MP's; Detail & Contrast

There's been a trend towards high magapixel sensor which record huge files for quite some time. I guess there's nothing wrong with that, but there are certainly negatives to it.

There's the obvious penalty of large files which require more storage space and processing power.

There's the less obvious penalty of smaller pixels being less sensitive to light and creating images with more noise. This is less obvious because when you print at a smaller size you compress all of that data and the noise is minimized.

There's the almost unknown fact that the number of MP's recorded has as much to do with the number of MP's projected onto the sensor as it does with the number of pixels on the sensor. A point of light will have a certain size on the sensor based solely on the aperture. Smaller apertures create larger spots on the sensor. For example the green/yellow wavelength (the most important) will project ~ 16MP of detail for a FF set to f/11, and 29MP at f/8 from a perfect lens (Nikon doesn't make a perfect (truly diffraction limited) lens that I'm aware of). That's equivalent to 7 and 13MP for an APS sensor (the smaller sensor size will fit fewer spots onto it). And that's regardless of how many MP's you use to record those dots. For a lot more detail on this topic you can read THIS ARTICLE.

Then there's the fact that smaller pixels require a lens with greater resolution/ MTF. This is also somewhat negated by the oversampling/smaller printing *IF* the image is used full size (i.e. not cropped). This is the reason that smaller pixels reduce contrast. But if you put more smaller pixels on the same sized sensor they get compressed for the same size print. So, in the end the larger sensor always delivers more contrast (because for any given number of pixels they will always be larger in size on the larger sensor). 

Here are three images taken with the same lens (400mm f/2.8 +2x TC) using three different bodies. I used the Nikon D4 a 16MP FF camera (the largest pixel size), the Nikon D800 a 36MP FF camera, and the Nikon V2 a 14MP 1" sensor (2.7x crop factor and the smallest pixels). The bird was about 90yds away. These images all had the same basic processing done to them in LR.




You can click on each and view it at 1024px wide, the size I output from LR. Can you tell the difference?


The first image is from the V2 14MP, uncropped. Due to the pixels size I used ISO250 and f/7.1 (trying to balance lens sharpness, diffraction, detail, noise) which gave me a SS of 1/250...for an effective focal length of almost 2200mm, WAAAY too slow. This image has no cropping due to the effective FL.

The second image is the D800 36MP. Also due to the pixel size I restricted the ISO to 1600 for noise and stepped up to f/8 for lens sharpness which gave me a SS of 1/250 for an 800mm lens...Still way too slow. Image cropped to match the V2 image. 

The last image is the D4 16MP. Here I used ISO 3600 (could have used 6400 if needed), f/8 for lens sharpness, and that gave me a SS of 1/800...way more acceptable. This image is cropped the most heavily. 

To be honest, they are all pretty bad at full size, 100% view prior to editing/output. And they would all look the same comparatively when viewed at any size. Of the three I would choose to use the D4. It gave me the most SS and I still had room left on usable ISO. The D800 image is arguably slightly better but getting consistent results at such slow SS's is VERY difficult. And I'm certain I could bring the D4 image up to match.


The point of this is that, just like almost everything else in photography, selecting a sensor size and MP count is a compromise. There is no "best" and there is no "one size fit's all" answer.


And keep in mind that the COC standard for sharpness only requires less than 2 MP of data... That standard is quite old and arguably way underestimated; some say you need 12-14MP max. But beyond that you are gaining detail and not really "sharpness." It's kind of like a drawing done with a fine pen (or newspaper dot print) vs a drawing made with a bold marker. When you look closely at the fine line image (or dot print) there is more detail and fine lines (dots) discernable. But the image isn't really percieved to be sharper. Our perception of "sharpness" is based a lot more on contrast than it is based on actual detail.



An edited D4 image of the same owl/situation.





FaceBook  Twitter