Oft-sought camera not announced again

[combatentropy]

That's not how lens mounts work. The 4/3 mount will not cut off the imaging circle of whatever lens you attach. The camera pictured in my first post is the JVC GY-LSX2, and it is exactly that, an S35 sensor with a 4/3 mount, and has both modes. Neither would crop modes "make the processing much more complex." It's a common feature among cameras. Nor would it substantially degrade the image (3K to 2K is what Alexa does; 2K to 2K is what Ikonoskop, Blackmagic Pocket, and Digital Bolex do; which look great to me and many other people, but if that's not good enough for you, just don't use 4/3 or C-mount lenses). Please stop throwing out conjecture and muddying this thread with misinformation.

 

[harddrive]

Please stop throwing out conjecture and muddying this thread with misinformation.

I said quite clearly at the start of my last post "I may have got it completely wrong, but I don't think that will work ......". That is a world away from "muddying this thread with misinformation", it's stating an honest concern, making it clear there is uncertainty, and inviting response. If I can refer to your own link, that itself says:

As Edgar Shane from JVC put it this lens mount is MFT and many people rightfully say that MFT lenses shouldn’t cover its super35mmm sized sensor. But apparently there are quite a few lenses that will cover it very well and the Olympus models Edgar mentioned are just the tip of the iceberg.

Some examples of lenses that cover super35 or more include......

In other words, what you describe certainly will work in some cases, but it seems by no means certain that all lenses suitable for s35 coverage will still cover an s35 sensor when adapted via a 4/3 mount and convertor? Or can you guarantee that's not the case? Because if not, and if you want to use an s35 sensor, it seems far simpler to just use something like an EF mount on the camera, doesn't it?

Neither would crop modes "make the processing much more complex." It's a common feature among cameras. Nor would it substantially degrade the image (3K to 2K is what Alexa does; 2K to 2K is what Ikonoskop, Blackmagic Pocket, and Digital Bolex do; which look great to me and many other people, but if that's not good enough for you, just don't use 4/3 or C-mount lenses).

The advantage to 3840x2160 (for now ignoring anything to do with 4K itself) is that it *EASILY* forms a top quality 1080 image of full HD quality. It needs little processing, which is a big factor why power consumption for such as the C300 is so low. Simply form every output pixel directly from a unique 2x2 Bayer block by a direct read of values for R,G,B - end of story.

In your example of the Alexa, and 3K-2K processing, it's quite true that it may achieve comparable (and high) quality - but it comes at the expense of far higher processing complexity, that's the whole point. Look at the price of Alexa, look at the power consumption. As compared to the very simple way the C300 gets it's result, if you start with something like 2880x1620, that has to be properly deBayered in real time, then the resultant 2880x1620 RGB matrix downconverted in real time to 1920x1080. To do that well in real time, without introducing too many downconversion aliases, is certainly possible - but demanding.

An alternative is to record RAW, and leave all that to be done in a computer (with high computing power and no real time restraints) later - which is what most of your other camera examples tend to do. Fine - but that brings constraints of it's own. Extra work in post, and high data rates needing relatively large amounts of high speed memory.

Do the bigger crop, and you won't get full quality 1080 results from a 2K sensor crop after deBayering. There is very sound reasoning behind why cameras like the Alexa, the F3 etc use an oversampled sensor to get 1080.

Perhaps I should have worded the previous post "make the processing much more complex and/or yielding lower quality" (rather than "whilst yielding lower quality"), but the theory behind it is perfectly sound.

Combatentropy, what you're proposing may work - but I don't really see the big advantage to it? If you must use 4/3 lenses, fine, but use them with a 4/3 sensor to get the most out of them. Ideally one with a 4096x2160 photosite count. If you want to use a 4096x2160 s35 sensor, then why not simply build the camera with a more appropriate mount, and use lenses which will get the best possible out of that?