Add your thoughts in the question about megapixels in the comments.
The battle of megapixels. Thoughts from Radozhiva
Nowadays there are quite a few cameras with matrices with a very large number of pixels on board. Many amateur photographers are sure that the more pixels the better, but there are nuances and subtleties.
Some disadvantages of cameras with a large number of pixels:
- The more megapixels there are on the camera's matrix, the smaller these pixels are. In general, the smaller the pixel on the camera's matrix, the less light it can absorb, and the stronger the noise in the picture can appear. In general, a small pixel size (large number of megapixels) results in more noise. Fortunately, scientific progress does not stand still, and each new generation of cameras simultaneously increases the number of pixels and tries to reduce the level of noise produced. But there are certain limitations that are very difficult to go beyond.
- Cameras with a lot of pixels produce huge files, especially when shooting in RAW format. For example, files from a 36-megapixel monster Nikon D800 weigh about 70MB. The process of copying the shot material from a small photo shoot to a computer with an HDD with a hard drive can take tens of minutes.
- Processing images consisting of a large number of megapixels requires very powerful computers.
- On cameras with more megapixels, it is more difficult to achieve 'per pixel' sharpness. For super sharp images, you need use good resolution lenses... These lenses are usually very expensive. Therefore, it is customary to say that a matrix with a large number of pixels is more difficult to "feed" with ordinary lenses. Often times, closing the aperture well is enough to achieve a super sharp image.
- A matrix with a large number of pixels has a lower diffraction threshold. This means that on cameras with a large number of pixels, it is more difficult to achieve a sharp image with closed apertures. This is a serious problem with such cameras. More details about this can read here.
- When using matrices with a large number of megapixels, it is more difficult take off due to pixel-wise movement. On such cameras, you have to reduce the shutter speed to avoid grease.
- In fact, modern digital cameras do not have real pixels, in the usual sense of the word, you can read about a kind of `` wrapping '' pixels in the section about Bayer pattern.
Advantages of cameras with a large number of pixels:
- The possibility of a strong crop of the picture without much loss in image quality. This is very convenient for post-crop. But at the same time, in order to take advantage of this advantage, it is necessary that the original image be very well detailed, which in turn imposes restrictions on the optics.
- Ability to print on materials of a very large format, for example, on sheets of A1 format and more. This is a very strong point of such cameras. Sometimes this is the only serious argument in favor of cameras with a large number of MPs.
- In most cases, multi-pixel cameras have very wide dynamic range and large color depth, It is very important for accurate color reproduction and post photo processing.
- Cameras with high density sensors may not use an AA filter. At the same time, the sharpness of the image can significantly increase, but the possibility of obtaining moire in photographs also increases.
My experience:
As a photographer, personally for myself once and for all concluded that megapixels play a much smaller role than many people think about it. I very rarely need strong framing, I print canvas with wedding photos from my 12 MP camera without any problems, I have a huge number of printed A4 glossy photos from 6 megapixel cameras. Larger pixels allow the use of non-top-end lenses, and batch processing of photos takes much less time :).
On the topic of megapixels, I also advise you to look in the sections'Pixels and Subpixels", "Gigapixels'and'Megapixel Math'.
Material prepared Arkady Shapoval. Training/Consultations | Youtube | Facebook | Instagram | Twitter | Telegram
Thank you, it was interesting) The D90 itself, and the friend D40, but the picture is in his blood and he has an order of magnitude more good shots.
An interesting and informative article! Yesterday I was just comparing photos with d40 and d90 - sometimes the pictures are sharper and more vivid with d40 than with d90. But on the other hand, the d90 offers convenient functionality and a bunch of trendy things. And the race for megapixels is explained by the marketing strategy of camera manufacturers. In principle, 4 megapixels is quite enough for an A6 print, but the truth is for complex photo processing - cropping and subsequent deep retouching, 6 megapixels is not enough.
At the D40 itself, so far with a whale lens, I am completely satisfied with it, the only thing missing is a series with different exposures, what would HDR do)
Who prints from such cameras and where? if in a newspaper (a supposed place for small-sized shots), then megapixels are almost not important at all. And the size there is much smaller.
If the banner is 3x6m .. well, you understand.
The third option is internet placement. Noises are important here.
But thanks anyway for trying. Waiting for new articles!
If there is information, then from which cameras do posters print?
Avzhezh test, having confirmed the fact that it’s not in the foreign part of the city but in the third place and in the same area. The top two images look the same way even though the D40 has less than three trochies of light colors. I’m just watching Rock Yak before D300s. Change like metal and airtight housing, brackets and focus points. If you don’t know any camera, you can take it for yourself, and see my video card on my D40 on D300s. For me, it is important to image and dynamic sensor range. Chi on bagato vono vidriznyatsya at D300s vid D40? Do the D300s have the D3100?
DD quite seriously differs in the NIkon D40 Nikon 3100 from the D300s a little more than a step. Proof Link http://snapsort.com/compare/Nikon_D300S-vs-Nikon_D3100 и http://snapsort.com/compare/Nikon_D300-vs-Nikon_D40 Please note that the Nikon D40 has a little exposure meter, because in the semiautomatic device the photos will be slightly lighter.
Great article))
Creative photographer. Thank you for the article. Very interesting!!!
Really illustrative example.
Thank you for the article, Arkady! But in the review you touched on only one facet - noise and ISO. But I think the duel would not be complete if not to mention one more facet - the effect of diffraction. Take the same shooting conditions as in the previous test, but increase the f-number, not iso! Everyone knows that to increase sharpness and grip, you need to cover the diaphragm. But the tighter the aperture, the more diffraction appears (in the form of blurring and loss of small details). That is, there is a certain diffraction limit at which this very blur takes precedence over sharpness. And as it turns out, the size of the pixel is also of great importance here!
I experimentally established (having subtracted the theoretical calculations on one of the photo sites, now I’m trying to find it) that for my Nikon D100 (and its other 6MP brothers) this same limit occurs in the f11 region. Then sharply begins to fall. For d90 with its 12mp, this limit is in the f9 region.
For newfangled 16mp cameras, it’s completely at f7.1
For landscape painters, I think this is not the last value)
Thank you, we know for this, but the article was aimed precisely at the fact that the number of megapixels is not so important, especially for printing and viewing on a monitor. By the way, due to the diffraction and pixel size on the D700, the picture with the same lens is much sharper, since its pixels are larger than the 6MP line of old cameras.
you want to say that if I shoot not at 24 megapixels., but at 13 ... by reducing this in the camera settings, will I get a better picture?
Alex is a bunch of idiots who do not understand those and mate part ... .. everyone is saying that the pixels are a type of garbage. I have d3200 and 7000, which I sell ... so the 14bit 7th is not next to 12b 3200 * 6000 and 4000MP do their job, with the same parameters, detail, color, sharpness and noise are superior to d24 - I was shocked when I looked through and compared - here's the legend 7000k-old horse ala d7, the picture from the latter never pleased. Even on kit 90- 18 higher resolution and large pixels give just a fabulous picture, and if the tamron 55-17 is a complex scheme, then it is generally a feast for the eyes!
Eugene, and you do not take into account the fact that the d3200 is stupidly two years younger, respectively, has a newer matrix, more modern data processing algorithms?
If you don’t see that the D3200 is noisier than the D7000, this does not mean that there are a lot of idiots around
Alex, reduced the resolution in the settings, you will not increase the physical size of each pixel
> And as it turns out - the size of the pixel is also of great importance here!
Absolutely not!
The aperture at which the lens has maximum resolution is determined only by the lens itself. But the higher the resolution of the matrix, the more accurately you can determine this aperture. From this, different numbers are obtained.
If you make a 60 MP APS-C * matrix (technically it is quite possible), then according to your logic, the maximum sharpness will be at all at the open aperture. But this is not so. When aperture is 1-3 stops (depending on the lens), the sharpness will increase significantly. Yes, the diffraction is enhanced, but the aberrations are reduced, so that in the end the resolution increases.
Separately, it should be said about the depth of field. For 100% crop of the image and in truth, depth of field is inversely proportional to the square root of the number of megapixels. But for a picture inscribed on the screen or printed, the depth of field does not depend on the resolution. After all, all the same, the picture on the screen is compressed to less than 2 MP, and a 15x21 print is only 4 MP.
* - separately note that a 100% crop of an image from such a matrix will look wildly soapy, because to obtain pixel-by-pixel sharpness on it, you need a lens with a resolution of more than 200 lines (pairs of lines, if you will) per millimeter, and this is very difficult to find.
Everything is true about the lens resolution, but in any case, the impression is made and it is the impression that a smaller number of MPs gives a sharper picture, which is actually important for ordinary users.
Well, after all, nothing prevents to take a photo of at least 17 MP and reduce it to 6 MP :-)
Personally, I rarely overtake photos from RAW to JPEG with a resolution greater than 2304x1536, which is only 3,4 MP. Only if I plan to print in 20x30 or A4. But even here only 3600x2400 is needed, and this is 8 MP.
When moving pictures, you need to write about interpolation systems, algorithms that are responsible for size manipulation, etc. By the way, I also always leave a photo with a resolution of slightly more than 3mp for my memory, and often, if necessary, I print A4 from them.
Arkady, that is, a high-quality picture with 3-4 MP can be printed in A4 format? And what quality of the printed portrait will be "at the exit"?
“Separately, it should be said about the depth of field. For 100% crop of the image and in truth, depth of field is inversely proportional to the square root of the number of megapixels. " - whinnied until you drop. DOF is a concept related to the laws of optics, megapixels have nothing to do with it. Only the GEOMETRIC size of the matrix matters, not how many elements there are.
I am shooting on film. Then I scan to the HP SJ G3010. Scans are indistinguishable from digital originals starting from 6 MP. When the volume of the scan is more than 50 MP, the processing time on the computer sharply increases. All editors refuse to work with a volume of more than 100 MP.
All of the above is true when working with a “native” driver. Under Vuescan Pro, the scanner works quickly and without any problems even at 4800 dpi, giving a full frame of 35 mm transcripts a file of more than 24 MP of impeccable quality.
the main thing here is not megapixels, but the convenience of working with the camera, its speed, autofocus accuracy and so on. on this indicator, d40 loses d90, and he, in turn, d300. and the detail on the d90 is higher. noticeably.
Pixel density is also of great importance.
Marketing wars force manufacturers to cram as many diodes per square millimeter as possible. As a result, 8-12MP each is crammed into a “telephone” matrix the size of a match head, and instead of a picture, soap is obtained.
Here we are talking about mirrored cropped cameras.
here eurekis sounded (3x6 banners along the roads and not only). I apologize for the offtopic, but I'll tell you right away. It is for them that the file resolution for printing is 35-50 dpi - no more (when the vector does not work). And then the people love to stick 300 dpi - in the end, they will torture themselves and the printers. That is, the photo will give a ride to digital soap boxes. It is possible and more, but the printers themselves will throw off to the same 35-50 in most cases and will do the right thing: the viewer from three meters or more will not notice the difference very much - namely, from this distance it all looks normal. Interior printing is another matter. Here the problem is acute - interior designers, more often beginners, will twist on the walls of night cities - in their tiny 3D model everything looks great - there are a lot of small pictures in the internet, gnawed by a jeep. And then the client comes to the large-format printers ... And where can they get the night city with skyscrapers, in the windows of which all the details of the blonde who just came out of the shower are discernible? Netuti if you don't draw. Fortunately, in most cases, people understand that the main thing is the plot without the blonde on that 35th floor of that building is on the horizon, and the plot and the extended one works, and if not ... only the appearance of new cameras gives hope.
I had a Nikon D100 (6MP) + Sigma AF 24-70 / 2,8 D, then I bought a Nikon D90 (12,3MP).
At first I was disappointed, the portraits on the D100 looked nicer. But having orientated in the D90 settings, getting used to the camera, I will say that the sharpness / detail of the pictures is much higher than that of the D100 (with the same lens).
Maybe it's because of the CMOS matrix, or maybe 12 megapixels - I don't know.
And it also seemed that the D90's DD is wider.
Shot architecture / monuments D90 + Nikon AF 18-35 / 3,5-4,5 D ED, the pictures are very, very sharp within f5,6-16, f3,5 f22 is worse, this is a regularity of the lens operation, not diffraction , from the “bright thought” above it follows that f11 and f16 should have led to a loss of sharpness.
Here, diffraction also plays a role.
As for the tests in Dx0, I myself am not a fan of these tests, but sometimes I turn to them) So, they give 90 points to the D73 (http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Database/Nikon/D90), and D3200 - 81 (!). And in the latter, as we know, 24 (!!) Mp (http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Database/Nikon/D3200) They note low noise at high ISO in the D3200. What can you say to that?
Nice lyrics. D3200 refers to a newer round of camera development https://radojuva.com.ua/2012/02/evolyuciya-kamer-nikon/, it’s all logical that the matrix is better there.
Great article, Arkady)))
I completely agree on the count of MP.
The question is ripe as follows:
the camera has 10 MP in the arsenal, but the size of the photo when shooting is not maximum (medium or minimum)
What happens in this case? How does the camera use its 10 MP and produce a picture, for example in 6MP?
I take pictures for a family album, print 10x15 or 15x21 (A4 is extremely rare). It makes no sense for me to shoot huge pictures, but I can't figure out where these megapixels are going)) The camera simply does not use a certain number of MPs, or is saving the photo on the camera at a minimum size - is it nothing more than in-camera post-processing?
When shooting in JPEG, the image size is selected L - large, 10MP, M - medium, 6MP, S - small, 2MP. Also, the file compression level is set. High - low compression, Medium - medium, Basic - high. You can shoot in L size, but at the same time have the Basic compression level, the camera will create 10MP images, but their weight in megabytes will be small, since the camera's processor uses stronger compression with more data loss.
I asked the wrong question. I will try again:
When shooting in JPEG, the image size is S-small, the quality is maximum (low compression, without loss of details)
It is during shooting that the camera uses ALL of its 10 MP, then post-processing occurs (reducing the image to size S)
Or, at the moment the shutter is released, the camera does NOT use a certain number of MPs and the matrix “memorizes” six, not ten megapixels?
I'm just wondering how this happens.
Thank you very much for your reply )))))
All MPs are used when the shutter starts, and then the processor simply reduces the image using any algorithm (bicubic, Hoffmann, linear).
When discussing the battle of megapixels with my comrades, a dispute suddenly arose and I wanted to hear an opinion from the outside. What can you say about cameras like the Phase One 645AF or the Hasselblad H4D-60? Why are there so many megapixels?
Due to the size of the matrix.
Arkady, from your vast experience - have you come across a camera that gives the same stunningly sharp picture as d40? I have been trying to buy it for two months now - nowhere else / Instead of it, they offer cashmere 3100 and 3200, crammed under the top of the head with fake megapixels. Even d90 gives the picture dimmer and less sharp than d40 - this can be seen from this review. Is there some kind of intelligible replacement for d40, or is there no choice and do you need to take a d40?
Cameras of the d40, D70, d70s, d100, d1h, d1, d2h, d2hs types have a high diffraction threshold. Personally, on my D700 all 12mp are working, the picture is very sharp, but this is already a full frame.
Arkady, thanks a lot for the answer. Rummaged the entire Internet for information on the diffraction threshold. As a result, I dug an interesting link:
http://vladimirmedvedev.com/dpi.html
At the end of this article there is a very curious table containing the pixel sizes on the matrices of different DSLRs - According to this article, the D700 is the best FF for today, and among the crops, the championship is still behind the d40. D40 still found a new one in America, ordered it, I'm waiting for it to come with an opportunity. Arkady, thanks again for your answer. He helped me a lot.
There was still a question about the D3000. Is he really so seductive? The matrix is on it as on the D200. The pixel size is also decent, but for some reason this model was obstructed on professional sites. If you look at the reviews in Yandex Market, the situation is exactly the opposite: buyers spit on the d3100 and praise the d3000.
On D200, a matrix with two-channel signal readout, on D3000 and on D80 and D40x with single-channel. Why they don't like the camera - I can't tell, there are many different options.
Arkady, regarding the question of Vadim. You write that the camera uses all the megapixels, and then software compression goes through. I do not understand the logic. If in my D5100 it is possible to select an image size of 16,1 mp., And then as an image, the compression ratio. In your opinion, it turns out that if I select 9 mp. In the settings, then the camera will still use all 16 mp. And then do what I did in the first case using the quality selection (high, medium, low)? It seems to me that it’s more logical when choosing a smaller number of pixels in the settings to programmatically disable the unnecessary number and thereby reduce their density (which, according to the statements, affects the occurrence of noise). And in the quality settings already determine the degree of compression.
The pixel density will decrease - but their size will remain the same. The scheme for reading all pixels is more logical, since the organization of reading sampled pixels is more complicated, especially since the resizer program is easier to process the original file with a large amount of initial data.
I agree, the size of the pixels is the same, but the light entering the neighboring ones when closing the diaphragm will decrease. No?
Are you talking about the diffraction threshold?
Yes, Arkady, about him. As I understand it, the dialogue here is without practical use :) It’s not very interesting for readers. Thank you for taking the time to answer.
My personal opinion is that if your hands are straight, you can also make a good quality shot on a soap dish.
The number of pixels does not particularly affect the quality of the image, but optics. In most cases .
I have a SONY R1. I am a simple amateur photographer with a lot of photography experience and have always loved to shoot with a medium format film camera. He spun the film himself. And the hall that the quality of the image has always depended on the quality of the film, developers and optics. In digital it turned out to be different. But I realized one thing, that the number of MPs on the matrix does not always provide high quality of the original picture. My favorite format is A3 +. And SONY R1 coped with the tasks and continues to cope perfectly! The quality of photos printed on the “photo lab” is absolutely not inferior to the quality of photos taken by cameras of even famous brands, although it has only 10,3 megapixels.
Can I ask for the first two frames (D90, D40) in full size?
If I find, I will send links, for a long time it was.
Hi! Say it he, just embarrassed by the letter "X"
http://www.eldorado.ru/cat/detail/71006108/?show=response
Arkady, I read your reviews, but my head is spinning even more. I choose between D5100 / D3200 / D90. This will be my first DSLR. What do you recommend? Thank you!
Find selection tips here. https://radojuva.com.ua/2013/01/what-camera-nikon-shoud-i-buy/