Bill Pierce
Well-known
I believe we’ve gone over some of this before, but I just got asked twice about the advantage of high megapixel sensors. One of the questioners actually asked if the advantages were all PR BS. Indeed, he had heard that the images could look crappy unless you bought those expensive new lenses.
Glad to say he is wrong, but lets look at the real disadvantages of sensors with a lot of pixels and therefore smaller pixels. All other things being equal, there is going to be more noise in images from a sensor with smaller pixels, and, the higher the ISO, the more this will be obvious. How great this effect is varies from manufacturer to manufacturer. Guess what, some are better than others. The Sony A7R IV, the full frame camera with the highest megapixel count shows both noise and a green shift at its highest ISO’s. While it’s still essentially far ahead of what older and some current cameras do at these speeds and much of the effect is correctable or at least concealable in post production, if a huge amount of your shots are shot under “available darkness,” you are probably wiser to use a 24 megapixel camera instead of a 63 megapixel camera.
That’s about it. Your lenses aren’t going to get worse because you are using them on high megapixel sensor. At a pixel peeping level both the faults and the advantages of lenses will be more obvious. But when you make a print that won’t be true. Some early film lenses will have technically awful performance or character, however you choose to describe them. Most modern lenses will look the same.
So, why in the world would you want a camera with a high megapixel sensor? This I know I have written about here. It allows you to crop and still produce a print of high technical quality. In many kinds of photography - sport, street, news, even fashion and portraiture and, face it, family snap shots, that ability to tighten up the shot after you have taken it can be lifesaver. Leica realized that when they introduced the Q2 with a fixed lens and viewfinder frame lines for 28, 35, 50 and 75. But that same technique, achieved by combining a high megapixel sensor and a very good lens, minus the frame lines, is available to anybody with a high megapixel camera and really good lenses.
Any thoughts?
Glad to say he is wrong, but lets look at the real disadvantages of sensors with a lot of pixels and therefore smaller pixels. All other things being equal, there is going to be more noise in images from a sensor with smaller pixels, and, the higher the ISO, the more this will be obvious. How great this effect is varies from manufacturer to manufacturer. Guess what, some are better than others. The Sony A7R IV, the full frame camera with the highest megapixel count shows both noise and a green shift at its highest ISO’s. While it’s still essentially far ahead of what older and some current cameras do at these speeds and much of the effect is correctable or at least concealable in post production, if a huge amount of your shots are shot under “available darkness,” you are probably wiser to use a 24 megapixel camera instead of a 63 megapixel camera.
That’s about it. Your lenses aren’t going to get worse because you are using them on high megapixel sensor. At a pixel peeping level both the faults and the advantages of lenses will be more obvious. But when you make a print that won’t be true. Some early film lenses will have technically awful performance or character, however you choose to describe them. Most modern lenses will look the same.
So, why in the world would you want a camera with a high megapixel sensor? This I know I have written about here. It allows you to crop and still produce a print of high technical quality. In many kinds of photography - sport, street, news, even fashion and portraiture and, face it, family snap shots, that ability to tighten up the shot after you have taken it can be lifesaver. Leica realized that when they introduced the Q2 with a fixed lens and viewfinder frame lines for 28, 35, 50 and 75. But that same technique, achieved by combining a high megapixel sensor and a very good lens, minus the frame lines, is available to anybody with a high megapixel camera and really good lenses.
Any thoughts?
Rayt
Nonplayer Character
I didn’t think I needed a 40 megapixel camera until I got it. I have been doing street photography daily and need to shoot fast and often just zone at 1.2m with a 21 or 28mm. Often at the very least I need to crop to level the final image or to omit some distracting elements. There were times when I had to crop the 28mm image to a 50mm equivalent and glad I had the pixels. I do other types of photography with film where I compose precisely and never crop. It depends on your needs.
But print a 63mp photo at 24mp size (300dpi) and you are not going to see as much noise as you would at 63mp (300dpi).
Darthfeeble
But you can call me Steve
I would think if the noise is there, cropping would amplify the noise artifacts as they would become larger. Am I wrong?
JeffS7444
Well-known
I've found that cropping can also be used as a form of lens shift:
Because while Lightroom has perspective control adjustments, results can look odd if they're applied with a heavy hand.
Because while Lightroom has perspective control adjustments, results can look odd if they're applied with a heavy hand.
Rayt
Nonplayer Character
Mild perspective correction with LR looks good. More extreme correction requires the photographer to step back yielding a larger canvas to crop. Either method requires cropping. I just bought a cheap used Nikon 28mm PC to experiment to make use of live view now I have it.
Bill Pierce
Well-known
Yes, you are right. Just one more reason that a camera with fewer megapixels can be better for low level available light
Dogman
Veteran
At internet sizes, a crop from an 8mp camera can look good.
Ko.Fe.
Lenses 35/21 Gears 46/20
To me if you need to crop, you are not good enough.
Crop in street and family pictures? Why?
"You not good enough, if you are not close enough"
I understand where are some street hornies with teles and they might crop after to make legs looking closer.
Or OP is talking about surveillance on the street?
But where is the problem with family pictures?
Oh, got it. Cropping out ex.
Crop in street and family pictures? Why?
"You not good enough, if you are not close enough"
I understand where are some street hornies with teles and they might crop after to make legs looking closer.
Or OP is talking about surveillance on the street?
But where is the problem with family pictures?
Oh, got it. Cropping out ex.
BWF
Established
The only reason I got a higher megapickle camera was because it was the only way to get a higher resolution EVF. It makes it much easier to manually focus lenses, IMHO. In hindsight it’s really useful to be able to switch to the APS-C crop mode and get 1.5x crop on lenses, without any real loss of IQ. Also in hindsight, the focus assist on the newer Sony’s is just better anyway (thanks for the blue peaking color!).
Not at all. However, what I said is true too because you are not printing at anywhere near 100% size.
JeffS7444
Well-known
One of the joys of the A7R4 is the added "pop" that comes with elimination of the optical low-pass filter, and yet I'm rarely bothered by aliasing. And when we reach the point where sensors can out-resolve even the best optics, then the lenses themselves will effectively be low-pass filters.
agentlossing
Well-known
Something that often gets missed, however, is that observable noise can be substantially reduced by displaying an image at a more reduced size, and a 24MP image displayed at the same actual dimensions as an 8MP image will be hugely reduced in actual size. So whatever the benefits of the larger photosites, you are likely to see a cleaner image from the 24MP image. Cropping, of course, will magnify the noise.
Back in the day I saw an argument in favor of larger photosites, but having shot M4/3 for many years, I only ever saw improvements as the sensor megapixels increased from 12 to 16 to 20. Photosites obviously got smaller, but IQ improved, and noise was reduced. So one thing we need to keep in mind is improved technology, and the other is that noise becomes essentially invisible if it's small enough.
agentlossing
Well-known
Photographers crop where necessary. Not cropping at all is an arbitrary decision, if it works for your method then that's fine, but c'mon.
Darthfeeble
But you can call me Steve
From the beginning of my serious digital efforts I said that I needed full frame to crop as I'm not a very good Photographer, now I need mega for the same reason. I'm a nut for detail and when you add in a compulsive cropper the only answer is mega.
willie_901
Veteran
After data is recorded nothing can change the signal-to-noise ratio.
There are no miracles.
Nevertheless, there are useful methods that improve perceived image quality. These are incorrectly called noise reduction. All these noise filtering methods involve some type of mathematical pixel averaging. Pixels with higher SNR are averaged with pixels with lower SNR. These methods can be simple or highly sophisticated. AI methods make informed guesses about the missing signal for pixels in low-exposure (shadow) image regions. But the total image SNR does not change because the image information remains constant. Effective guesses about groups of noisy pixels does not create new information. It simulates missing information we wish we had recorded. However, a skillfully noise-filtered image often has superior perceived image quality.
Cropping in post production can affect how SNR influences perceived image quality. For example, if the crop is contains mostly shadow regions the perceived SNR will be inferior to a crop of the same image that is mostly highlight regions. Shadow regions have less signal. The noise is close to constant throughout the entire image. Increasing shadow-region crop brightness will result in an increase in perceived noise levels compared to viewing the uncropped image with an image brightness appropriate for brighter regions.
Note: capturing more of a scene than one actually uses followed by cropping can result in an inferior SNR compared to purposeful composition with no cropping. Shutter time and aperture estimates will reduce exposure for a higher scene area with uninteresting bright image regions. A purposefully composed image will result in shutter and aperture parameters that maximize exposure for image regions of interest with lower illumination. When the total scene illumination is low and the scene contains uninteresting brighter regions, the convenience of optimizing composition during post-production cropping has a price. The decrease perceived image quality for the needlessly underexposed - but desirable - darker regions increases.
Sensor fill factor (pixel area) has less effect than one might think. Time-dependent (psuedo-random) noise comes from two very different sources - photon noise and downstream electronic read noise. By now most cameras have very low read noise levels. Photon noise has nothing to do with the camera. Photon noise does not depend on sensor technology. Total image SNR is not greatly affected by sensor fill factor. Any differences in cropped images will be small. Time-independent, coherent artifacts (banding) are different. Perceived image quality perception can be affected by time-independent noise sources. This is particularly so for low lit scenes when darker regions are the regions of interest.
There are no miracles.
Nevertheless, there are useful methods that improve perceived image quality. These are incorrectly called noise reduction. All these noise filtering methods involve some type of mathematical pixel averaging. Pixels with higher SNR are averaged with pixels with lower SNR. These methods can be simple or highly sophisticated. AI methods make informed guesses about the missing signal for pixels in low-exposure (shadow) image regions. But the total image SNR does not change because the image information remains constant. Effective guesses about groups of noisy pixels does not create new information. It simulates missing information we wish we had recorded. However, a skillfully noise-filtered image often has superior perceived image quality.
Cropping in post production can affect how SNR influences perceived image quality. For example, if the crop is contains mostly shadow regions the perceived SNR will be inferior to a crop of the same image that is mostly highlight regions. Shadow regions have less signal. The noise is close to constant throughout the entire image. Increasing shadow-region crop brightness will result in an increase in perceived noise levels compared to viewing the uncropped image with an image brightness appropriate for brighter regions.
Note: capturing more of a scene than one actually uses followed by cropping can result in an inferior SNR compared to purposeful composition with no cropping. Shutter time and aperture estimates will reduce exposure for a higher scene area with uninteresting bright image regions. A purposefully composed image will result in shutter and aperture parameters that maximize exposure for image regions of interest with lower illumination. When the total scene illumination is low and the scene contains uninteresting brighter regions, the convenience of optimizing composition during post-production cropping has a price. The decrease perceived image quality for the needlessly underexposed - but desirable - darker regions increases.
Sensor fill factor (pixel area) has less effect than one might think. Time-dependent (psuedo-random) noise comes from two very different sources - photon noise and downstream electronic read noise. By now most cameras have very low read noise levels. Photon noise has nothing to do with the camera. Photon noise does not depend on sensor technology. Total image SNR is not greatly affected by sensor fill factor. Any differences in cropped images will be small. Time-independent, coherent artifacts (banding) are different. Perceived image quality perception can be affected by time-independent noise sources. This is particularly so for low lit scenes when darker regions are the regions of interest.
pluton
Well-known
This is the principle of oversampling, where, in some cases, the noise from the high MP sensor is less noticeable, because the little blobs of noise are smaller than the bigger blobs generated by the big pixels of the lower MP camera.
pluton
Well-known
I have a Nikon D800E camera, and six Zeiss ZF and Milvus lenses. Of those six lenses, only the 100mm f/2 "out resolves" the 36MP sensor. The other five lenses are, to varying degrees, low-pass filters.
Dogman
Veteran
The Henri Cartier-Bresson philosophy? I used to believe that crap too. Then I saw how Walker Evans cropped many of his photos. Some of his large format negatives he cropped with scissors. I just finished reading a book on Michael Schmidt and there were examples of how he cropped sections of photos, making more than one photograph from one negative. Many of the greatest photographers of the past cropped their photos and many of the best photographers of today crop their photos. While many a photo has been ruined by editors and their cropping tools, more have been improved by knowledgeable photographers working with their images, cropping to make them better.
So what if image quality declines with cropping. If it makes the photo better, no one will notice the image quality. Or care. Only those who get knots in their pixel panties over image quality pay attention to such things.
I find minor crops to be valuable, to clean up the edges sometimes (but I’d rather not if possible). However, when I try to fix a poorly composed image with a heavy crop it rarely works. If the balance is off, it tends to be off through out. Of course there are other reasons for heavy crops... I mean look at The Camera is God by Trent Parke.
https://www.magnumphotos.com/theory-and-practice/trent-parke-the-camera-is-god/
https://www.magnumphotos.com/theory-and-practice/trent-parke-the-camera-is-god/