PMW-400 gets XAVC for FREE

[harddrive]

Just wondering...how much difference is there in overall sharpness between the 320 and 350 and how different do you think the 400 will be from these two?

Resolution wise, don't expect a lot of difference, and same for the 400. Big difference between the 320/350 is due to the chip size and the 350 being about a stop more sensitive, and having about a stop more control over depth of field. There is also a lot more choice available for 2/3" alternative lenses than 1/2".

When I was buying, I'd have preferred the 350 - but the 320 is much cheaper (and does SD as well out of the box for no extra) so I went with it on cost grounds.

 

[Barry_Green]

the HPX600 is only a single chip, believed to be 1920x1080 Bayer CMOS.

Believed by whom?[/quote]

That means much less than 1920x1080 resolution after processing, and at least a stop worse sensitivity as well.

C'mon, harddrive, I know you never miss a beat to cast Sony in the best possible light and Panasonic in the worst possible light, but can you at least have some facts to back up your statements? How do you know the Panasonic doesn't have a 4K chip and is doing on-the-fly 4K-to-1080p downconverting?

Would it be too much to ask, to at least see some footage, before you start flinging mud?

 

[harddrive]

C'mon, harddrive, I know you never miss a beat to cast Sony in the best possible light and Panasonic in the worst possible light.......

If a 2/3" single chip camera had come from Sony, and an equivalent 2/3" three chipper from Panasonic, both on sale at the same price, then be quite sure I'd be casting Panasonic in a good light, Sony in a bad one. The fact it's the other way round is not my fault. It doesn't change the fact that all else equal, a three chipper will substantially outperform a single chipper, whoever makes which.

........can you at least have some facts to back up your statements?

I don't have definitive charts (hence my use of the word "believed"), but side by side comparisons on normal pictures seem to show the HPX600 substantially softer than other 2/3" 1920x1080 3 chip cameras. (Not just from Sony, either.) That's in line with what would be expected from a 1920x1080 sensor deBayered.

And I and many other people, both on forums and personally, have directly put the question to Panasonic "what are the sensor characteristics"? Result has been a stunning silence. Why?

How do you know the Panasonic doesn't have a 4K chip and is doing on-the-fly 4K-to-1080p downconverting?

I don't - that would certainly be a possibility, though I'd expect the end result to be sharper. But if you have access to anyone at Panasonic who actually knows the answer, then could you get a definitive answer, once and for all? Both as to the chip specifications, and why Panasonic chose to make it single 2/3" chip? If it is the way you suggest (which I don't deny, it could be) why aren't Panasonic proud to announce it, as may be expected?

The general assumption seems to be that with all the mystery, expect the worst. That a single 1920x1080 is most likely.

Would it be too much to ask, to at least see some footage, before you start flinging mud?

Regardless of the actual resolution characteristics, then for the same basic chip size and technology, using a single with Bayer sensor is inevitably going to mean much poorer sensitivity than using three with a beam splitter. That's basic theory, not "flinging mud". OK the other aspect is resolution, and with a 4K chip read in 2x2 blocks, you will get full 1920x1080 resolution - but if that's the case, why don't Panasonic just say so?

 

[Barry_Green]

It doesn't change the fact that all else equal, a three chipper will substantially outperform a single chipper, whoever makes which.

That is, of course, agreed upon. But the point is, you don't know whether things are equal or not.

The general assumption seems to be that with all the mystery, expect the worst. That a single 1920x1080 is most likely.

Well, maybe that's a difference between your culture and mine. Over on this side of the pond, we say "innocent until proven guilty." I don't know what chip they have in there. But I do know, that all the other manufacturers of cameras that are catching any attention at all, are using single chip. Including Sony. F5, F55, F65, F3, FS100, FS700, Red, BlackMagic, they're all single sensor. So maybe they felt that that's what the market wants now, is a single sensor, but they decided to make it 2/3" to be compatible with the existing lenses. I'm not saying it's a winning strategy, but it's a decision that, if that's what they decided, I can see how they got there.

Regardless of the actual resolution characteristics, then for the same basic chip size and technology, using a single with Bayer sensor is inevitably going to mean much poorer sensitivity than using three with a beam splitter.

Obviously.

That's basic theory, not "flinging mud".

No, that's not flinging mud. Flinging mud is immediately assuming the worst, and then spreading that as if it's known fact. It's not known fact. You are assuming. Just like you did with AVCI Class 200.

You do know what they say about what happens when you assume, right?

Sorry, but it's a real pet peeve of mine when someone makes an assumption and then starts spreading it around like it's fact or "commonly accepted" or whatever. BS is BS, whether it's "commonly accepted" or not. You're better than that, I know you are.

 

[vanvideo]

Gotta agree with Barry.


I understand the HPX600 has a 3K sized sensor. So, I assume it oversamples (is that the right word?) to get a 1920x1080 image. Is that better than three 1920x1080 2/3" chips? I dunno.

 

[harddrive]

That is, of course, agreed upon. But the point is, you don't know whether things are equal or not.

What I do know is that Bayer filtration MUST mean that over half the light let through the lens will get absorbed in the Bayer filters. The same is not true when a beam splitter and three chips are used. For a single chip design of the same basic size, the best possible case for sensitivity would be equivalent photosite numbers. Go to a 4K read in 2x2 blocks and expect better resolution, aliasing etc figures - but worse sensitivity (more light wasted in the inter photosite gaps). And nowadays basic sensor sensitivity is mostly limited by physical laws for a given photosite size. The basic CMOS technology used in the HPX600 would have to be over twice as efficient as the PMW400 even to get parity. I will put a pretty large bet with anyone that that is impossible.

Over on this side of the pond, we say "innocent until proven guilty."

Hopefully that's true in most democratic countries! But if the accused gives "no comment" at all to quite reasonable questions ("where were you on the night of...") in a courtroom, it isn't likely to impress any jury, is it? If the defendant is innocent, wouldn't you expect him to be open with his answers? And the same here. Many people (me included) have openly asked about a fairly basic specification - and got no sensible answer.

Could it be that they don't want to give details of the chip dimensions for normal commercial reasons? Not really likely. Anyone with the camera and suitable charts and knowhow could work out the basics pretty quickly, it'd be child's play to any of the other manufacturers without even taking the cover off.

I don't know what chip they have in there. But I do know, that all the other manufacturers of cameras that are catching any attention at all, are using single chip. Including Sony. F5, F55, F65, F3, FS100, FS700, Red, BlackMagic, they're all single sensor.

But *ALL* the examples you give there are large sensor, mostly s35. There are very good engineering reasons why 3 chip s35 is problematic, especially at higher resolutions, not least being size of the beamsplitter. Likewise, the size is such that losing a stop through being single chip is more than made up for by sheer size. An s35 sensor is about 8x the area of 2/3". The size difference should make +3 stops difference, the single chip nature -1 stop. Single chip s35 will still be about 2 stops better overall than 3 chip 2/3".

Can you name me one single other camera in this class that is single chip? One single other shouldermount 2/3" camera? There are sound reasons for making s35 cameras single chip - sound reasons for making 2/3" and below three chip.

Sorry, but it's a real pet peeve of mine when someone makes an assumption and then starts spreading it around like it's fact or "commonly accepted" or whatever. BS is BS, whether it's "commonly accepted" or not. You're better than that, I know you are.

I can understand your annoyance in principle - but it's not a snap of the fingers "I've had a thought!" matter and immediately spread around without any checking. The HPX600 was announced quite a while ago. (A year?) The question of basic specifications has been put to various Panasonic people many times on many occasions by many people - yet no answer at all to what most people consider a fairly basic question. A bit like going into a car showroom and asking the engine size of the model you're interested in.

But let's move on. With AVCI class 200, it was poorly written press statements that led to confusion, along with an unannounced change of terminology. But you managed to clarify matters through your contacts.

So is there any hope you could do the same again? Get us a definitive answer about the basic specifications of the HPX600 sensor?

Or has anyone had the chance to look at a zone plate with it? I'm afraid anybody I know who may have the means to do such is just not interested enough in the HPX600 to warrant giving it full testing.

 

[harddrive]

I understand the HPX600 has a 3K sized sensor. So, I assume it oversamples (is that the right word?) to get a 1920x1080 image. Is that better than three 1920x1080 2/3" chips? I dunno.

No is the simple answer. If that was to be the case, it would give decent 1920x1080 resolution - but would still have an inherent sensitivity disadvantage of more than a stop.

It would also be more complex to process than either a single 1920x1080 chip, or the 3840x2160 idea proposed by Barry. Both of those would yield a 1920x1080 output signal directly (at full res in the case of the latter - like the Canon C300), a 3K chip would need downconversion - and that tends to need more power to do well.

It's really difficult to see the point of it. Why bother developing such a chip and giving yourself the effort of the downconversion, when 3x1920x1080 chips and beamsplitter is established technology needing less processing and with an inherent sensitivity advantage?