NEW more detailed info on the DVX200 (4/24/2015)

NEEL,

Just to clarify, do you need the 10bit 4:2:2 in 4k or HD?

Frankly speaking, this is for HD and I don't know for 4K yet. But taking a look at this document (see 1.2.2) I'd say it'd be more than likely that 422 and probably 10Bit will be recommended soon for any UHD acquisition.

https://tech.ebu.ch/docs/r/r118.pdf

As far as I understand the "roadmap", the EBU decision will be taken later this year. With the FS7 and the onboard 10 bit it feels more save at the moment though ;-)

Edit: The chapter with the recommendations for bit depth etc in the EBU document is 1.2.2, not 3.14 as I previously wrote - sorry for this.
 
Last edited:
Frankly speaking, this is for HD and I don't know for 4K yet. But taking a look at this document (see 3.1.4) I'd say it'd be more than likely that 422 and probably 10Bit will be recommended soon for any UHD acquisition.

https://tech.ebu.ch/docs/r/r118.pdf

As far as I understand the "roadmap", the EBU decision will be taken later this year. With the FS7 and the onboard 10 bit it feels more save at the moment though ;-)

If it is for HD then just shoot 4k and down convert to HD in post. This gives you HD at 4:4:4 which will rival most native HD cameras that record 4:2:2. In fact it gets you very close to the quality of a HD 3 chip camera with even higher chroma detail. HD from the DVX200 would likely look much better to shoot 4k then it would be compared to shooting HD in camera even if it did have a 4:2:2 HD codec. If you need 4k at 4:2:2 that is another story but honestly from a 4k bayer sensor you don't really have 4:2:2 worth of color anyway. Just up sample the chroma in post and you have virtually what a 4:4:4 recording from the bayer sensor would have been anyway.
 
This is where I diverge from the "4k crowd". Yes, 4k downsampled is great but lets face it, a "video" person more than often will need to shoot for long periods of time or need quick turnaround. As well as dealing with file sizes, backups and possible multicamera shoots. ALL of this nullifies the shoot in 4k and downsample image benefits imho.

All they need to do is add a 10bit 4:2:2 HD recording option and all of the "pro" bases will be covered.
 
There's another little elephant in the room: The 5 axis stabilization - an utterly utile tool for run and gun - apparently will only work with HD, not UHD. So, for 10 bit 4:2:2 I not just have to downconvert my 4K footage, I might also be obliged to use stabilization in post for similar results like shooting directly in HD. The render times would be monstrous. For any serious broadcast work with this camera, an internal 4:2:2 codec in HD is a must.
 
There's another little elephant in the room: The 5 axis stabilization - an utterly utile tool for run and gun - apparently will only work with HD, not UHD. So, for 10 bit 4:2:2 I not just have to downconvert my 4K footage, I might also be obliged to use stabilization in post for similar results like shooting directly in HD. The render times would be monstrous. For any serious broadcast work with this camera, an internal 4:2:2 codec in HD is a must.

Could you define what, exactly, is "serious" broadcast work? I'm not trying to pick a fight or be snarky, I'm genuinely curious. When I left my previous job at a major market network affiliate 2.5 years ago, all the news was shot on DVCPro25 and uprerezed to 1080i for broadcast. DVCPro25 is 4:1:1. Most of the local commercials we shot were shot in AVCHD on an AF100 or GH2... 4:2:0. I helped with an interview for a big PBS doc that was also shot on a GH2. One competing station was shooting on EX3's (4:2:0), and the other two news stations were still on Betacam.

I get for certain broadcasters like Discorvery HD have very specific requirements, but I'm curious if this is something all European broadcasters require? Because I've seen plenty of serious broadcast work shot on something less than 4:2:2 and no one cared. All that mattered was whether the footage looked good to the eye.
 
Could you define what, exactly, is "serious" broadcast work? I'm not trying to pick a fight or be snarky, I'm genuinely curious. When I left my previous job at a major market network affiliate 2.5 years ago, all the news was shot on DVCPro25 and uprerezed to 1080i for broadcast. DVCPro25 is 4:1:1. Most of the local commercials we shot were shot in AVCHD on an AF100 or GH2... 4:2:0. I helped with an interview for a big PBS doc that was also shot on a GH2. One competing station was shooting on EX3's (4:2:0), and the other two news stations were still on Betacam.

I get for certain broadcasters like Discorvery HD have very specific requirements, but I'm curious if this is something all European broadcasters require? Because I've seen plenty of serious broadcast work shot on something less than 4:2:2 and no one cared. All that mattered was whether the footage looked good to the eye.

Yeah, here in the US, I'm not sure about this 'broadcast requirement'. I've done
a ton of broadcast work in less than '50mbs 4.2.2' codec. I think this is an oft
cited requirement that comes from the BBC (and other specific channels quite possibly
have similar requirements?). Not sure how important it is here in the US for general
'broadcast work.'
 
Go to the very beginning of this thread. The Panasonic spec list says that the lens at the wide end is 28mm in FHD and 29,5 in 4K. That's where I drew my conclusions from.
Ahh, I didn't spot that. Even more interesting from that list is that 4K at 60p has an angle of view equivalent of 37.2mm.

The way I read that is that for 4K at 24p, it's using the sensor full width, and for 60p it's doing the same as the GH4 - just using the central 3840x2160 photosites of what must be a 16Mp (approx.) sensor. (The equivalent focal length ratios work out pretty well exactly correct for that.)

Maybe not the end of the world, but it does mean if you're buying it for the 4K 60p mode, it's widest angle of view is going to be pretty tight for a wide end. Same applies with the GH4 - but that can always have interchangeable lenses to get over it, a "very wide" becoming a "wide". With the DVX200 you're stuck with quite a narrow field of view in 4K 60p.

And so much for the talk of having a native 8Mp sensor! It seemed unlikely.......
 
Frankly speaking, this is for HD and I don't know for 4K yet. But taking a look at this document (see 1.2.2) I'd say it'd be more than likely that 422 and probably 10Bit will be recommended soon for any UHD acquisition.

https://tech.ebu.ch/docs/r/r118.pdf
Most interesting - it's a while since I last saw that document and there has been at least one substantial change since then. That is concerning colour space, and instead of simply requiring 4:2:2, period, it now says for general HD acquisition:
4:2:2 Interlace 4:2:0 progressive
Which is very, very sensible and something I've been trying to get across for a while. The reason 4:2:2 is traditionally preferred for acceptability is because of interlace - use 4:2:0 and it's the interlace that can cause problems with repeated coding/reencoding. Use progressive and that's no longer an issue, so the 4:2:2 requirement goes away.

And lok what it says for 4K (progressive all the time):
4:2:0 but 4:4:4 preferred
In other words, forget about 4:2:2 moving forwards. It's whole raison d'etre was in the days of interlace and moving into 4k and progressive it has little point. For most acquisition 4:2:0 makes sense - when only the best will do, it should be 4:4:4.
 
Most interesting - it's a while since I last saw that document and there has been at least one substantial change since then. That is concerning colour space, and instead of simply requiring 4:2:2, period, it now says for general HD acquisition:

Which is very, very sensible and something I've been trying to get across for a while. The reason 4:2:2 is traditionally preferred for acceptability is because of interlace - use 4:2:0 and it's the interlace that can cause problems with repeated coding/reencoding. Use progressive and that's no longer an issue, so the 4:2:2 requirement goes away.

And lok what it says for 4K (progressive all the time):

In other words, forget about 4:2:2 moving forwards. It's whole raison d'etre was in the days of interlace and moving into 4k and progressive it has little point. For most acquisition 4:2:0 makes sense - when only the best will do, it should be 4:4:4.


very interesting. I would love to hear other people's opinions on this startling, but sensible, development. Do you think the picky broadcast channels will adopt these recommendations?
 
Most interesting - it's a while since I last saw that document and there has been at least one substantial change since then. That is concerning colour space, and instead of simply requiring 4:2:2, period, it now says for general HD acquisition:

Which is very, very sensible and something I've been trying to get across for a while. The reason 4:2:2 is traditionally preferred for acceptability is because of interlace - use 4:2:0 and it's the interlace that can cause problems with repeated coding/reencoding. Use progressive and that's no longer an issue, so the 4:2:2 requirement goes away.

And lok what it says for 4K (progressive all the time):

In other words, forget about 4:2:2 moving forwards. It's whole raison d'etre was in the days of interlace and moving into 4k and progressive it has little point. For most acquisition 4:2:0 makes sense - when only the best will do, it should be 4:4:4.


can you please link to the new proposal?
 
Could you define what, exactly, is "serious" broadcast work? I'm not trying to pick a fight or be snarky, I'm genuinely curious.

I get for certain broadcasters like Discorvery HD have very specific requirements, but I'm curious if this is something all European broadcasters require? Because I've seen plenty of serious broadcast work shot on something less than 4:2:2 and no one cared. All that mattered was whether the footage looked good to the eye.
Two things.

First is that the EBU guidelines are just what the words says - guidelines. They aren't rules, let alone laws - they're guidelines. The aim is to try to define a form of "best practice" for all concerned. To try to ensure that standards are upheld as much as possible, without setting impossible to reach targets. The idea is that everyone knows where they are and a contributor knows in advance what will and will not be considered acceptable for various types of programming. Hence the tiering of cameras. What is considered best practice for drama is certainly not insisted upon for news!

And being "guidelines" they are flexible such that if a case is made for using a sub-standard camera for very good reasons (eg nature of work means a very small camera must be used) that may well be granted. But using a small camera just because it's cheap (so more profit to the film-maker!) will not be seen as a valid reason. :) Same applies to using poor quality archive or amateur footage. If it's necessary to the programme because of content it's one matter - it doesn't mean a low quality camera will therefore be seen as acceptable for purpose shot footage.

Secondly, whether "footage looked good to the eye" is only part of the story. A broadcast chain may involve repeated recoding of the material, more so when archive starts being used and reused. And it's been found that some material can look fine straight out of the camera - visually little different to from a much higher specced camera - but fall apart as it goes through the broadcast chain. How big a problem in practice depends on the total chain - the idea is to set a standard that should cope reasonably well with all but the worst cases. And this is why aliasing is considered such a big issue in the broadcast world. A lot of aliasing may not be very objectionable in camera footage - but can really cause problems in compressors further down the chain.
 
can you please link to the new proposal?
It's the link that NEEL posted earlier ( https://tech.ebu.ch/docs/r/r118.pdf ), and the same section he referred to: para 1.2.2 - the quotes I gave came directly from there. Which are highly sensible - honest! Such as 4:2:2 and 4:1:1 really have their relevance in interlace systems, and certainly in Europe most countries went for 1080i/25 when HD first started. Initially, progressive material got transmitted psf hence still the 4:2:2 requirement for full acceptance, but now I think the transmissions chains handle true 1080p/25 - so far less need for 4:2:2 acquisition for such.
very interesting. I would love to hear other people's opinions on this startling, but sensible, development. Do you think the picky broadcast channels will adopt these recommendations?
Since R118 is formulated by the broadcasters, and correlated by the EBU as "best current practice", that would seem highly likely....... :)
 
Secondly, whether "footage looked good to the eye" is only part of the story. A broadcast chain may involve repeated recoding of the material, more so when archive starts being used and reused. And it's been found that some material can look fine straight out of the camera - visually little different to from a much higher specced camera - but fall apart as it goes through the broadcast chain. How big a problem in practice depends on the total chain - the idea is to set a standard that should cope reasonably well with all but the worst cases. And this is why aliasing is considered such a big issue in the broadcast world. A lot of aliasing may not be very objectionable in camera footage - but can really cause problems in compressors further down the chain.

Totally understand and agree... which is, in part, where my question came from. All of the commercial stuff shot in house at my last TV station at less than 4:2:2 was usually ingested straight into master control's servers after the edit was finished. News was played back from the news servers. But local news and last week's local Kia car lot commercial in the US, or my friend's call-in basketball TV show in Uganda aren't going to have the same archive requirements as a BBC drama.

I guess my concern is when a statement like "serious broadcast work" is said, that everyone is on the same page as to what that means. My definition of "serious" might be a lot different than someone else. I would have no problem shooting a Kia car lot commercial on a DVX200's codecs. :)
 
I notice that the standard calls for H264 200mbs UHD 0.5s GoP and 400mbs UHD intra all. So the question is what is DVX200 top mbs in 4k, 150 or 200mbs?
 
I notice that the standard calls for H264 200mbs UHD 0.5s GoP and 400mbs UHD intra all. So the question is what is DVX200 top mbs in 4k, 150 or 200mbs?
Isn't DVX200 offering the same bitrate as GH4? Namely, 100 Mbps in 4K 24p (and 200 Mbps in 1080 24p)?
 
....Maybe not the end of the world, but it does mean if you're buying it for the 4K 60p mode, it's widest angle of view is going to be pretty tight for a wide end. ..

And so much for the talk of having a native 8Mp sensor! It seemed unlikely.......

My only reason for getting a DVX200 would be for 4K 60P that has better performance than my FDR-AX1 so is looking less attractive !!!
 
I guess my concern is when a statement like "serious broadcast work" is said, that everyone is on the same page as to what that means. My definition of "serious" might be a lot different than someone else. I would have no problem shooting a Kia car lot commercial on a DVX200's codecs. :)

OK , "serious" sounds a bit kooky in this context. But there are some projects you might run directly into a bureaucratic brick wall if you don't check the codec requirements before - even if your 4:2:0 footage looks perfect by any criteria. The above linked specs book woke up my doubts even more as I was convinced 4:2:2 was required for HD acquisition for mayor TV channels here in Europe. Now I'm going to check this with someone who really is in the knowing and will forward the result.
 
EXACTLY, if true or does not change before release UDH would not meet spec.
The bitrate is only one aspect of the EBU approval spec, and it's worth looking at what's required for a camera to be fully approved for 4K (3840x2160) acquisition.
2.1 UHD1 Tier 1
These cameras offer the full UHD1 3840 x 2160 resolution. They are usually single sensor cameras with an image size approximately the same as that of 35 mm film. These cameras have a shallow depth of field and are primarily used for Fiction (Drama) productions.
• minimum sensor size 1",
• minimum sensor pixel count is a minimum of 3840 x 2160 in each of R & G & B,4
• minimum 10-bit processing but 12 (or higher) preferred,
• 4:4:4 capable,
• interchangeable lenses with a quality commensurate with the cameras use,
• timecode gen-lock.
On-board codec can either be a full resolution RAW format or a UHD H264 based codec of 200 Mbit/s or better.

2.2 UHD1 Tier 2
These cameras have a UHD output but a sensor that does not offer the full UHD 3840 x 2160 resolution. They are usually single sensor cameras with an image size approximately the same as the 35 mm film image size. These cameras have a shallow depth of field and are primarily used for Fiction (Drama) productions.
• minimum sensor size 1",
• sensor pixel count must be greater than 2880 x 1620
• 10-bit processing but higher preferred,
• 4:2:0 but 4:4:4 preferred,
• interchangeable lenses with a quality commensurate with the cameras use,
• timecode gen-lock.
On-board codec can either be a full resolution RAW format or a UHD H264 based codec of 200 Mbit/s or better.

The point is that these are broadcasters specs, and as such their interest in 4K *at the moment* is solely at the very highest end. Hence the tiering (currently) only applies to top end cameras - there is no way any (4K) camera of the type of the DVX200 will be accepted currently for *4K* acquisition for many reasons other than bitrate. For HD acquisition, then probably yes, at least at some tiering level.

For 4K, even an Alexa will only be considered Tier 2, by virtue of it's less than 3840x2160 sensor.

But the most interesting thing is how "4:2:2" has now started to disappear from the requirements as progressive becomes the order of the day with 4K. Entry level becomes 4:2:0, higher level becomes 4:4:4. It was interlace that demanded 4:2:2, and 4:2:2 makes much less sense in the all progressive world of 4K.
joe1008 said:
I was convinced 4:2:2 was required for HD acquisition for mayor TV channels here in Europe. Now I'm going to check this with someone who really is in the knowing and will forward the result.
In it's simplest form, and for an interlaced HD format (such as 1080i/25), the requirement for 4:2:2 does hold good...... But it's never been quite as black and white as that. Firstly, 4:2:2 is only one requirement, and I think what you're referring to is what the EBU refers to as "general unrestricted HD acquisition". The exact definitions are fluid, but that doesn't mean *ALL* material in any single programme has to be such. There is generally a percentage of "non-approved" material allowed within any programme without the need for special clearance - and "non-approved" covers archive, upconverted SD, and lower tier HD cameras (such as 720 native, minicams etc). It is also possible to exceed this percentage by a significant amount - provided advance approval is obtained. Examples where such would be likely to be granted would be where large amounts of archive are necessary, or there is a pressing case for such as small cameras to be extensively used for genuine practical reasons.

Note that it isn't - never has been - an all encompassing rule, and hence the whole idea of tiering rather than just stamping any camera "approved" or "non-approved". Hence a camera may be considered fully approved for (say) news, but not for (say) drama. So a 4:2:0 interlace camera may be considered fully acceptable for news, but not for general programming.

And finally, the minimum 4:2:2 requirement has been relaxed to 4:2:0 for all progressive material, though is still considered in force for interlace. And I confess I didn't know that had happened until relooking at R118 yesterday.
 
Back
Top