What would the Ideal MP be for a FF 1:1:1 Foveon DSLR?

I'm talking real geometric pixel resolution, not the X3 obfuscation

Respondents should consider the following assertion:

"Only few output media like large fine art printers support resolutions in excess of 10 MP."

Found in Section 1.3.1 here:


Those respondents who check 'More' should at least tell us where to find SA mount lenses to match ;-)

[No message]

. . which begs the related question:

Who should Sigma cater to?

Me?

Or someone who prints bill-boards and examines them up close or, equally, someone who shoots bugs from 100 yards and crops down to just the bug?

Between those extremes lies those folks who, for example, print B or C size (A3 or A2) and hang them on a wall and view from the couch.

On an Epson printer, an image looks great at 180 dpi, better at 360 dpi, even better at 720 dpi, etc.

The more information I can provide to a fixed-dimension output space (a print as opposed to a monitor), the better. Anyone who claims otherwise is just settling for "good enough."

The print is all about information: as much information as possible processed correctly.

Punkt!

xpatUSA said:

I'm talking real geometric pixel resolution, not the X3 obfuscation

Respondents should consider the following assertion:

"Only few output media like large fine art printers support resolutions in excess of 10 MP."

Found in Section 1.3.1 here:

http://falklumo.blogspot.com/2012/06/true-reasons-full-frame.html

Those respondents who check 'More' should at least tell us where to find SA mount lenses to match ;-)

Click to expand...

The Quattro layout makes a lot more sense on a FF sensor where 12MP half size is almost as high as a Merrill, yet allowing 48MP for those who are willing to trade speed for even more detail. So I'd like to check two boxes: 12 and 48 ;-)

Johan Borg said:

The Quattro layout makes a lot more sense on a FF sensor where 12MP half size is almost as high as a Merrill, yet allowing 48MP for those who are willing to trade speed for even more detail. So I'd like to check two boxes: 12 and 48 ;-)

Click to expand...

A good point, Johan, which is why I was careful to put 1:1:1 in the title

One thing I've found re: LO res images on SD9 thru DP2M is that, generally speaking, a full res capture down-sampled 50% is potentially 'better' - because modern re-sampling algorithms such as Lanczos et al are better than the simple averaging afforded by on-chip pixel binning.

Since all of the other numbers are one, the numerical constancy part of my would like the megapixels captured to also be one.

Since megapixels do not actually mean anything, the resolution could be 20000 x 20000.

Kendall Helmstetter Gelner said:

Since all of the other numbers are one, the numerical constancy part of [me] would like the megapixels captured to also be one.

Click to expand...

That'll be a vote for 1MP, then? ;-)

Kendall Helmstetter Gelner said:

Since megapixels do not actually mean anything, the resolution could be 20000 x 20000[px].

Click to expand...

As Roland would say:

"Hmmmmm . . . ."

12 Big pixels the bigger the better.

Roger J.

It should be more than 24 Megapixels, since some Foveon critical
people would argue, that even with the Foveon advantage 24 Million
Foveon pixels will not be equivalent to the current 36 Megapixel Bayer
sensors.

However, it should have pixels that are (slightly) larger than the Merrill
pixels in order to minimize noise. (The Quattro pixels exhibit visibly
more noise in most situations except for blue sky...)

Based on these consideration I would suggest:

6480 x 4320 = 27993600 or 28 Megapixels

with pixels that have 23.4% more area than Merrill pixels.

Greetings,

Robert

36 x 24 mm with 4500 x 3000 Pixel ~ 8µm pixel size ~ 13.5MPx

I chose 36 because there was no 39. I think that the sensor in the Merrill cameras could handle slightly more density, and 39 MP makes a lot of sense to me, since reducing the horizontal resolution in half will result in a perfect fit for a 4K screen. It will be a perfect fit in the future for an 8K screen. 4K screens are already here, and I believe that in 5 years the 8K screens will be here. Some day people will look back on the Merrill photos and think they look so low res. Recently I was looking at some old 4 MP photos, wishing I had an 8 MP camera at that time . . . but I didn't get my 8 MP camera until about a year after I shot those photos.

So I believe that if you want to "future-proof" the resolution and stick with that resolution for 5 years or more, the full-frame sensor should be 39 MP, at least on the top layer, if not on all layers.

Roger said:

12 Big pixels the bigger the better.

Click to expand...

Perhaps I should have included ye olde original SD9 pixels!

That would be 10MP to the nearest MP :-D

33MP (P standing for photosite), each photosite consisting of 3 stacked photodiodes (or maybe 7).
Essentially an array of 7680 x 4320, so as to be a multiple of most common displays.
For prints, that gives a width of 19.2" (~49cm) at 400dpi

Chris

I would think it should be some optimal relationship between pixel pitch and lens resolution, which would be based on available lenses and calculated precisely (hopefully). And this should be kept in mind:

http://www.lensrentals.com/blog/2012/02/sensor-size-matters-part-2

since my eyes are not evolving from 4k to 8k that fast for me something between 12MP and 16MP x 3 is just good enough ... ;-)

OrdinarilyInordinate said:

I would think it should be some optimal relationship between pixel pitch and lens resolution, which would be based on available lenses and calculated precisely (hopefully).

Click to expand...

A reasonable view. My liking for lower MP and bigger pixels stems from my motley collection of lenses, plus a tendency to camera-shake, plus I don't print. On an SD9 or SD10 (9.12um pitch) I can actually ignore the effect of diffraction up to f/16 (the diffraction exists but is not resolved by the sensor). Plus, 'small' aberrations of the order of fractional wavelengths are also unresolved, IMHO.

For the majority of folks, R N Clark (see below) has calculated an optimum relationship resulting in a 'best' pixel pitch of about 6um (Bayer system, though). For a Bayer FF camera that would be exactly 24MP - to which a 16MP Foveon would be approximately equivalent. There is, however, a poor correlation between 16MP and the poll results so far.

OrdinarilyInordinate said:

And this should be kept in mind:

http://www.lensrentals.com/blog/2012/02/sensor-size-matters-part-2

Click to expand...

Yes, one of the better articles and well worth a read. Roger Cicala is a member of the DPR fora and is well-respected.

This link from his article is also worth a read.

http://www.clarkvision.com/articles/digital.sensor.performance.summary/index.html

Perhaps a little dated, now that software provides claimed ISOs up to 256 gazillion, but basic principles remain good to my old eyes.

xpatUSA said:

Those respondents who check 'More' should at least tell us where to find SA mount lenses to match ;-)

Click to expand...

I see two votes for 'More' but little to no justification.

Suppose 'More' implies 96MP. That would a pixel size of 3um!

Do head over LensRentals and see what Roger Cicala thinks about that:

http://www.lensrentals.com/blog/2012/02/sensor-size-matters-part-2

For starters, diffraction kicking in at f/5.6; More noise, pardon my general use of the word; Next to no MTF at Nyquist;. Lots of sensor self-heating; You name it; The list is endless.
--
Cheers,
Ted

xpatUSA said:

xpatUSA said:

Those respondents who check 'More' should at least tell us where to find SA mount lenses to match ;-)

Click to expand...

I see two votes for 'More' but little to no justification.

Suppose 'More' implies 96MP. That would a pixel size of 3um!

Do head over LensRentals and see what Roger Cicala thinks about that:

http://www.lensrentals.com/blog/2012/02/sensor-size-matters-part-2

For starters, diffraction kicking in at f/5.6; More noise, pardon my general use of the word; Next to no MTF at Nyquist;. Lots of sensor self-heating; You name it; The list is endless.
--
Cheers,
Ted

Click to expand...

Ted I haven't read what it says in the LensRentals article, but I've read many articles in the past with all sorts of theory about diffraction and such. Usually these articles were written with a slant that seemed like they were justifying an opinion that it didn't make sense to step up from 12 MP to 16 MP on an APS-C sensor or that it was a complete waste to step up from 16 MP to 24 MP with an APS-C sensor. Now we know that is a bunch of B.S., because we have practical experience. What I do know is that there are lots of tiny little 20 MP sensors out there right now, and they are working just fine. One example is the little 1" sensor in the Sony RX100 series cameras. Please tell me how quadrupling such a sensor (in area and photo-sites) would make it beyond reasonable in performance vs. limitations, and if that would NOT be beyond reasonable then why would it not be reasonable to make a full-frame camera with even higher resolution, considering the 1" sensor has a 2.7x crop factor, so it is less than 1/4 the area of a full-frame sensor.

You mention the 3um pixel size, but aren't there plenty of point-and-shoot cameras with smaller photo-sites? I'm talking about the 16 MP cameras with the tiny little 1/2.3" sensors. Don't those cameras perform just fine at f5.6? One that I'm thinking about is the Nikon AW200, which I checked out at length, comparing the photos to other samples I've seen, and being quite impressed with its 16 MP quality (for such a small sensor and tiny 5x zoom lens).

I'm not suggesting that I think we need a sensor that is more than 39 MP. If you read my entry in this thread above, you know why I think that would be a good number, but I really don't think that in the future we will see sensors NEVER step up to 100 MP. I think 100 MP is coming . . . at least in full-frame sensors, if not in APS-C too. The reason I believe we will see this some day is that oversampling is actually a good thing, and with diffraction-limited lenses at f2.8, the best lenses really will be able to resolve that much detail or close to it.

Right now our computers are a limitation. Hard drive space is so cheap now though that hard drive space is NOT a problem. In five years our computer processors will be so fast that 100 MP digital raw files even at 16 bits per color will not be a problem. With UHS-2 the bottle-neck of the memory card and card slot speeds are not an issue anymore either, and those cards will be gradually adopted until they are mainstream in five years. I think that ultimately we will see a quadrupling of image processing speed over the next few years, even if we see our sensor resolution double. And then it will happen again, and we will have 100 MP sensors. If Sigma is a little before their time with their full-frame Quattro sensor, I don't think that would be a really huge problem. But like I said, I don't think it's warranted . . . not yet anyway.

I would say 36mp, as I see nothing wrong with the pixel size of the Merrill. However maybe 24mp would make the camera usable above ISO 400? Also that's a good size image and it would be more forgiving on lenses and technique. So I voted 24.