[reh321]

So at what point does a FF have the same size pixels or pixel spacing or pixel pitch or whatever, as an APS-C? 50MP? 60MP? What I am asking is if this FF advantage for SNR is lost when the pixel size becomes the same? Which I think for 24MP APS-C is about a 55MP FF.

When the 50mp Canons were announced, I offended some Canon fans by saying my back-of-envelope calculations showed their sensor pitch was in APS-C territory - and when they did come out, they turned out to be weak in low-light situations. I have commented on several occasions that we cling to the "one best camera" illusion; I doubt if anyone would have claimed that one camera / film combination was best for every situation. We are probably at the point where a person will have to tailor his/her camera purchase(s) based on what tasks are most important to him/her. For example, an MILC will be the preferred camera for situations where quiet is paramount - such as photography during performances, ceremonies, etc.

[osv]

this is an extreme iso12800 noise comparison, showing what rondec was talking about, k3ii vs. a7rii vs. 645z vs. 5dsr

they are all downrezzed to the same print size, so the more mp you have to work with, the cleaner the picture will be... i think that you can see that with the k3ii pixel shift "normal" down arrow?? choose "pixel shift" and watch what happens to the k3ii noise artifacting... pretty impressive i think, mp matters.

however, the canon 50mp 5dsr gets killed by the 50mp 645z, and the 42mp a7rii is somewhere in between.

Studio shot comparison: Digital Photography Review

[normhead]

this is an extreme iso12800 noise comparison, showing what rondec was talking about, k3ii vs. a7rii vs. 645z vs. 5dsr

they are all downrezzed to the same print size, so the more mp you have to work with, the cleaner the picture will be... i think that you can see that with the k3ii pixel shift "normal" down arrow?? choose "pixel shift" and watch what happens to the k3ii noise artifacting... pretty impressive i think, mp matters.

however, the canon 50mp 5dsr gets killed by the 50mp 645z, and the 42mp a7rii is somewhere in between.

Studio shot comparison: Digital Photography Review

Which all goes to show, there's no free lunch.

[derekkite]

Isn't the deterioration as density increases a limitation of the technology and fabrication? The back lit sensors try to move the circuitry behind the wells as opposed to between, presumably to allow more wells and more light hitting the wells. Give it a few iterations to improve the fabrication consistency and a marginal improvement from slightly less noise and more signal will show up. There is also the exigency of fast readout for video which becomes the design parameter as opposed to low noise and high dr.

The 645z sensor seems to have hit the sweet spot of density and size. Big lenses letting in lots of light onto a big surface requiring less manipulation and bending.

It would be interesting to compare the lenses transmissibility of light in these discussions. I wonder if the characteristic of full frame vs apsc to transmit light densities to a sensor surface is different? Size really does matter.

[biz-engineer]

It would be interesting to compare the lenses transmissibility of light in these discussions. I wonder if the characteristic of full frame vs apsc to transmit light densities to a sensor surface is different? Size really does matter.

- Lens coating plays on a few percent of light transmission improvement
- Sensor improvement within similar techno plays within a +-20% range improvement (BSI or ISO tweak)
- Sensor format change from u4/3, APSC, FF, MF plays with an order of magnitude of 100% per 1.5x increase in size.
- Lens one stop wider aperture provides 100% more light.

So, light transmission could well be that tree that hides the forest: you look so close to the tree that you don't see that the tree hides the forest (I'm not criticizing, it's just an image). Typically, I can see a number of people having purchased one of the Sony A7 for its advertised sensor perf. without considering a system, ending up with the best FF sensor in the world but a lot of use case limitations. Canon gets criticized because of more noise/less DR in their sensors, but the wide choice of lenses make the system more valuable.

In other words, there is not way that the big marketing highlights of tiny improvements or gadget features (such as the difference between SMC and HD coatings...) can match the photographic capabilities of a larger format (even a 6 years old camera) coupled with a one stop faster lens (even a 15 years old lens).

[Rondec]

Yes, I understand that. So at what point does a FF have the same size pixels or pixel spacing or pixel pitch or whatever, as an APS-C? 50MP? 60MP? What I am asking is if this FF advantage for SNR is lost when the pixel size becomes the same? Which I think for 24MP APS-C is about a 55MP FF.

OK, I guess that makes sense. Sort of. But, still, it is the same number of pixels, right? Is the printer going to print the FF pixels bigger than the APS-C pixels? I thought a pixel was a pixel, a fixed dot with color information telling the printer what ink to put there.

I don't know, maybe this stuff is just too far over my head. Sorry to be so dense.

From what I can tell, a megapixel is a megapixel at low iso. The problem is that once you get away from iso 100 or 80 or whatever is the lowest iso, you start having noise enter in to your image and with APS-C cameras, you will see that noise quicker. As Ichabod Crane said, a bigger sensor doesn't have to be magnified as much as a whole, in order to create the same size print. Even a Canon 50 megapixel full frame has better SNR than current APS-C cameras, although it isn't as good at dynamic range at low isos.

The real problem I see with sensors with a high density of pixels is that it really does become more difficult to get a pixel sharp image. It is like when people moved from a K10 to a K5 and they started complaining about how their images were soft. You needed better technique to get "sharp" photos.

[normhead]

Isn't the deterioration as density increases a limitation of the technology and fabrication? The back lit sensors try to move the circuitry behind the wells as opposed to between, presumably to allow more wells and more light hitting the wells. Give it a few iterations to improve the fabrication consistency and a marginal improvement from slightly less noise and more signal will show up. There is also the exigency of fast readout for video which becomes the design parameter as opposed to low noise and high dr.

The 645z sensor seems to have hit the sweet spot of density and size. Big lenses letting in lots of light onto a big surface requiring less manipulation and bending.

It would be interesting to compare the lenses transmissibility of light in these discussions. I wonder if the characteristic of full frame vs apsc to transmit light densities to a sensor surface is different? Size really does matter.

Don't even suggest that designing sensors for video would harm still performance. The thought train around here is that all you need for good video is for the lazy software engineers to stop throw paper balls into the waste basket and write some software. It doesn't cost a cent. And a noise performance hit for sensors designed for better video... come on, that's simply a technical impossibility. It's been stated over and over again, it's actually cheaper and better to produce cameras with video than it would be to do it without.

Have you actually even been reading the forum?

Heresy like that can get you burned at the stake.

The first rule of discussing video... never suggest there is a cost to video, and that still photographers are paying for it. Subsidizing those who want it.

I understand, no one wants to be thought of as a leech, but isn't it better to say thanks for buying cameras with video guys, subsidizing my demands, instead of pretending the costs don't exist?

It's the old, the best defence is an good offensive, and I gotta say, video adherents are truly offensive.

[Kunzite]

Normhead, I agree with everything you said - in the spirit in which it was said - except for the " it's actually cheaper and better to produce cameras with video than it would be to do it without" part, which is more often true than not.
There's no general rule, but a DSLR with consumer-level video could be sold in larger volumes than one without; when even Leica introduced video, an ILC without would be seen as 'lacking'. And not implementing video - while the sensor and the processor is video capable, software components developed for other models - doesn't lead to big cost savings. There's already live view in there, right?
OTOH, a professional-level video has requirement not met by the usual stills camera. I don't think we'll see a 4K-capable video monster. I really doubt it would be a 4K RAW video capable monster.

[normhead]

But can it be sold in larger numbers than say a $50 drop in price would sell. I'm going to have to see some hard numbers before I buy that one. Usually in economics, lower price = higher sales. MY guess is, the Pentax approach of putting the minimum into video and keeping costs down is the more effective strategy. And I'm sure they only did that because Canon made a big deal of it. I'm simply not going with the notion that people are buying high end still cameras for their video capability. It's unsupported by any data I know of.

[jatrax]

From what I can tell, a megapixel is a megapixel at low iso. The problem is that once you get away from iso 100 or 80 or whatever is the lowest iso, you start having noise enter in to your image and with APS-C cameras, you will see that noise quicker. As Ichabod Crane said, a bigger sensor doesn't have to be magnified as much as a whole, in order to create the same size print. Even a Canon 50 megapixel full frame has better SNR than current APS-C cameras, although it isn't as good at dynamic range at low isos.

Thanks for trying to explain all this. Not sure I get it yet, but fortunately I don't have to in order to make nice pictures Maybe I'll just stick with making pictures instead of trying to understand how it works. So bottom line: APS-C 'better' than m4/3, FF "better" than APS-C, and 645z kicks everybody else into the turf. Got it.

[derekkite]

But can it be sold in larger numbers than say a $50 drop in price would sell. I'm going to have to see some hard numbers before I buy that one. Usually in economics, lower price = higher sales. MY guess is, the Pentax approach of putting the minimum into video and keeping costs down is the more effective strategy. And I'm sure they only did that because Canon made a big deal of it. I'm simply not going with the notion that people are buying high end still cameras for their video capability. It's unsupported by any data I know of.

It would be more complicated than that. It isn't simply a matter of not having the functionality, it would be designing a sensor chip without video in numbers large enough for economies of scale, and with the advantages that could get you for stills. I suspect that would be a very expensive failure.

I'm certain that Pentax is looking at this, if the 645z had a workable video mode, or even a very good video mode how large would that market be? Vs the additional cost.

Video is a fact of life. I was in a camera shop in Kelowna a few weeks ago; they had been struggling for a long time and seemed to have turned it around. What was remarkable is how full the store was of video accessories for lighting and other aspects. The stills stuff was there but the bread and butter was video.

[IchabodCrane]

But can it be sold in larger numbers than say a $50 drop in price would sell. I'm going to have to see some hard numbers before I buy that one. Usually in economics, lower price = higher sales. MY guess is, the Pentax approach of putting the minimum into video and keeping costs down is the more effective strategy. And I'm sure they only did that because Canon made a big deal of it. I'm simply not going with the notion that people are buying high end still cameras for their video capability. It's unsupported by any data I know of.

I dunno Norm. My guess is video capability is the reason m4/3 buyers who chose Panasonic instead of Olympus did so.

[normhead]

I dunno Norm. My guess is video capability is the reason m4/3 buyers who chose Panasonic instead of Olympus did so.

Hmmmm... well that sounds solid. I used to, and am currently looking a a Panasonic Lumix as opposed to an Olympus, strictly on the photo quality, zoom ratio, the fact that Panasonic has always been the best at reducing shutter lag, and user ratings... but I guess video could be a reason. These guesses usually amount to the the same thing as anecdotal evidence in medicine. They might be on to something, but usually they aren't.

Sorry, I just don't like people's guesses. Especially when I could have said the exact opposite, based on my research.

[Kunzite]

But can it be sold in larger numbers than say a $50 drop in price would sell. I'm going to have to see some hard numbers before I buy that one.

But if you can sell the same quantity without reducing the price?
A mere $50 drop for a camera sold let's say 100,000 units per year means $5,000,000; the amount includes taxes, distribution expenses and distribution network's profit but still, it's non-negligible.
And I think you underestimated the number

[RonHendriks1966]

But if you can sell the same quantity without reducing the price?
A mere $50 drop for a camera sold let's say 100,000 units per year means $5,000,000; the amount includes taxes, distribution expenses and distribution network's profit but still, it's non-negligible.
And I think you underestimated the number

Well the 100.000 units dslr might be the number for Pentax to sell this year.