Request 4K 10-bit 4:2:2 internal for the DVX200!

[combatentropy]

Jaime Valles's request is entirely reasonable in 2015.

Chroma subsampling was needed 10 or 20 years ago, because compact media had such a narrow throat. Even DVCAM and DVCPRO were 4:1:1.

Today's commodity media can handle it. Some SD cards guarantee they can write 30 megabytes per second, which is 240 megabits per second. Furthermore some actually write twice that or more, 60-80 megabytes per second (480-640 megabits per second). Even those still cost less than $2 per gigabyte.

For comparison, ProRes 422 HQ 1080p 24fps is 22 megabytes per second. I'm assuming 4K would be 4x that, at 88 megabytes per second. So just out of reach of the best SDXC cards. But just barely, and this is ProRes, a mild, intraframe-only compression. If you start talking H.264 or H.265, you can keep the data rate of even 4K 10-bit 4:2:2 well within the limits of high-end SDXC cards.

That's today. In a year, it will be even sillier, as SD cards probably double in speed.

 

[Mike Harvey]

And some folks would while others wouldn't. DVX200 is a run&gun camera, where presumably footage isn't going to be graded heavily and, with a straight-out file, 8 bit is sufficient. Raise the price by $500 and suddenly your market niche goes that-a-way.

This camera seems less run & gun and more jack-of-all-trades. But, I would think run & gun scenerios would be even more reason to include it. I almost always have to do some sort of color correction because I'm often shooting where I get partial control of lighting at best, and the camera has a hard time white balancing the mixed lighting. Plus, the overwhelming number of wedding videographers that I see (who would make up a huge market for this camera) do some sort of color grading. Vlog + 10bit 4:2:2 could be a huge selling point for them.

And is the difference between, say, $4,700 and $5,000 really going to be a big sticking point between the for customer base for this camera?

 

[David Jimerson]

I agree; I don't get at all where it can be labeled (dismissed as?) a "run & gun" camera.

 

[Barry_Green]

Chroma subsampling was needed 10 or 20 years ago, because compact media had such a narrow throat.

Today's commodity media can handle it.

For 1080p this argument may have merit, although -- are you saying everything should be recorded 4:4:4? Because practically nothing is. 4:2:2 is chroma subsampling, it is considered an effective and efficient way to save space without noticeably impacting the visual fidelity of the picture. And, for many folks, 4:2:0 has served to be sufficient as well.

But here's the kicker -- asking for 4:2:2 4K is just silly. As harddrive and I have harped on forever -- you cannot cannot cannot get 4:2:2 or 4:4:4 out of a Bayer sensor that means anything. You can't get higher than 1920x1080 color resolution out of a 4K bayer image sensor. Sure, you could record it as 4:4:4, but that's just silly, it'd be effectively digital up-rezzing. If you want a true 4:4:4, you need three chips, all recording the native frame size. Or, it takes a massively oversampling sensor, such as the RGB Stripe system as employed in the Sony F35, where there are 5,760 pixels on each line of the sensor, used to record 1920 4:4:4.

4:2:0 in 4K will be plenty to record anything this camera, or any 4K single-chip camera, can produce. There is zero point in asking for more.

So 4:2:2 (or 4:4:4) should be considered as possibilities for 1080p recording, but not for 4K. And, even then, 10-bit 4:4:4 1080p is within the realm of possibility when recording 4K source material; downconverting 8-bit 4:2:0 4K down to 1080p will result in 10-bit 4:4:4 1080p. Not entirely convenient to do it that way, but it is possible to do so, even if the onboard recording were only 8-bit 4:2:0.

 

[Stephen Mick]

For 1080p this argument may have merit, although -- are you saying everything should be recorded 4:4:4? Because practically nothing is. 4:2:2 is chroma subsampling, it is considered an effective and efficient way to save space without noticeably impacting the visual fidelity of the picture. And, for many folks, 4:2:0 has served to be sufficient as well.

But here's the kicker -- asking for 4:2:2 4K is just silly. As harddrive and I have harped on forever -- you cannot cannot cannot get 4:2:2 or 4:4:4 out of a Bayer sensor that means anything. You can't get higher than 1920x1080 color resolution out of a 4K bayer image sensor. Sure, you could record it as 4:4:4, but that's just silly, it'd be effectively digital up-rezzing. If you want a true 4:4:4, you need three chips, all recording the native frame size. Or, it takes a massively oversampling sensor, such as the RGB Stripe system as employed in the Sony F35, where there are 5,760 pixels on each line of the sensor, used to record 1920 4:4:4.

4:2:0 in 4K will be plenty to record anything this camera, or any 4K single-chip camera, can produce. There is zero point in asking for more.

So 4:2:2 (or 4:4:4) should be considered as possibilities for 1080p recording, but not for 4K. And, even then, 10-bit 4:4:4 1080p is within the realm of possibility when recording 4K source material; downconverting 8-bit 4:2:0 4K down to 1080p will result in 10-bit 4:4:4 1080p. Not entirely convenient to do it that way, but it is possible to do so, even if the onboard recording were only 8-bit 4:2:0.

 

[Mike Harvey]

Pat Lamb just responded to say that they're passing along my feedback (inclusion of a 10bit 4:2:2 option, at the very least AVC-LongG 50) to the appropriate product managers.

 

[combatentropy]

Thanks. AVC-Long-50 is a superb codec and I'm surprised it was not already going to have it.

 

[DPStewart]

View attachment 100219

And soda just came flying out of my nose I laughed so hard at that!

 

[Barry_Green]

View attachment 100219

Okay, let me try it this way...

Asking for 4K 4:2:2 on the DVX200 is functionally the equivalent of putting a 300mph speedometer on your VW Beetle. It may make you feel special, but it isn't going to change the car's capability whatsoever... it ain't goin' 300 mph...

 

[DLD]

... And is the difference between, say, $4,700 and $5,000 really going to be a big sticking point between the for customer base for this camera?

That and the cost of executing this request is something no one outside of Panasonic - and folks with a professional knowledge of electronic engineering - knows.

Which is why I asked how much folks were willing to pay for this. combatentropy posted a link recently to a disassembled GoPro, which revealed its components costs at 43% of its final retail price. (i.e., the retail price was ~ cost x 2.5). Given the tag for DVX200, this ratio is probably somewhere around 3 to 3.5. Now, suppose, the 422 upgrade requires pulling an ASIC board out and slotting an FPGA board in. Then you could be looking at the substantial cost/price hike.

Disclaimer - I am not saying that Jaime's wish has no merit but that it requires a certain assumption of this particular product design. In general, cameras - or cars, for that matter - specs aren't usually in a straight line. If you want to do X, you may have to create Y and Z and self-contained products like DVX200 are usually designed with a proper component fit in mind. This is why you often see price-doubling in a pro-video market when you're making a qualitative step up. One can make more incremental changes too but the manufacturers usually choose not to.

Okay, let me try it this way...

Asking for 4K 4:2:2 on the DVX200 is functionally the equivalent of putting a 300mph speedometer on your VW Beetle. It may make you feel special, but it isn't going to change the car's capability whatsoever... it ain't goin' 300 mph...

One can also install a turbo/supercharger into a Beetle but he's not going to turn it into a Porsche Boxter. At some point, the model's innate ecosystem is its own limit.

 

[Jaime Valles]

For 1080p this argument may have merit, although -- are you saying everything should be recorded 4:4:4? Because practically nothing is. 4:2:2 is chroma subsampling, it is considered an effective and efficient way to save space without noticeably impacting the visual fidelity of the picture. And, for many folks, 4:2:0 has served to be sufficient as well.

But here's the kicker -- asking for 4:2:2 4K is just silly. As harddrive and I have harped on forever -- you cannot cannot cannot get 4:2:2 or 4:4:4 out of a Bayer sensor that means anything. You can't get higher than 1920x1080 color resolution out of a 4K bayer image sensor. Sure, you could record it as 4:4:4, but that's just silly, it'd be effectively digital up-rezzing. If you want a true 4:4:4, you need three chips, all recording the native frame size. Or, it takes a massively oversampling sensor, such as the RGB Stripe system as employed in the Sony F35, where there are 5,760 pixels on each line of the sensor, used to record 1920 4:4:4.

4:2:0 in 4K will be plenty to record anything this camera, or any 4K single-chip camera, can produce. There is zero point in asking for more.

So 4:2:2 (or 4:4:4) should be considered as possibilities for 1080p recording, but not for 4K. And, even then, 10-bit 4:4:4 1080p is within the realm of possibility when recording 4K source material; downconverting 8-bit 4:2:0 4K down to 1080p will result in 10-bit 4:4:4 1080p. Not entirely convenient to do it that way, but it is possible to do so, even if the onboard recording were only 8-bit 4:2:0.

I'm confused, Barry. If 4:2:2 4K doesn't do anything, why is it that GH4 4K footage recorded on a Shogun in 10-bit 4:2:2 4K looks better and grades better than 8-bit 4:2:0 4K recorded internally?

 

[ugafan]

 

[joe 1008]

I'm confused, Barry. If 4:2:2 4K doesn't do anything, why is it that GH4 4K footage recorded on a Shogun in 10-bit 4:2:2 4K looks better and grades better than 8-bit 4:2:0 4K recorded internally?

10 bit over 8 bit has a quality effect even in 4K and so does the higher bitrate of the recorder compared to the quite slow 100Mbps of the internal codec. I'm sure the DVX200 will have an 8 bit 150Mbps 4:2:0 codec, which will be an impovement compared to the GH4 and will deliver excellent image quality. In fact in 4K 8bit 4:2:0 recording bears much less problems than in HD. But it would be extremely useful for fast turnarounds and broadcast shooting to have a HD codec that has at least 8bit 4:2:2 recording. 10 bit of course would be much better.

But there is one point that preoccupies me. The specs we know indicate, that the sensor surface for HD is bigger than for 4K. That smells like line skipping and aliasing as the GH4 has in HD. If so, we would have to go the post production path for the best image anyway and a request for a better HD codec would be somewhat obsolete.

If this is not the case and the DVX200 delivers an excellent HD image, I would be even willing though to pay for a better codec upgrade. Even more if I could use it in post with the same ease like DVCPRO or ProRes.

 

[Bassman2003]

One thing I think needs to be stated and is kind of an elephant in the room is that this camera is clearly made for work usage. It is not so much a "cinema" camera but a "video" camera with a larger chip if you know what I mean. To me this means that 4k will NOT be the first choice of recording a lot of the time. My hope is that Panasonic is not drinking the 4k Kool Aid thinking that just because it is there everybody is going to be running around shooting in 4k for their live events to be delivered on the web and DVD. Yes you can get great 1080p from 4k but file size and computer playback factor in a lot for this type of work.

I say all of this to re-state the need for a real 1080p/720p codecs as these are the modes non-cinema stuff is shot on. This and proper color controls in the menu are more important to me than anything 4k.

joe 1008, I am expecting this camera to have a crackin' 1080p image. What we have going against us is the "AG" designation rather than the "AJ". This might prove to keep the camera more towards the consumer side which will make us unhappy.

 

[Ron Evans]

....... What we have going against us is the "AG" designation rather than the "AJ". This might prove to keep the camera more towards the consumer side which will make us unhappy.

Yes it is maybe a large sensor version of the HC-X1000 with a few extras ?

 

[Barry_Green]

10 bit over 8 bit has a quality effect even in 4K and so does the higher bitrate of the recorder compared to the quite slow 100Mbps of the internal codec. I'm sure the DVX200 will have an 8 bit 150Mbps 4:2:0 codec, which will be an impovement compared to the GH4 and will deliver excellent image quality.

Yep.

But it would be extremely useful for fast turnarounds and broadcast shooting to have a HD codec that has at least 8bit 4:2:2 recording. 10 bit of course would be much better.

Totally agreed.

But there is one point that preoccupies me. The specs we know indicate, that the sensor surface for HD is bigger than for 4K. That smells like line skipping and aliasing as the GH4 has in HD.

I'm not worried about this in the slightest. The DVX200 isn't going to be a GH4. It is, more likely, going to have at least as good technology as the HC-X1000 has. And the HC-X1000 has absolutely brilliant 1080p. It's made from a full scan of the 4K sensor and properly scaled on the fly, with zero line skipping or aliasing.

Given that the GH4 is limited to 1080/30p, and the HC-X1000 does 4K/60p, and the DVX200 is doing 4K/60p, it makes me think that the DVX200 is going to be at least as advanced, if not moreso, than the X1000.

[/quote]If this is not the case and the DVX200 delivers an excellent HD image, I would be even willing though to pay for a better codec upgrade. Even more if I could use it in post with the same ease like DVCPRO or ProRes.[/QUOTE]
There's always the option of using your Shogun/Decklink/Pix-E/7Q+ etc as an external recorder with the DVX200. We have already had it confirmed that it will output 10-bit 4:2:2 from its HDMI port. While I am not sold on the value of that 4:2:2, the 10-bit should give the potential for better recordings than the onboard codec.

As for ease of use, consider that the X1000 records directly into Quicktime files, or .MP4 files. Any editor should be able to drag and drop those files right to the timeline. They won't edit as quickly as an intraframe codec like AVC-I or ProRes, that's true... but again, if you were okay with using an external recorder, you could get that as-is.

Not saying I wouldn't want a 10-bit codec onboard! Don't get me wrong. I'm just pointing out that there are options even if it doesn't happen.

 

[Barry_Green]

Yes it is maybe a large sensor version of the HC-X1000 with a few extras ?

I wouldn't be surprised if the internal processing was based off the same basic engines as used in the X1000; it is a current 4K model from Panasonic, and it is currently their only professional 4K handycam, so it seems reasonable to expect that the DVX200 would share familial traits with it.

Hopefully though, it will feature a lot of extras. We already know that it has more features (such as V-Log, and variable frame rates, and 1080 up to 120p, and a timecode input port). We don't know what additional features it may have yet, but hopefully it will include the return of the waveform monitor, etc.

 

[joe 1008]

I'm not worried about this in the slightest. The DVX200 isn't going to be a GH4. It is, more likely, going to have at least as good technology as the HC-X1000 has. And the HC-X1000 has absolutely brilliant 1080p. It's made from a full scan of the 4K sensor

... ... ... ...

There's always the option of using your Shogun/Decklink/Pix-E/7Q+ etc as an external recorder with the DVX200. We have already had it confirmed that it will output 10-bit 4:2:2 from its HDMI port. While I am not sold on the value of that 4:2:2, the 10-bit should give the potential for better recordings than the onboard codec.

I strongly put my faith in that. Aliasing is a big downwer for me, above everything else.

Concerning an external recorder: in the case of the DVX200 I would absolutely favour an internal solution, as anything added externally to this camera would go against its philosophy - at least if I don't shoot a feature with it. I want to be able to carry this camera through rough terrain and don't worry about external add ons, while producing footage accepted by TV stations for example. The Sony PXW-X70 has such a codec for almost half the price. So the DVX200 should have it, too - and exceed cameras with smaller sensor in every aspect.

 

[ugafan]

I want to be able to carry this camera through rough terrain and don't worry about external add ons, while producing footage accepted by TV stations for example.

while the closest i'll ever get to rough terrain is climbing over the mound of crap i've got in the garage, i agree wholeheartedly with this. i don't want to have to pay for external solutions, i want a camera that does it from the start.

 

[Jaime Valles]

while the closest i'll ever get to rough terrain is climbing over the mound of crap i've got in the garage, i agree wholeheartedly with this. i don't want to have to pay for external solutions, i want a camera that does it from the start.

+1,000,000