camcorder would. It's only when it's shooting "HiRes" pictures that it uses MPEG.

I'll just ammend that, standard DV recording is 4:3 720x576 at 50i. That's what JVC's UK website says anyway.

Does anyone who has one of these beasts have anything to say about it, or have we put them all off?

Pretty sure it's 4:2:0 - and in answer to the first question.... errrr, yes, good question! I can only think it's an offshoot of the American true HD camera, and a LOT has got lost on the way across the Atlantic! It does have a 50p mode I seem to remember, for which the HDV format is necessary, but....

And, yes Alan, it would indeed be good to hear from an owner about the reality of some of the points raised hypothetically here.

You said about the zero lux rating "You can certainly get pictures with zero lux illumination, but it's a trick. Silicon is sensitive to IR as well as visible light.....". A cynical person may say "way of disguising bad figures on spec sheets" instead of "trick" ....

The effect is not new to silicon either. In my engineering days I remember working on location drama with tube cameras, and the arrival of a set of ND filters. The addition of a filter and reracking would cause a huge unexpected colour change - only after quite a lot of head scratching we got transmission graphs for the filters to find uniform absorption across the visible spectrum, but perfect IR tranmission! Hence adding a filter led to a large relative increase in IR, and that reflected by foilage caused grass to go quite red.

I wondered if it would be a way of getting better "day for night" results. At the time the camera CCU's derived Y,R,G,B outputs from the three tubes and fed these to separate coders. My idea was to use an ND to enhance the IR sensitivity and feed the R o/p to the Y input of the coder, additionally underexposing and desaturating. Hopefully it would mimic the most succesful day for night done in monochrome days with underexposed IR film stock. Unfortunately the cameras got replaced before I could do a trial - and the new cameras had IR absorption filters!

HDV is most definitely 4:2:0. If you want to see the spec of the HDV family, just enter "HDV" in Google, there's a site run by (I think) Sharp that gives the full table of spec data.

It's plain that the PD1 is a reworking of the HD10 for the 50Hz market, and I reckon they aimed it at the Ozzie market because they've got 50Hz HDTV and so can show the 50p footage. I still don't understand why JVC are marketing it in Europe. Nor can some of the JVC people I know (but more of that after I've retired, not long now).

Hello all,

I looked into the URL >>[url]http://sharp-world.com/corporate/news/030704.html<<[/url]
in search for a HDV-MPEG2 spec with 576 lines at 25p and still dont have any good clue what the Hi-Res mode of the JVC GR-PD1 is recording on the tape. :confused:

Seltsam

Go to www.jvc.co.uk and search for the PD1, it'll tell you all that JVC UK knows about it.

Hi All,

My first post...

Sorry to dig up an old thread, but being the owner of a GR-PD1 I was also supprised that the PD1 does not actiually record in HD (I know...I should have done more research, but the specs / details are confusing). The North American HD1 seems as though has the same camera housing, optics and electronics, so I beleive the only difference would be firmware. I have written to JVC Australia regarding firmware update to support HD recording to tape. I will let you know as soon as I get a response..

Regards

Alex

The HD version is the HD10, which superficially looks the same, but isn't inside. I wouldn't hold out any hope of firmware updates, or even of any meaningful response, if I were you.