Sony 4K Handicam pic leaked

[Glen Vandermolen]

To me, the article is talking about two different rumors regarding two different cameras. I don't think it's ambiguous at all. Or at least that's how I'm interpreting it. And remember, translating info from asian sources doesn't always give you precise info. Which is why the author gave the medium frame (format?) rumor a low confidence rating.

Besides, can Sony make a 20X zoom for a medium frame sized sensor in a small camcorder body? Assuming the zoom info is correct, doesn't a 20X zoom, based on the picture, almost guarantee a small sensor? If we assume the sensor is a "medium format" at around 1", even 2/3", how big will a 20X zoom have to be? And all this for under $5,000? The only 20X zooms I've seen on camcorders have always been on 1/3" chip cams, 14-16X zooms on some 1/2" cams.

[David Heath]

Originally Posted by Glen Vandermolen

Besides, can Sony make a 20X zoom for a medium frame sized sensor in a small camcorder body? Assuming the zoom info is correct, doesn't a 20X zoom, based on the picture, almost guarantee a small sensor?

I'd say no - the 20x zoom does not guarantee a small sensor. (With the caveats below!)

It all comes back round to the arguments in the thread on the IMEC 2/3" sensor thread.

All lens design is a compromise, and one of the factors is max f stop and how it ramps over the zoom range.

It follows that in general principle, it's perfectly possible to have a :

1] Large sensor
2] Big zoom range
3] Size as expected here

as long as the designer is prepared to accept a fairly small max f no That's the compromise that the laws of optics force.

It's only if you want 1] Large sensor, 2] Big zoom range *AND* something like f1.8 that it MUST be much larger and more expensive.

Before you say "but I don't want an f5.6 max lens! That'll be no good in low light" then (as said before) for a given size of lens, then sensor size is irrelevant independently to low light ability. All else equal, a 1" sensor may be 3 stops more sensitive than one of 1/3" - but for the same physical size of lens, that will be 3 stops less when used for 1" rather than 1/3". (Say f5.6 rather than f2.)

[Glen Vandermolen]

Originally Posted by David Heath

I'd say no - the 20x zoom does not guarantee a small sensor. (With the caveats below!)
)

Nah, I'm not buying it. I'm betting no larger than a 1/2" sensor, more likely a 1/3" sensor set. Not with a 20X zoom, not with the sensitivity needed for all the pixels crammed into a 4K chip. What is that, about 8 million pixels a chip?

Trying to market a maximum F5.6 lens? No way. In full 20X zoom, what would a F5.6 lens ramp down to? Why bother making it? The physics may work for a F5.6, 20X zoom for a large 4K sensor, but actually trying to sell one? Under $5,000? It doesn't make any sense, not for a Handycam.
But, that's assuming the lens really is a 20X zoom.
A 1/3" 4K image may not look the best, but this is marketed as a Handycam, not as a competitor to an Epic.

And if this is a response to the BM4K, this prototype came out at CES 2013, before the BM4K came out at NAB 2013.

[Cliff Totten]

If that lens is the same one that has been around for several years, it's traditionally used on 1/3 sensors in Sony cameras.

I strongly suspect this is still the case.

But what sensor this time? 1/3 inch 4K? (low light performance will be challenged)
Will it be a single bayer sensor?
Will it be a full raster 3 sensor block? (no Bayer problems)
Will the pixel count be lower than 4K and then "pixel shift" up to 4K? (Sony has done similar things in other products)
Will it be a single sensor that has a 20% oversample for Bayer removal?

If this a "Handycam" it MUST be considered "low end" and MUST be the "least" Sony offers. It's safe to say that they have higher spec plans for better cameras up stream to meet the current "NXCAM-style" market.

So, the question still is; How with they limit and cripple this camera? There are at least a dozen ways they have done this in the past.

Fascinating!

Now we know why Sony released almost nothing at NAB this year. They are working on their "next-gen" product fleet.

CT

[David Heath]

Originally Posted by Glen Vandermolen

Nah, I'm not buying it. I'm betting no larger than a 1/2" sensor, more likely a 1/3" sensor set.

I'd actually take you up on that bet, but probably to no more than about $10. :-) My money would be on 1", possibly 2/3". Either way, I think we'll know for sure which of us is right pretty soon - my guess would be IBC.

Originally Posted by Glen Vandermolen

Not with a 20X zoom, not with the sensitivity needed for all the pixels crammed into a 4K chip. What is that, about 8 million pixels a chip?

If it's 4k, it's just under 9 megapixels, but surely that argues for a bigger chip size? A 1" chip is about 16mm horizontal, a 1/3" is about 6mm. Assume 4,000 photosites horizontally and that gives a photosite spacing of about 4 microns for 1", but only about 1.5 microns for 1/3". That makes the area of the photosites on the 1" chip far bigger - even more than simple numbers suggest when you take inter-site spacing into account.

Originally Posted by Glen Vandermolen

Trying to market a maximum F5.6 lens? No way. In full 20X zoom, what would a F5.6 lens ramp down to? Why bother making it? The physics may work for a F5.6, 20X zoom for a large 4K sensor, but actually trying to sell one?

Here I'll go along with you. In real terms, f5.6 on a 1" sensor and f2 on a 1/3" sensor should give the same low light performance - but yes, unfortunately it's simple numbers that marketing people like. That said, exactly this has been happening with the DSLR lens market for a while. Lenses for such are typically much slower than they used to be (to get zoom range up, whilst keeping cost and size down) - but since it's permissible to set a DSLR to a far higher ISO than you'd previously have considered for film, it's been seen as OK.

Practically, the f5.6 number was meant as an illustration of the principle, and to compare 1/3" with 1". Maybe a better comparison is the 1/2" PMW200 with an f1.9 lens. Keep the same front element dimensions and a 1" version of that may be expected to be f3.5 (2 stops down). Maybe make it a little bigger and you're at about f3.2....... sound better for marketing? :-)

We'll see......

[David Heath]

Originally Posted by Cliff Totten

If that lens is the same one that has been around for several years, it's traditionally used on 1/3 sensors in Sony cameras.

Highly unlikely it's exactly the same, I'd expect it may have similarities, but almost certain to differ to meet 4K requirements.

It's also possible to consider the separate parts of a zoom lens individually (at least in principle). The same front section can be used with differing rear sections according to the sensor area to be covered, which obviously affects the effective f stop.

Now we know why Sony released almost nothing at NAB this year. They are working on their "next-gen" product fleet.

Errr, they might not have actually made the announcement at NAB, but I think you may be forgetting the F5/55.....? True, announced a couple of months before, but they were just starting to ship around that time and I'm told were what got the most attention on the Sony stand at NAB.

And what about the PMW400? Didn't that get announced very close to NAB, if not actually at the event?

[Glen Vandermolen]

Originally Posted by David Heath

I'd actually take you up on that bet, but probably to no more than about $10. :-) My money would be on 1", possibly 2/3". Either way, I think we'll know for sure which of us is right pretty soon - my guess would be IBC.
..

OK, $10 says the chip(s) is no larger than 1/2". But I'm guessing there's no betting on this site, so not sure we can ever collect. ;-)

[David Heath]

Originally Posted by Glen Vandermolen

OK, $10 says the chip(s) is no larger than 1/2". But I'm guessing there's no betting on this site, so not sure we can ever collect. ;-)

:-) Note that I wasn't prepared to bet more!

I'm not going to comment anymore other than to say that in a camera at this price point, if the sensor is 2/3" or bigger, this would be a very, very interesting camera. If you're right, and it's 1/3", it's nowhere near as attractive. Maybe I was a little harsh in what I said in post #6, but it's as Monday Isa said earlier on - if it is 1/3" you would need to be at about f2.8 or wider to get full benefit of 4K resolution. That would give it a usable iris range of likely little more than one stop!

[David Heath]

Originally Posted by David Heath

........ but it's as Monday Isa said earlier on - if it is 1/3" you would need to be at about f2.8 or wider to get full benefit of 4K resolution. That would give it a usable iris range of likely little more than one stop!

I wasn't going to comment any more on this until definite news, but one point that's just been made to me in a discussion is something I'd overlooked. Namely that whilst the basis of what I'd previously said was sound, then it's down not to the sensor size as such that really determines diffraction limiting, but rather the actual physical size of the lens components.

So compared to an f2 lens for 1/3", then if the physical dimensions of the lens remain roughly the same, whilst it may indeed translate to an f5.6 lens for 1" that won't gain you any advantage. It'll still start to diffraction limit stopped down the same amount - so if the usable range for 1/3" is (say) f2-f2.8, the corresponding figure for 1" may be about f5.6-f8. You still only have a one stop usable range.

The truth is that as the system resolution increases, the laws of physics dictate that the only way to maintain much of a usable aperture range is to physically increase lens size.

I was also made to think about exactly what some of the spacing numbers actually mean. As said earlier, for 4k type resolutions, a 1/3" chip must mean photosite spacing of roughly 1.5 microns (1500 nanometre), for a 1" chip it's roughly 4 micron (4000 nanometres). Since the wavelength of red light is around 700 nanometres, 4k resolution on a 1/3" chip effectively means spacing approaching the wavelengths of light.

[Monday Isa]

The reason I believe it's 1/3" chips is the 20X zoom and also the sensor is EXMOR R. EXMOR R is a back illuminated sensor which till now have been for smaller sensors then 1/3" which pack a lot of pixels on a small size. 4K on 1/3" sensors means Sony would have to use back illumination. The NX70 uses EXMOR R but it's a 6MP sensor on a single 1/2.88 chip. Sony could use this size as a 3 chipper. We'll find out soon enough. I honestly believe its a small sensor and those that know about diffraction and how it effects image quality will have a hard time exposing with this cam as you need 2.8 and higher to get a sharp image where you can start to get into diffraction problems.

[Ron Evans]

Interesting comments about sensor and lens sizes. My own experience. I have a NX5U ( looks just like the 4K version) with 3 1/3" sensors, I have several small Sony AVCHD cams too. Also a HX30V still camera with an even smaller lens and 1/2.3 sensor and 20x optical zoom. In good light this still camera shoots lovely 60p video as good as my CX700 or NX30U. The 1/2.88 sensor in the CX700 or the similar sensor in the NX30U easily out perform the NX5U in low light with a much bigger lens and 3 1/3 sensors. The small cameras do of course have newer "R" sensors compared to the NX5U with about 6M pixels. Operating these small cameras requires a different approach and I run them in semi auto mode so that the designers knowledge that is in the on board processor has control not me. I just tell what to focus on and how light or dark I want the image. Touch focus and AE shift is the key. I think it will be a single sensor a bit bigger than 1/2 so that the depth of field is reasonable for camcorder use. If users want the shallow depth of field they will go the NEX range.

I used my FX1 when it came out in HDV mode but edited in DV so that I could pan around the image and am keen to do the same with this 4K version if it does in fact arrive.

Ron Evans

[Jack Zhang]

What we may be overlooking is "ClearVid."

The diagonal pixel arrangement that claimed better sensitivity, at the cost of interpolation.

[David Heath]

Originally Posted by Jack Zhang

What we may be overlooking is "ClearVid."

The diagonal pixel arrangement that claimed better sensitivity, at the cost of interpolation.

As I recall, in the early days of 1/3" HD, the first two camera sensors (3 chip) either used 960x540 sensors (HVX200) or 960x1080 with horizontal pixel shifting (Z1). (Approx 0.5 megapixel and 1 megapixel respectively.) Then came the EX1 with 1920x1080 (2 megapixel) - but went to a 1/2" chip to keep the photosite size up.

My understanding of the reasoning behind the diagonal ClearVid technology is that 1 megapixel on 1/3" was seen as desirable for photosite size reasons a good compromise on 1/3" chips between 0.5 megapixel and 2 megapixel - but undesirable as it wasn't horizontally/vertically symmetrical in resolution terms. (That mattered less for the Z1 and 1080i - more so as progressive scan became more common.)

The point of the diagonal arrangement is that it allows a 1 megapixel sensor to have equal horizontal/vertical characteristics (an effective resolution of about 1440x675 after processing), whilst keeping fairly straightforward processing arrangements (it treats the diagonal matrix as two interleaved 960x540 diagonal matrices, and processes them to a 1920x1080 output)

As far as this thread goes, then if the main boast of this product is 4k, I don't see how any ClearVid arrangement is going to help, without lowering resolution such that it can't truly be called 4k?

[Ron Evans]

Originally Posted by David Heath

As far as this thread goes, then if the main boast of this product is 4k, I don't see how any ClearVid arrangement is going to help, without lowering resolution such that it can't truly be called 4k?

I see no reason why a Clearvid arrangement could not be devised to create an interpolated 4 K output just like it did for 1920x1080. I fully expect this first Sony product to be interpolated in some way as the benefits in sensitivity are certainly real and to me more important than a definition that requires sensors of 4K. If the sensor and on board electronics generate a recordable 4K output its 4K !!!! Of course it will not have the tested resolution of a very expensive production camera under a range of conditions. I don't expect it to. But I also expect it to be very acceptable in most conditions that the interpolation will work effectively and accept that the conditions will effect the performance of the interpolation just like it does in all the Sony AVCHD cameras I have now. Another reason to let the camera decide settings in a semi auto mode as it knows how best to setup for optimum performance. Tell the camera what you want in focus and how light or dark you want the image. I think if one is interested in controlled depth of field etc one is going to have to spend a lot more money !!! At least an NEX FS700 +

Ron Evans

[David Heath]

Originally Posted by Ron Evans

If the sensor and on board electronics generate a recordable 4K output its 4K !!!!

Hmmm - becomes a bit of a question of where you draw the line..... ? At what point do you say "that's not 4k! It's just 1080 upscaled!" :-)

Even cameras like the F5/55 (which have a 4096x2160 Bayer sensor) are considered by some to be stretching the definition of the term "4k" - they've got a 4k sensor - but after deBayering the max resolution is going to be only 80% max of what 4k is really capable of.

I'm inclined to give those cameras the benefit of the doubt, but any less.....?

I suppose a lot may depend on whether it's 3 chip or 1 chip, how practical 3 chip is at these sizes and resolutions. In which case (if 3 chip) maybe 3 ClearVid 4.5 megapixel chips? Which should give resolution to about 2880x1620. Would that satisfy to be called 4k?

I think if one is interested in controlled depth of field etc one is going to have to spend a lot more money !!! At least an NEX FS700 +

It's not just depth of field, it's usable aperture range. In pocket HD cameras it's already the case that the iris hardly changes to vary exposure, it can't without the image becoming unacceptably soft. They rely on a continuous variable ND moving in the iris plane to control the amount of light transmitted. That may be OK for a small pocket camera, but not for anything more professional.

But as far as spending more money to get something more professional, then yes, you're right. I suppose it should be remembered that if the report is true, this is going to be under $5,000, so perhaps one should be realistic with expectations.