IMEC codevelops 4k/2k CMOS 12 bit sensor w/Panasonic for 2/3 inch

[David Heath]

Originally Posted by Alister Chapman

Might be an interesting camera for those needing to shoot 4K run and gun, but sensitivity is likely to be an issue, you really want a small aperture for run and gun for a deep DoF but diffraction effects will force you to use larger apertures.

Getting 4K with deep depth of field is going to be a problem, period, and the sensor size is irrelevant if you work it through. For a given resolution and angle of view, the ONLY factor that is relevant to dof is the physical size of the aperture (not the f stop, but actual dimensions).

A given size of aperture will be equivalent to (say) f4 for 2/3", f11 for s35. And in each case the dof will be the same. Start then to stop down in each case, and they will both start to diffraction limit the same no of stops down - when they are both the same physical size.

My bet would be for a single chip bayer design. I don't know what the practicalities of creating a high enough quality prism are? Might be too expensive.

I commented on whether it will get utilised in a single or 3 chip form early on in this thread, and well, we'll have to wait and see.

If it is 3 chip, then this is a way this (2/3" 4K) could have a definite unequivocal advantage over a single chip 4K s35 sensor, at least in terms of low light performance, and the reasoning follows on from the above.

For a given size, weight etc of lens then we'll assume the physical size of the aperture is the same whether it's used for 2/3" or s35. As regards f stop, the s35 version must then be 3 stops lower - but the sensor will be 3 stops higher intrinsically, so performance should be the same, the factors cancel each other out.

The only way to improve the performance of either is to either increase the physical size of the lens - or go from single chip to 3 chip. Put another way, for a given size of lens, a 3 chip 2/3" camera will have at least a stop advantage over a single chip s35. (I'm assuming that 3 chip s35 is out of the question.)

[Alister Chapman]

A prism is a non starter at s35 as the flange back would be too deep for DSLR or PL lenses. Any wide lenses would need to be retro-focal to accommodate the very deep flange back. Even at 2/3" with a prism to get a wide FoV the lens must have a retro-focal design (one of the reasons 2/3" zooms are expensive compared to DSLR zooms). This is one of the reasons why I think the Pana 4K will be single chip, the lens will be simpler and much cheaper, the lens design for this camera will be a challenge.

I'm not arguing against David's science on the lens size for sensor size sensor/aperture argument, but there is also the signal to noise ratio to be considered, which with smaller pixels will most likely be higher than with the large pixels on most s35 sensors. So the camera will be less sensitive for any given SNr. Even if the camera is a 3 chip prism design, that alone only brings around a 1 stop advantage (prisms are not loss less they normally include absorption filters for passband trimming).
The reality is that right now you can easily get f1.8 primes and f2.8 DSLR zooms that are neither large or bulky for s35. To get the same on a 2/3" sensor your looking at f0.6 and f1.0 lenses and I have yet to see a decent f1.0 zoom, so in practice s35 will remain the better performer in low light. My guess would be f1.8 as is typical of most normal 2/3" B4 lenses. So you won't get as much light on to the sensor and the sensor will have a higher noise floor, so overall I think s35 will still remain a much better performer in low light.

[Glen Vandermolen]

Originally Posted by Alister Chapman

A prism is a non starter at s35 as the flange back would be too deep for DSLR or PL lenses. Any wide lenses would need to be retro-focal to accommodate the very deep flange back. Even at 2/3" with a prism to get a wide FoV the lens must have a retro-focal design (one of the reasons 2/3" zooms are expensive compared to DSLR zooms). This is one of the reasons why I think the Pana 4K will be single chip, the lens will be simpler and much cheaper, the lens design for this camera will be a challenge.
.

And there's the HPX600 with a single 2/3" chip design. Interesting.
I wonder....

[David Heath]

Even if the camera is a 3 chip prism design, that alone only brings around a 1 stop advantage

Yes, as a rule of thumb, for the same sensor size, a 3 chip design will have an advantage of about a stop compared to an equivalent single chip design.

The reality is that right now you can easily get f1.8 primes and f2.8 DSLR zooms that are neither large or bulky for s35. To get the same on a 2/3" sensor your looking at f0.6 and f1.0 lenses and I have yet to see a decent f1.0 zoom, so in practice s35 will remain the better performer in low light.

The lenses you describe are likely to either be prime or short zooms, which may be just what one market sector (let's call it the "digital cinema" market) wants and needs. I agree with what you say above, so yes, here s35 is best suited for the reasons you say.

BUT let's think about another market sector, of which live sport etc may be one example. The type of work traditionally where 2/3" cameras have been seen to be best. Here the expectation is that long range zooms are the norm, and cost/size/weight dictate the max aperture - to say nothing of cost. For a given size of lens, it may be f1.8 when designed for 2/3", but 3 stops down when designed for s35. Comes to the same thing, same performance, as the s35 chip is likely to have an ISO 3 stops higher - it's not true to say long range zooms are prohibitive for s35.

THEORETICALLY, it may be possible to design a long range s35 zoom with max aperture of f1.8 - but the cost is likely to be prohibitive, to say nothing of size and weight!

The implications are that if both sensors (s35 and 2/3") are single chip, the only way to get better low light performance is to make a lens with a bigger front element - hence bigger, heavier, more expensive etc. OR to go 3 chip - which will give a stop improvement for the same lens. And since (as Alister says) that is not practical for s35, then if implemented for 2/3" would immediately give such a camera a benefit relative to s35 for such lenses.

And that's why whether this sensor gets implemented in single or 3 chip form is so important. If in single form, it gives no benefit over 4k s35 sensors for use with long range zooms - and lacks the versatility of being able to be used with fast primes or short zooms. But if in 3 chip form it immediately gives the camera an edge if the primary usage is seen as with long range zoom lenses. For comparable lens sizes, it will give it a stop sensitivity advantage compared to s35, AND will give true 4:4;4 full 4k resolution, better than a 4k bayer can manage.