Sensor Pixel density question

[aztwang]

As I look at all the sensors out there, they come in all kinds of combinations; 35mm with 36mp, the new Nikon 45.6mp on the D850, MFD full frame 60,80 and 100mp and crop frame MFD in 50mp....and there's more.
Given these few I have mentioned can ya'll clarify in lay-mans terms what the advantage or dis-advantage of pixel size has on image quality, dynamic range and ??...The IQ350 has 50mp on 1452 mm of sensor real estate. Pixel size is 5.3x5.3 microns. Phase's iq3100 has 100mp on 2160 mm of sensor real estate and pixel size measures at 4.6x4.6 microns. The D850 sensor has 864mm of real estate and pixel size is 4.35 micros. Then I read a bit about fat pixel backs, some sensor pixels over 9 microns! Any help in understanding the above is much appreciated...Cheers...Don

 

[jerome_m]

can ya'll clarify in lay-mans terms what the advantage or dis-advantage of pixel size has on image quality, dynamic range and ??...

In lay-mans terms: no, not really. There are various reasons why larger pixels will bring some advantages to smaller ones of the same technology, but they are somewhat complex. Furthermore, because pixel size generally decreases when technology advances, it is difficult to find two devices of the same era with different pixel sizes. Last but not lease, you won't use the same lenses on different sensors and lenses are essential.

 

[Paul2660]

As I have followed the tech over the years, the main issues I see with a smaller pixel size are:

Diffraction problems and low light capability or gathering.

The issue of diffraction has been pretty well documented, and with the IQ3100 it's quite noticeable past F11 for me. I realize it's showing sooner than F11 but most of the work I do I can handle what diffraction I get in C1 or LR or with other sharpening tools. In fact I often use F14 to F16 at times as the F8 to F11 range offers a much shallower DOF.

Low light gathering, smaller the pixel the less light a single pixel can receive. Of course you can use a microlens to assist, and thus we see the other issues that the larger pixel backs have, color shift and cross talk with a tech camera. The new tech of BSI (back side illumination) that Sony brought out with the A7RII may help on this issue in time. Personally I was not impressed with the A7RII and noise improvement due to BSI, but others have been.

When the next round of MF sensors comes out it's been written they will be 150MP, which will be 50% more dense than the current sensor, thus diffraction will start even sooner. (unless P1 uses a physically larger sensor). I am assuming that the 50MP back will upgrade also, to what size not sure.

The Canon 5ds and new Nikon D850 both have increased issues due to diffraction and if Sony comes out with an even larger pixel count 35mm sized chip, they will again have increased diffraction issues to work out, and light gathering.

It's important to go back to the Sony A7sII, and look at just how good the files look, they are clean. Cleaner than anything I have seen from digital, but they are only 12MP, and thus your print size IMO is very very limited. Here is what I feel is the single greatest trade off, as no software has been created that can interpolate a 12MP image to a native 24MP or 39MP capture and still hold the same details. Years ago, I figured this would happen, but the cost of large pixel cameras has dropped so much that I guess there is no longer any interest.

Paul Caldwell

 

[mjr]

As a dumb ***, I have absolutely no understanding or interest really in things like pixel density, I have had fat pixels, thin pixels, rubbish ccd pixels and super amazing cmos pixels, I think there are so many variables from lenses to processing techniques to how I expose and what I point it at, that the chances of an image I take being better or worse based on the density of the pixels is unlikely. I have read people say that smaller is best, some say bigger is best, I just don't see the point of worrying about it! I'm sure it's important and even fascinating to some but what the hell, I just can't be bothered. I think if someone looked at an image and said, nice, what density were the pixels used in that shot, I'd probably give up!

Mat

 

[Godfrey]

As I look at all the sensors out there, they come in all kinds of combinations; 35mm with 36mp, the new Nikon 45.6mp on the D850, MFD full frame 60,80 and 100mp and crop frame MFD in 50mp....and there's more.
Given these few I have mentioned can ya'll clarify in lay-mans terms what the advantage or dis-advantage of pixel size has on image quality, dynamic range and ??...The IQ350 has 50mp on 1452 mm of sensor real estate. Pixel size is 5.3x5.3 microns. Phase's iq3100 has 100mp on 2160 mm of sensor real estate and pixel size measures at 4.6x4.6 microns. The D850 sensor has 864mm of real estate and pixel size is 4.35 micros. Then I read a bit about fat pixel backs, some sensor pixels over 9 microns! Any help in understanding the above is much appreciated...Cheers...Don

Consider a pixel to be like a sail: the more unit area a pixel has, the more energy it can register with the least pixel-to-pixel cross-talk in the sensor in the same amount of time. That can be expressed as more sensitivity or more dynamic range, depending on a lot of other factors in the design of the total system. You need a certain number of pixels to obtain adequate resolution, you need a certain amount of dynamic range to obtain adequate tonal gradation, and you need a certain amount of sensitivity free from noise to obtain adequate speed in use for given subject matter. So you pick what format, what pixel density, etc, to suit those three variable factors and provide the right range of sensitivity, the right dynamic range, and the right resolution for your intended output products.

That's as complicated an explanation as I generally like to go with on this topic; it's a decent first order approximation. In practice, you find a camera with a sensor set up to be a good match to your lenses and accept whatever its advantages are while working around whatever deficiencies it has. Worrying about the particulars of pixel pitch and format too much can be counterproductive to the work of learning to see and making great photographs. Most of what's available in digital capture today can produce very very high quality results...

G

 

[Shashin]

There are too many variables to that question. The only answer is with "equal" technology, which of course is hard to find. However, one area is more constant which is image contrast. Fat-pixels sensors are limited to lower frequency detail, which means that contrast can be higher. Lenses can be optimized for resolution or contrast, but not both at the same time. The greater the resolution demands, the more a lens will be optimized for that, losing contrast. Did you ever have an old film lens you really liked, but then put on a high-resolution sensor which made it appear soft? It probably had really nice contrast, but people are nervous when they pixel peep, even though it would probably not been seen at normal viewing distances (which is also the problem of people's perception to diffraction--no one can tell which images I shoot at f/16 or f/22, but I am constantly told those apertures are too small. Sensors too are part of this resolution/contrast tradeoff. Which is why people still find something special about those 9um backs.