Is this technically correct or just marketing hype?

[Ford Prefect]

Hi Folks,
A colleague recently purchased a new cam which was (and still is) described as "Full HD 1080i".
He was mortified to find out (worse still, he hadn't noticed until someone else pointed it out to him) that the sensors are actually 1440x1080 and not 1920x1080 as he expected.
The cam was a Sony HDR-FX1. He has contacted sony and they told him it's full HD.
The shop offered him an upgrade to the "latest" FX7", but I have told him to be careful as although it's very easy to find the Sony specs, it's very difficult to find the true resolution of the sensors and I believe the FX7 is also 1440x1080 but has CMOS sensors. I believe Sony marketing are making it confusing on purpose.
I wasn't confused personally as both cams have model numbers which start with "HDR", a bit of a giveaway, i.e. HD ready.
I was recently looking at some high-end professional JVC stuff, the GY series, can't remember the model which was also 1440x1080.
As a photographer and videographer, I always take into account the size of the chips as well as the resolution but is there an agreed definition of what really constitutes HD?
Hence my question in the title.
Regards.

[Alan Roberts]

Right, here's the straight stuff.

HDTV, as broadcast standards defined in ITU and SMPTE standards, is either 1920x1080 or 1280x720. However, the broadcasters may not be doing that. Since much of the recorded HD has come through HDCAM or DVCPROHD, the image dimensions will be 1440x1080 (the image dimensions of the recordings), so there's little point in broadcasting 1920. As far as I'm aware, Sky HD is 1920x1080 and BBC HD is 1440x1080 (although I expect that may change eventually). Broadcasting 1920 needs either HD-D5, HDCAM-SR or XDCAM HD at 50M b/s for capture (or JPEG2000), all else is 1440 or less.

"HD Ready", on a tv set, means only that it will accept a broadcast HDTV signal, and make pictures/sound from it. It does not define the screen resolution, except that the manufacturers are regarding "HD Ready" as at least 720 lines. Assuming the screen pixels are square, that means 1280x720 and bigger is "HD Ready".

"Full HD" on a tv set, means that the display panel has 1920x1080 screen pixels. However, it doesn't mean that you'll get pixel-mapping from the camera to the screen, since internal processing in the display may be at lower resolution. I've been throwing test signals at displays recently, and have found some intriguing results that I don't fully understand yet. But more of that anon.

So, display manufacturers regard "Full HD" as 1920x1080. And so I'd expect that camera manufacturers would do the same, particularly since they're mostly the same companies. But, you have to take into account the resolution of the recording systems. HDV is 1440x1080, not 1920x1080, so a camera with 1920x1080 sensors will probably make the best job that can be done within the format, going for 1920x1080 would make pictures that are a bit better, but only if the camera's scaling filters are good, and I doubt that can be done at the FX1 price. And the situation in single sensor cameras is much less simple.

So, while the FX1 is not technically Full HD, it's as good as it needs to be for HDV recording.

[foxvideo]

Alan, that's the first explanation of HD/HDV I've actually understood - many thanks ;)

[infocus]

He was mortified to find out ....... that the sensors are actually 1440x1080 and not 1920x1080 as he expected.
Firstly, 100% agreement with all Alan says, with one addition - the EX cameras are XDCAM at 35Mbs, but do record a true 1920x1080 raster, and have three chips of that resolution.

The 1440x1080 sensors aren't necessarily the thing your friend should be worried about anyway. Many cameras employ what's called pixel-shift techniques to get more resolution out of the system than the native sensor resolutions would suggest. Figures vary as to how effective it is, but an improvement of about 1.25x is in the right ball park. Hence (if used) pixel shift could yield resolution here of up to about 1800x1080. But even if it was used, HDV recording resolution is (as Alan says) 1440x1080, so ........

The situation seems a lot clearer with display panels as regards "Full HD", which is generally now taken to mean a panel with 1920x1080 R,G and B pixels. With cameras the situation is a lot more woolly, though I suspect if you tried to take anybody to court they'd argue that their definition of "Full HD" was a 1080 system rather than 720.

[steve]

Actually, I think the FX1 has 3 CCD sensors each with 960x1080 cells. The green sensor is shifted 1/2 a cell laterally, as infocus suggests, giving approximately 1.5x luminance resolution, which matches the HDV standard of 1440 horizontal samples per line fairly well. There's little point having 1920 cells per line if it will be sub nyquist sampled with all the aliasing problems that Alan R is talking about. The FX1 also has what some regard as a slightly soft lens, particularly when compared with the single chip HC1/A1 HDV camera. Alan R who has an A1 has frequently mentioned that the lens is a little too sharp, and the camera tends to produce artifacts that end up as picture disturbances even on well lit scenes.
Of course the consumer camera manufacturers know well how number obsessed their lay customers are so are happy to ply the market with products that quote numbers to feed this frenzy. just look at the ridiculous resolutions being quoted for pocket digital still cameras and mobile phones that are fitted with cheap plastic lens.
The FX1/Z1 is a very well engineered compromise that hit the market nearly four years ago and got the best out of the HDV standard at a prosumer price.

Steve

[Don Clay]

Thanks Alan!

Don Clay

[infocus]

Actually, I think the FX1 has 3 CCD sensors each with 960x1080 cells. The green sensor is shifted 1/2 a cell laterally, as infocus suggests, giving approximately 1.5x luminance resolution, which matches the HDV standard of 1440 horizontal samples per line fairly well.
You're quite right - 960x1080 per sensor. I think the 1.5x figure is generally regarded as high, though a lot seems to be depend on whether the increase is to a useful level, or "ooh look, there might be a faint vestige of even finer detail there". Either way it matches the HDV spec reasonably well.
Of course the consumer camera manufacturers know well how number obsessed their lay customers are so are happy to ply the market with products that quote numbers to feed this frenzy.
Yes, nothing wrong with most of the hardware in value for money terms, and the same here, but it doesn't excuse misleading advertising statements. I suppose it revolves around what "Full HD" actually defines, and I suspect the claim here would be 1080 as opposed to 720. but.......

[JGNattrass]

Thanks to Alan for all the tech stuff but how do I fit in with shooting on a Z7 at 1440x1080i but then my final export files are Apple pro res HQ 4:2:2 at 1920x1080i?

I hear a lot about anti HDV for broadcast up here in north north east but DVcam at 25mbs to Digi beta is perfectly acceptable.

surely the same applies to taking 25mbs HDV and then outputting at pro res 1920x1080i

Also my HDV M35 deck has an HD/SD-SDI output so what is coming out of that to XDCAM/Digibeta decks for TX masters? Surely the same as going Dvcam to Digibeta or better resolution.

As an audio guy I find the HD formats thing full of bullshit as for years 44.1 or 48k 16 bit was the bench mark for audio but these days anything goes inc mp3 and very compressed audio source. Surely the same will apply to the pictures in time due to the diminishing budgets available for production.

I just did this video today as a test for macro shooting with the Z7 and nikkor lens but even though this is compressed to oblivion the master was Z7 to pro res 4:2:2 1920x1080 from 1440x1080:http://www.vimeo.com/1672101

It just doesnt add up to knock 1440x1080 as non broadcast when progs like you been framed can take a VHS from 20 years ago and make it sync and broadcast safe just by poping it onto digi beta, yes the quality is grainy and crap but to dismiss the greatest opportunity for change in programme shooting can not be left to pure tech specs and perceived HD pixel and data rate snobbery.

I see the same happening in audio as sample rate and mic pre-amp one upmanship is more important than actually recording the music, people on audio forums like this get bogged down in what mic/what pre amp/what sample rate is best and they forget that content is king. Do we ask what mic the classic recordings were made on or what video format or camera classic TV shows were produced on?

Is 2 inch ampex better than 1 inch sony better than Betacam sp better than digibeta or is ariiflex better than aaton at 16mm or super 16 or 35mm, kodak/fuji/reversal/negative/ ampex/3m/racal/basf....???????????? DOH@! what was the storyline again?????

Is it any wonder we have 1000 channels with nothing on???? Make some watchable content for christs sake!!!!!

[StevenBagley]

It just doesnt add up to knock 1440x1080 as non broadcast when progs like you been framed can take a VHS from 20 years ago and make it sync and broadcast safe just by poping it onto digi beta, yes the quality is grainy and crap but to dismiss the greatest opportunity for change in programme shooting can not be left to pure tech specs and perceived HD pixel and data rate snobbery.

No one is knocking HDV for its 1440x1080 pixel count, after all both HDCAM and DVCproHD run at the same pixel count and are regularly used in broadcast. The problem with HDV is purely the compression is pushed too far and that the pictures lain down on tape are flakey at best. They may look ok at the first-gen (although with HDV it is touch and go), but by the time they've been through the number of codecs used in a modern broadcast environment (which could easily be 4 or 5: MPEG2 on HDV tape, editing codec, codec on final master tape, high-bitrate MPEG2 on broadcast server, H264 on transmission channel) the images look nasty. This isn't solely a problem with HDV -- the same thing was happening with the coverage of the Olympics, the concatenation of codecs was beginning to show by the time the HD pictures reached the home (and things were even worse in the US -- where the picture just broke up into a million scintillating blocks a lot of the time).

Now in the case of the Olympics a lot of it was out of the BBC's hands (they got the pictures fed from the IBC in Beijing after they had been fed back over MPEG2 links) and so they had to run with those images but in the case of someone shooting something new then there is the opportunity to mandate that they use equipment that will produce the best possible quality. That isn't snobbery -- it's called ensuring the viewer gets the HD experience, and so HDV, along with Super16 film is outlawed. This isn't the problem for the producer either -- if his programme is comissioned then it will be allocated a budget that allows him to hire the correct kit to do the job.

Of course, there are lots of situations where HDV is fine as a medium, for producing SD shows (where the compression artefacts will be filtered out in the downconversion) or for non-broadcast HD work.

You've been framed is not a good example -- its only reason for existence is to showcase home videos, to use that as an argument would mean we could through out any notion of framing shots, direction etc. IT would have been commissioned on the basis that it was going to feature lots of below broadcast quality sources.

but DVcam at 25mbs to Digi beta is perfectly acceptable.

surely the same applies to taking 25mbs HDV and then outputting at pro res 1920x1080i

Also my HDV M35 deck has an HD/SD-SDI output so what is coming out of that to XDCAM/Digibeta decks for TX masters? Surely the same as going Dvcam to Digibeta or better resolution.

No, because the DV codec works in a very different way to the MPEG2 codec used on HDV. For starters, DV is only compressing 158mbit/s into 25mbit/s rather than 1.5Gbit/s and so there is significantly less data thrown away.

Is 2 inch ampex better than 1 inch sony better than Betacam sp better than digibeta

2" is better than 1" no doubt about it and stands up to multi-generational use far better and in terms of quality wasn't matched until D3 and not surpassed until digital component formats like D1, D5 and Digibeta came along...

Is it any wonder we have 1000 channels with nothing on???? Make some watchable content for christs sake!!!!!

Yes and do it with the proper tools! And isn't it telling that the channels that take the time to get the toolchain correct, tend to be the ones with things worth watching on. Do you really believe that Planet Earth would be as breathtaking and interesting if it was shot on a Z1???

Steven

[John Disdle]

Just to be different, my Toshiba 37WLT66 has a resolution of 1366x768???
I can`t as yet view in HD, although it is HD ready.

Several years ago in the early days of Freeview, I bought a new TV and queried as to why it was not fitted with a digital tuner to receive Freeview. The answer was that I could buy a box that would do this, as was advertised all round the shop.

It`s the same with the Toshiba. It says HD Ready, and yet I have to buy a box to get it. But it does have a Freeview tuner.
So! In a few years time I expect to be able to buy a TV that is fitted with a HD tuner, or am I expecting too much?

[JGNattrass]

Thanks for your comment Steve they are much appreciated, I take your points but I just get frustrated that mainstream TV content is dreadful and I am sure that a lot of this is down to the high budgets that are imposed on producers.

Surely once a format is transfered to a more robust medium it will stand up better to the compression or is that not the case? That is why I used you been framed as an example that is very low quality pictures transfered to digibeta and then put through the tx mill and I dont see any technical break up of frame scan etc. YES the subjective picture quality is still crap but the pictures dont break up do they?

Agreed in a perfect world everything would be shot full high quality at he best bit rates available but at the end of the day it all comes down to cost. In these times where people like ITV are wanting to just buy in all there content and take the advertising revenue pie surely making new programming on the most accessable formats that have ever been available is far better? Using HDV as a shooting medium and then doing post up to the full rates available to make a master that is robust for delivery.

It not as if this kind of thing is new, mores used to be blown up from 16 to 35mm for international delivery and the blair witch project was shot on domestic video and then blown up to 35mm for delivery.

[StevenBagley]

Surely once a format is transfered to a more robust medium it will stand up better to the compression or is that not the case?

No, this isn't an analogue world anymore. Once something has been digitally compressed then the artefacts are locked in for life. If you transfer it to another format then the problems are still there -- so the HD-SDI out on the M35 has all the artefacts locked in by the HDV compressor, even though it is an uncompressed signal. If you then try and recompress that 'uncompressed' signal, the artefacts from the HDV compressor that are locked in to the images (although possibly unnoticed to the human eye at this point) cause the compressor to have to waste bits encoding the artefacts and so the picture as a whole suffers. Do this through 5 or 6 different encoders and by the time the viewers get the picture the result is a mess and the show is no longer really an HD experience (a good example of this, even though little of it is shot on HDV would be 'Britain from Above' -- which didn't look that nice in HD).

To put it in audio terms, it's like trying to produce Dolby Digital audio for a DVD (at 192kbps) by recording everything initially at 128kbps MP3. Yes, the source MP3 sounds ok (but not great), but once its been recompressed as a 192kbps AC3 signal, it sounds worse. The concatenation of the MP3 compressor and the AC3 compressor produces a worse signal than if the AC3 compressor was fed with raw PCM data. Converting the audio to uncompressed (which has to be done to recompress it anyway) over SPDIF (the audio equivalent to HD-SDI) makes no difference, the artefacts are still in the audio signal and still confuse later compressors. Then when you consider that HDV is more like a 32kbps MP3, you can see why it isn't accepted for HD acquisition.

Steven

[infocus]

I just get frustrated that mainstream TV content is dreadful and I am sure that a lot of this is down to the high budgets that are imposed on producers.

In these times where people like ITV are wanting to just buy in all there content and take the advertising revenue pie surely making new programming on the most accessable formats that have ever been available is far better? Using HDV as a shooting medium and then doing post up to the full rates available to make a master that is robust for delivery.

But if it imposes high budgets on producers, it imposes the necessity to pay more from the broadcasters. Surely if cheaper production was generally allowed (such as HDV), it would just lead to such as ITV paying less, so back to square one, with poorer quality? And I don't think HDV is banned, period, it's allowed when a good case can be made for it. "It's cheaper for me" is not considered a good case.

And if you think mainstream TV content is dreadful, I'd argue that (generally) on the smaller channels it's much worse.
It not as if this kind of thing is new, mores used to be blown up from 16 to 35mm for international delivery and the blair witch project was shot on domestic video and then blown up to 35mm for delivery.
I don't think that's a very good example ;) I think of that as the most disappointing film I've ever paid good money at the cinema to see (and dragged my wife along to.....). But a good example of what marketing hype can do. Main objections were the content though, although the (very) poor technical quality didn't help.

[JGNattrass]

All points taken and I appreciate the tech side of it now but I grew up in TV with dolby A mis tracks on audio and dreadful u-matic and betacam std so at least nowadays there are more clean formats.

Of course I suppose all this applies to current digital delivery methods as they have to compress a lot to get deliverable bandwidth, in tests for the IPTV I am researching the delivery from HDV to user has a lot less compression than terestrial tv and therefore at times a cleaner picture than off air freeview etc.

Here's hoping that latest reports of new codecs to get 50mbs down twisted pair std fone lines holds true, they are also quoting 50mbs up and downstream times for wimax now with 5km coverage so it will be interesting to see how delivery platforms for HD develop.

[Alan Roberts]

Bit of a follow-up.....

Yes, the FX1/Z1 has sensors of 960x1080, and has horizontal "precision-offset" whereby the G image is half a pixel offset from R and B. It's widely assumed that this delivers up to 50% more horizontal resolution in the luma channel (a weighted sum of gamma corrected R' G' and B'), but it comes with coloured aliasing and can't be suppressed even with a decent bi-refringent filter. The effect is similar to that from single-sensor Bayer pattern cameras (e.g. A1/HC1). The idea is to keep the pixel size as big as possible in order to keep the signal levels high (i.e. to improve the sensitivity or reduce the noise level). The problem it brings is that of the coloured aliasing, which causes confusion on the recording compressor, consuming bit-rate unnecessarily. The high end cameras are now all showing 1920x1080 sensors, and concept of precision-offset will quietly go away, because it shouldn't be needed, it's there only for economy. Notably, it's in the HDC1500 studio cameras, but not in the F23 which otherwise has the same front parts. Also, I think it's not there in the HDWF900R and 790, but can't be sure of that. At the high end, it causes problems when recording 4:4:4, because the coloured aliasing is much more visible and creates problems in colour keying.

My comments about XDCAM were aimed at cameras such as the PDW700, which has 1920x1080m sensors and records in 4:2:2 at 50Mb/s. That ticks all the boxes for the broadcasters, provided they can handle the Prodisks.

There's years of research showing that the cleaner the original picture, the lower the compressed bit-rate can be, and I've got some lovely demos of the effects. Produce aliases in the camera, and the compressor will always do a poor job.

[infocus]

Bit of a follow-up.....

Yes, the FX1/Z1 has sensors of 960x1080, and has horizontal "precision-offset" whereby the G image is half a pixel offset from R and B. It's widely assumed that this delivers up to 50% more horizontal resolution in the luma channel ............
I previously wrote:
I think the 1.5x figure is generally regarded as high, though a lot seems to be depend on whether the increase is to a useful level, ........
It's been suggested to me that I'm confusing the use of horizontal AND VERTICAL pixel-shift techniques (such as the Panasonic HVX200 uses) with horizontal only (such as used by the FX1), and that whilst 1.5x is a reasonable approximation for the latter, my figure of around 1.25x (horizontally and vertically) may be more applicable to the former.

The more I think about it, the more I'm inclined to regard it as true, though with a lot of "approximatelys" in! :) I'd be interested to hear your opinion, Alan.

[Alan Roberts]

Yes, sounds right to me. With horizontal offset, you get more luma resolution and some horizontal colour aliasing. Ditto for vertical. Do both, and you get less benefit in each direction, and less coloured aliasing in each direction. 25% for both sounds a reasonable estimate to me, although it's hard to be precise because we're trying to balance the benefits of resolution against the problems of aliasing.