Panasonic rumors getting interesting

[John Camp]

Rumors around the new Panasonic FF are getting interesting -- the most recent rumors suggest that the new camera will use the Leica SL mount and that Panasonic, Leica and Sigma are all involved in the new system. The rumors suggest that there will be substantial number of native lenses at launch, that the sensor, designed by Panasonic, will probably have as many or more MPs than the Z7. Some of the rumors also suggest that the m4/3 lenses may be used in some manner. I don't follow camera rumors too closely, since there's usually a pretty large BS factor, but I've had a Panny m4/3 system for years and really like what they've done with it. If these rumors are true, the new camera could be one I'd be interested in. (I have a Nikon system in addition to my Panasonic GX8s, and I'm going to get a Z7 if the new Panasonic turns out to be less than rumored. But now, I'm waiting.)

[BernardLanguillier]

Interesting indeed!

I hope that the lenses are more appealing than the SL ones that I find way too bulky, way too expensive and not that amazing image quality wise.

Cheers,
Bernard

[johnvanatta]

I had thought the Z7 was so perfect for me that nobody else could possibly match (Canon certainly didn't). But this rumor at least gives me pause. I'd expect a Panasonic body to be the most mature of all the entrants crowding in; they've been doing mirrorless for a long time now. And they'd have the ability to get a wide range of lenses produced, too, from Sigma. What I think could really set it apart is a Foveon camera. It's a neat tech that hasn't really found a clear home yet; and I don't know if it can match a Bayer as the *primary* camera. But having a Pansonic body for general purpose and a Sigma Foveon for when the situation calls for it? Could be very interesting, especially if Olympus joins in--they're a clear leader in tech.

[Kirk_C]

What I think could really set it apart is a Foveon camera. It's a neat tech that hasn't really found a clear home yet; and I don't know if it can match a Bayer as the *primary* camera.

I can't imagine we'll see the Foveon sensor from anyone except Sigma. Too many problems with color balance, with acceptable skin tones being particularly hard to achieve.

Panasonic will be going after the broadest possible market is they choose to enter FF and it will need to be a very good sensor with all the latest technologies incorporated.

[JaapD]

The Organic sensor possibly ?

[davidgp]

The Organic sensor possibly ?

The announcement of the prototype is quite new : https://m.dpreview.com/news/1440456457/panasonic-unveils-industry-first-8k-organic-image-sensor-with-global-shutter , so... it will depend how far they are to mass produce it, Panasonic didn’t comment that in their press release, and rumors say nothing. Adding a extra step in a CMOS process line can take some time (years)... but maybe they were already working on that for some time...

P.D.: there was already a thread opened about this Panasonic FF rumor...



http://dgpfotografia.com

[Dan Wells]

If the rumors about Sigma being a part of this are true, they have two built-in sources of excellent lenses . The fly in the ointment is that the Sigma Art lenses are heavy and bulky, and their FE-mount versions are just DSLR lenses with a permanent extension tube, not lenses actually designed for the shorter flange distances of mirrorless cameras. There has been some speculation, notably by Lloyd Chambers,that the front-heavy Art lenses might put a lot of torque on the camera side of the mount (big, heavy lens hanging off an extra inch or more of tube away from the mount). Sigma is certainly capable of designing native mirrorless lenses, but they haven't chosen to do so yet (with the exception of a few very compact APS-C/Micro 43 primes that are rarely seen).

The second concern beyond potential lens bulkiness (if they're starting with the Sigma Art plus Leica SL lineup, there's not a small lens in the bunch) is that the SL mount is originally an APS-C mount, and has a relatively small diameter - it's 2mm larger than the Sony E/FE mount, but 6 mm behind Canon and 7 mm behind Nikon. The difference in size between Nikon's new Z mount (the largest of the FF mirrorless mounts, although Canon is not far behind) and the medium format Hasselblad X mount is actually smaller than the difference between the SL mount and the Z mount. Like Sony, they may be trapped by a small APS-C derived mount, with a harder time building fast lenses and very compact lenses.

Neither Sony's nor the Panasonic coalition's decision to use existing mounts made all that much sense (Sony's stuck now, because of all the great FE lenses that exist - but they could have introduced a new mount with the original A7 series - there weren't NEX lenses that people really cared about on the A7). Unless there are a lot more ($5000 each, twice the weight of Canon/Nikon/Sony equivalents) SL lenses out there than I'm aware of, the SL mount also offers little in the way of extra compatibility.

Things are certainly getting interesting - not least because this is going to be a superb video camera! No IBIS is a real pain for that application! 4Kp60 with a data rate high enough that the fastest SD cards can't handle it is the latest rumor - if true, there is nothing short of a RED that does that (the Sony FS7 series comes very close with a smaller sensor).

One important question is how good the high pixel count sensor is in factors other than pixel count. Both the Nikon D850/Z7 sensor lineage and the closely related Sony A7r mkII/III sensor lineage are really excellent. They have very high dynamic range (DxOMark claims over 14 stops at base ISO, which is impossible in a 14 bit file,  while Photons to Photos says around 12 really usable stops, which is believable)  and excellent color. The even higher-resolution Canon 5Ds sensor loses a consistent couple of stops of dynamic range to the Sony/Nikon duo at all ISOs below about 1600, where it catches up. At base ISO, the Canon sensor actually falls behind the best APS-C cameras. Canon has closed the gap somewhat with the sensor in the 5D mk IV and EOS-R, but there's no high-resolution version of that yet. Will the new LeicaSigmaSonic sensor perform more like the Sony/Nikon, or more like the 5DsR? To be credible as an ultra-high resolution still camera, it has to perform like Sony and Nikon

Dan

[Rory]

They have very high dynamic range (DxOMark claims over 14 stops at base ISO, which is impossible in a 14 bit file

Why is this impossible?

[Dan Wells]

Absent any compression (these cameras are all 14-bit uncompressed, and compressed options that some have start with a 14-bit signal), it takes 1 bit of data to represent 1 stop of dynamic range (per pixel, and per color channel). The only theoretical way to get more than 14 stops of signal is to use a 16-bit (or higher) analog to digital converter, which is much more expensive, slower and unnecessary for all but a few medium format sensors.

According to a prior discussion on Luminous, the only current sensor to have a 16-bit data path is the Sony 100 MP "645 full-frame" sensor in a few Phase One backs. Previous medium-format sensors that claimed "16-bit" files were apparently just taking a 14-bit output and adding a couple of zeros on the bottom (you could take an iPhone image with about 7 bits of real data, add 9 zeros and claim it as 16-bit) Since that discussion was last year, the very latest 150 MP sensor in the high-end Phase backs was not around yet, but I would guess it's also 16-bit. Another candidate that we should see soon is the new 100 MP 33x44 mm sensor that will show up in the successors to the Hasselblad H1D and Fuji GFX - that is unclear, because it is much more of a mass-market sensor - sort of. The medium format mirrorless cameras probably sell in the low tens of thousands per year, while the top Phase backs sell in the hundreds or maybe low thousands - for comparison, Nikon can make 240,000 Z7's per year, and I suspect A7rIII's, EOS-R's, etc. are in the same range, as are top-end DSLRs - say a million total cameras at or over 36 MP annually.  Also, no medium-format camera shoots much over 1 FPS - there's time to move the extra data around.

The current Sony 24x36mm sensor seems to be right on the verge of producing more than 14 bits of real, usable data. Measurements of photographic dynamic range (which can underestimate - there are decisions in determining where the signal fades into noise, and Photons to Photos is conservative) show over 12 stops of highly usable dynamic range, and you can stretch another stop in the shadows. A 16-bit converter won't help this generation, but it might be necessary for the next generation.

One possible approach is that the Nikon D3x, the first camera outside of medium format to actually need 14-bit processing (because it could record more than 12 bits of real data) didn't have a true 14-bit converter, but it had some way of reading the analog values several times on the same exposure and running it through the analog to digital converter at different gains. It slowed the camera down by a factor of four, but it got 14-bit data out of a 12-bit converter. I wouldn't be at all surprised to see someone do that with an upcoming sensor. The D3x in slow mode was something like a 1.3 fps camera, but it sure produced a nice file... By the time the D800 came along, Nikon had a true 14-bit converter that ran at a more standard frame rate.

Dan

[Telecaster]

You could, given capable tech, create a sensor with a dynamic range of 20 stops and then run its output into an 8-bit ADC. There's no law that says your converter can't be coarse in its processing of voltages. 

-Dave-

[Dan Wells]

Do you know of any camera that remaps values at the ADC stage? A sort of analog domain compression... Wouldn't adjacent voltages, by definition, look alike, creating banding instead of blowing highlights or blocking shadows. You still have only 256 values per color channel on an 8 bit ADC, you're just stretching them over a wider range, reducing resolution between them... Or am I missing something? I thought all cameras tried to be linear at the ADC stage (then they apply a curve at the JPEG engine, as do all RAW converters - a linear image looks very dark).

Dan

[Telecaster]

I'm not saying 8-bit conversions of 20-stop DR image data would look good.    Tonally they'd be rough: imagine undithered 6-bit conversions from a 14-bit source. (Or have a look if you've got software capable of undithered down-conversion.) In the real world I can't imagine why anyone would do such a thing. But it is technically possible.

-Dave-

[johnvanatta]

Arguably the situation on desktop monitors is something like that--8 bit color isn't really enough for modern wide gamuts, but 10 bit support is still spotty. If you go looking, you'll find banding--but of course most people don't look.

[davidgp]

I'm not saying 8-bit conversions of 20-stop DR image data would look good.    Tonally they'd be rough: imagine undithered 6-bit conversions from a 14-bit source. (Or have a look if you've got software capable of undithered down-conversion.) In the real world I can't imagine why anyone would do such a thing. But it is technically possible.

-Dave-

Dan, what Dave is saying is that DR response of a camera is an analog thing, independent latter on of the precision of the digatilization process you put behind of it: 8 bits, 12 bits, 14 bits, 16 bits...

For example, for video people this is quite normal, they notice when the camera has better DR, 12 stops instead of 14 stops even if the output is 8 bits or 10 bits video...


http://dgpfotografia.com

[Christopher Sanderson]

With the SL lens mount, virtually every Leica M lens ever made will work in manual. This makes the “new FF” camera pretty compelling.

[KLaban]

With the SL lens mount, virtually every Leica M lens ever made will work in manual. This makes the “new FF” camera pretty compelling.

But will they work well with the Panasonic sensor?

[Christopher Sanderson]

But will they work well with the Panasonic sensor?

Well, if the Leica SL lenses work which should be a fairly safe bet, then the M lenses will work. No?

[faberryman]

Well, if the Leica SL lenses work which should be a fairly safe bet, then the M lenses will work. No?

The SL lenses are different optical formulations than the M lenses, where the light rays exit the back of the lens more parallel to the sensor. With the M lenses, particularly the wide angle ones, the light rays exit at steep angles, and unless the sensor microlenses are designed to accept light arriving at such steep angles, as is the case with the digital Ms and the SL, there will be smearing. We don't know whether the Panasonic sensor microlenses will be optimized for M wide angle lenses.

[D Fuller]

Well, if the Leica SL lenses work which should be a fairly safe bet, then the M lenses will work. No?

Depends on what you mean by “work”. They’ll fit on the camera, most certainly, but Leica sensors have their micro lenses turned a bit to mitigate the off-axis smearing that wide lenses designed for film exhibit. It’s one of the reasons why Leica M lenses work better on M cameras. If you trust Sean Reid, better than on the SL, even. The other is that the glass stack is thinner.

[BJL]

While it is possible to fit over 14 stops of DR into 10 or even 8 bit output, that requires nonlinear encoding _after_ ADC: the gamma of 8-bit JPEG or the similar processing with LOG video encoding.

But the ADC itself is giving linear output, and that can only preserve as many stops of DR as the raw ADC output has bits.