High Megapixel Camera Development Announcement Mentioned Again [CR1]

[l_d_allan]

I'd be more interested in higher DR value than more mega-pixels.

I'd be pleased in Canon integrates the Sony Exmor 36 mpix sensor. Canon was willing to use the Sony 1" sensor in the recently announced G7x.

However, I'd think that use of high mega-pixels would tend to slow the FPS spec down, perhaps significantly. Or not?

 

[Maiaibing]

Not surprisingly you're of course guessing.

?? I made clear on what I was guessing - to what extent it was either D800 or possibly other Nikon models helped Nikon land OK sales 2013/14) the rest are from Nikon/Canon financial releases - which therefore should be painfully accurate - and a statement from Nikon that D800 sales where very good 2013/14.

 

[OnteoEOS]

I do not mind the MP increase, 22MP are ok for me and the 5DIII is giving me great service.

I would like to see some nice ISO 32 as a base for sensors. Landscape and studio photographers would be very happy with a camera in this range:

ISO from 32 to 6400 in a FF sensor
50MP with high dynamic range and dual pixel AF
4FPS in full resolution and 7FPS in MRAW
WIFI and direct upload plugins to specific dropbox
Dual cards
GPS
4" articulate screen with tap to zoom capabilities
USB 3
Pixel shifting mode for even higher resolution pictures
Capability to store pictures directly over wifi in case CF/SD card breaks
Shutter from 60s to 1/16000s
4K RAW video with pixel binning
Capability to input editing profiles from adobe raw straight into the camera's OS

And many more things...

 

[bdunbar79]

Not surprisingly you're of course guessing.

?? I made clear on what I was guessing - to what extent it was either D800 or possibly other Nikon models helped Nikon land OK sales 2013/14) the rest are from Nikon/Canon financial releases - which therefore should be painfully accurate - and a statement from Nikon that D800 sales where very good 2013/14.

Ok sorry. You're right.

I guess what I was saying is that you have no idea (or me) how well the D800 did on the market. We know how it did vs. the 5D3 on amazon. That's all. The statement from Nikon regarding the D800 doesn't really mean anything at all, IMO.

I'm just not convinced I guess that Canon's lacking a high MP body is going to push away enough people to hurt them, when most people don't want it. Some do, most don't. I do.

 

[dadgummit]

My biggest issue with much higher MP is if you take the current crop sensor at 20mp and multiply it by 1.6 squared (20 x 1.6 x 1.6) you get 51mp sensor. This means a 46 + mp sensor will have a similar pixel density to their current crop.

Would this sensor have the same noise that all of the FF owners are complaining about when they see the 7D2 samples?

 

[Khalai]

My biggest issue with much higher MP is if you take the current crop sensor at 20mp and multiply it by 1.6 squared (20 x 1.6 x 1.6) you get 51mp sensor. This means a 46 + mp sensor will have a similar pixel density to their current crop.

Would this sensor have the same noise that all of the FF owners are complaining about when they see the 7D2 samples?

I'd be more worried about diffraction as well. At f/11 it's already kicking in on the crop and f/16+ is barely usable at all.

 

[keriboi]

http://www.seeimpossible.usa.canon.com/

 

[spinola]

41:09:20!!!!!!!!!!!!!!!!!!!!!!!!!!!
41:09:19!!!!!!!!!!!!!!!!
41:09:18!!!!!!!!!!

 

[Marsu42]

Would this sensor have the same noise that all of the FF owners are complaining about when they see the 7D2 samples?

Sure, and that would be the reason why Canon didn't upscale their crop sensors yet. Of course that's only at 100% magnification, and if you downsample and denoise to 20mp crop the ff sensor would still have a large edge - but color fidelity and postprocessing elasticity would still suffer.

I'd be more worried about diffraction as well. At f/11 it's already kicking in on the crop and f/16+ is barely usable at all.

True and this is a huge issue for macro, but aren't most people with ff sensors rather using smaller apertures than f11? The one problem I can imagine would be "infinite dof" landscape, but then there's focus stacking... how do d800/d810 owners deal with diffraction?

 

[heptagon]

Would this sensor have the same noise that all of the FF owners are complaining about when they see the 7D2 samples?

Sure, and that would be the reason why Canon didn't upscale their crop sensors yet. Of course that's only at 100% magnification, and if you downsample and denoise to 20mp crop the ff sensor would still have a large edge - but color fidelity and postprocessing elasticity would still suffer.

I'd be more worried about diffraction as well. At f/11 it's already kicking in on the crop and f/16+ is barely usable at all.

True and this is a huge issue for macro, but aren't most people with ff sensors rather using smaller apertures than f11? The one problem I can imagine would be "infinite dof" landscape, but then there's focus stacking... how do d800/d810 owners deal with diffraction?

Diffraction is not as bad as it sounds. Since it's distributed evenly across the image it's easier to sharpen in postprocessing than a bad lens. There are a few examples around that show that with sharpening a f/16 image can look better than a f/8 image. Since diffraction has a soft onsetting i think that a sensor with a diffraction limit of f/5.6 can easily be used at f/11-f/16 without a high disadvantage after sharpening. And we're far away from diffraction limit at f/2.8 on full frame sensors. (Phone cameras are different though and they make good use of those tiny pixels they have.)

 

[jrista]

Using a f/11 on a sensor with an f/5.6 DLA is no different than using it on a sensor with an f/8 DLA. Diffraction cannot make a higher resolution sensor perform worse than a lower resolution sensor. At WORST, they perform the same, on average, a higher resolution sensor will still outperform a lower resolution sensor at smaller diffraction-limited apertures.

The difference is simply that the lower resolution sensor doesn't "see" diffraction that the higher resolution sensor can. Remember, diffraction is always present, a single point of light is always going to be an airy disk at the sensor plane...as the lens is stopped down, the size of that airy disk increases. It's the same size for both sensors at any given aperture....it's just that the sensor with f/5.6 DLA is "seeing" smaller airy disks than the sensor with f/8 DLA.

 

[LetTheRightLensIn]

I realize this is all speculation and may come to nothing. But if there is a high MP that costs $8000, it is hard to imagine why someone would not just fork over $9000 and get a medium format Pentax. Or spend $3000 for a Nikon D810 (which at $5000 less would surely cover the cost of replacing lenses). Still, this is an unsubstantiated rumor and may come to nothing.

yeah, as I've said, if it will be 1Ds price and form factor it's gotta include the kitchen sink, drive 36+ MP at a fairly great frame rate as in like 8-12fps, have top quality 4k and 2k video and even the 4k can be internally recorded and 10bits 4:2:2, have exmor DR, etc.

otherwise you could buy A7S+A7R+Ninja Shogun+D810 and a lens for like the same price or a D810 and a bunch of Nikon lenses and keep your current Canon stuff and be no worse off than buying this crazy 9k 1DsX or an A7R+adapter for a ton less. So the 1DsX would have to have really out of this world specs.

 

[LetTheRightLensIn]

The real question, is what level/model of the Canon line the higher MP sensor will be in! My guess, is that it could show up in both the 1-series and the 5-series bodies. However, I wouldn't rule out it being in a totally new model!

The rumor above and my personal guess is that Canon isn't in a hurry to deliver high mp cameras to the masses, or they would have done so by now. Their consumer tech advantage is dual pixel af, meaning they'll put more metapixies in their very high-end gear first, and until this trickles down to the rest of us it will be about 2020.

Not a big issue for me since for me since for what I do 20mp is enough. Not that I wouldn't like more mp for macro, but it's not that vital. But if you are set upon more resolution, a *rumor* about more mp means you either have to be rich, have to switch brands or have to wait for a long time.

If 2020 is their plan for better DR (and more MP) count me gone. I'll pick up the A7R soon and then maybe start getting A7S type stuff and whatnot and maybe even Nikon but won't spend more on Canon stuff at this point. Although i'll keep the 5D3 for regular stuff and the canon lenses for now at least.
But if it may be some time before Canon ever gets more money from me. I'll just start by putting my money to the A7R now and then slowly more and more to Sony and perhaps Nikon too if Canon really looks hopeless and I tire of the mixed 5D3+Sony mess.

 

[LetTheRightLensIn]

My biggest issue with much higher MP is if you take the current crop sensor at 20mp and multiply it by 1.6 squared (20 x 1.6 x 1.6) you get 51mp sensor. This means a 46 + mp sensor will have a similar pixel density to their current crop.

Would this sensor have the same noise that all of the FF owners are complaining about when they see the 7D2 samples?

I'd be more worried about diffraction as well. At f/11 it's already kicking in on the crop and f/16+ is barely usable at all.

Lots of nature shots can get by with f/8, some even f/6.3 and a great many at f/11. And when you use more MP for reach, lots of wildlife/sports stuff can be shot f/2.8-f/5.6 (although a few things need a lot more, like if you had to have two large birds both in focus at once or something).

 

[Marsu42]

The difference is simply that the lower resolution sensor doesn't "see" diffraction that the higher resolution sensor can.

Sure, I guess we understand this (did we?). However, it is a valid consideration - not to dump high mp sensors and say "18mp is enough", but to watch out for. At least for macro, I have experienced myself that using small apertures has a very bad effect on the sharpness, and that's what you're after with macro. So high-mp macro will mean more focus stacking, and with smaller steps.

 

[jrista]

The difference is simply that the lower resolution sensor doesn't "see" diffraction that the higher resolution sensor can.

Sure, I guess we understand this (did we?). However, it is a valid consideration - not to dump high mp sensors and say "18mp is enough", but to watch out for. At least for macro, I have experienced myself that using small apertures has a very bad effect on the sharpness, and that's what you're after with macro. So high-mp macro will mean more focus stacking, and with smaller steps.

Not really. You can always downsample the higher resolution macro to the same image dimensions as a lower resolution macro. The end result would be increased sharpness and finer resolved detail. Think about it...there is no way that smaller pixels could ever be bad (all else being equal). At worst, they would only be as bad as larger pixels...but in no way can they ever really be worse (assuming proper use of an AA filter or equivalent functionality...it's possible to overdo anti-aliazing, such as with Fuji's X-Trans sensor...overdo AA, and you throw away useful information and diminish or eliminate the value of having more pixels.)

Remember that to have an effective comparison, you have to normalize. On a normalized basis, higher resolution sensors are still resolving more detail at native size, even if that does not seem to be the case when pixel peeping at 100% native size...once you do downsample, the difference in resolving power and sharpness should pop out at you like night and day (especially if were talking about something like the 5D III vs. D810/A7r, where the resolution difference is fairly significant.)

 

[Marsu42]

You can always downsample the higher resolution macro to the same image dimensions as a lower resolution macro. The end result would be increased sharpness and finer resolved detail.

Even if your lens has an aperture setting that introduces visible diffraction? I imagine downsampling this would only result in digital "semi"-detail as a simply algorithm doesn't try to reconstruct data that isn't there.

Think about it...there is no way that smaller pixels could ever be bad (all else being equal). At worst, they would only be as bad as larger pixels...but in no way can they ever really be worse (assuming proper use of an AA filter or equivalent functionality...

Of course, and I'm no enemy of higher mp sensors. However, if for some purpose the best case is being on par with a lower resolution sensor, it probably isn't worth the hassle (and price...) for these applications.

Remember that to have an effective comparison, you have to normalize. On a normalized basis, higher resolution sensors are still resolving more detail at native size, even if that does not seem to be the case when pixel peeping at 100% native size...once you do downsample, the difference in resolving power and sharpness should pop out at you like night and day

This is indeed a good point, and that's why I'm happy with my "not-so-sharp wide open" 70-300L. If I downsample the results to usual print/view sizes the results are very good, so no real need there for a much heavier and more expensive lens that excels at 100% magnification "pixel peeping".

 

[jrista]

You can always downsample the higher resolution macro to the same image dimensions as a lower resolution macro. The end result would be increased sharpness and finer resolved detail.

Even if your lens has an aperture setting that introduces visible diffraction? I imagine downsampling this would only result in digital "semi"-detail as a simply algorithm doesn't try to reconstruct data that isn't there.

When you reach hevily diffraction limited apertures, the detail is not there anyway. When your at levels of diffraction where the spot size is between the pitch of smaller vs. larger pixels, then smaller pixels CAN pick up more detail. It's actually possible for them to resolve more even when the diffraction spot is larger than the larger pixels. It's a progression of diminishing returns, eventually (say f/32) you get to the point where the differences are so small as to be meaningless. It's not that detail is there, then suddenly gone, the moment diffraction hits. This is a common misunderstanding (although I'm not sure your making it...just for the edification of everyone here).

If you have 10µm and 5µm pixels, and your diffraction spot is 6µm in diameter. The 5µm pixels are going to "see" that diffraction spot...the sensor is diffraction limited, however your not losing resolution relative to a sensor with larger pixels because of it. However, as far as the 10µm pixels are concerned, it still just sees a spot of light smaller than it can resolve. The same would be true for a 7µm, 8µm, 9µm and 10µm diffraction spot. The 5µm sensor pixels can be rather thoroughly diffraction limited, and they will STILL be resolving more detail than the 10µm sensor pixels. That's because the 10µm pixels simply cannot see diffraction spots smaller than themselves.

The data IS there...it may be starting to get softened by diffraction on the smaller pixel sensor, but that doesn't change the fact that until your diffraction spot is larger than 11µm, your smaller pixels are still resolving more detail. Even at 11µm, the smaller pixels will still be resolving a rounder spot than the larger pixels (which will be resolving a rather blocky spot).

Diffraction doesn't reduce resolution. Diffraction LIMITS resolution. A higher resolution sensor is a higher resoluti on sensor. The resolution of an image created with a higher resolution sensor can be limited by diffraction...but it cannot be reduced by diffraction.

I don't know if that helps. This is one of those things that I have explained to people so many times over the years, and I guess it's just a difficult concept.

 

[jeffa4444]

Jrista last post is bang on the money.
The limiting factor I feel of cameras (as opposed to lenses) is sensitivity particularly at the higher end. On a normally lit summers day most cameras struggle without the use of ND grads and or polarisers. Dynamic range and the reproduction of colors within that dynamic range as our eye sees them (this does vary with age and eye acuteness). Our eyes are able to adjust to reflected light better than any sensor and its in the area of dynamic range & color reproduction that cameras have still a way to go. I read a very interesting article on this subject and the Optometrist that wrote it said she had over her years of research continuosly upgraded her view of our visions ability to seperate colors and it was between 20-22 stops based on camera ISO in terms of sensitivity although this diminished with age.

 

[Marsu42]

The data IS there...it may be starting to get softened by diffraction on the smaller pixel sensor, but that doesn't change the fact that until your diffraction spot is larger than 11µm, your smaller pixels are still resolving more detail. Even at 11µm, the smaller pixels will still be resolving a rounder spot than the larger pixels (which will be resolving a rather blocky spot).

Thanks for the explanation, I guess these analog (light) to digital (data) concepts aren't self-explanatory and that's why a lot of people are confused about how the conversion works, including me