How Many Pixels Do I Need?
Megapixel versus Resolution
How big a print can I get?
Mega Pixel versus Resolution
The common perception that "higher the mega pixel, better will be
the resolution". Interestingly this idea is not accurate. The word
resolution means the ability to capture small details. It depends on many
variables, like sensor technology, dark count, sensor size, optics and
so on!
It is true that starting with 640X400 pixel web cam, the resolution steadily
increases with the increase of pixel count, till it reaches a threashold.
According to my opinion, once you cross the level of 2 Mega pixel, higher
|
|
|
|
|
High density |
Medium density |
Low Density |
mega pixel does not directly translates into better resolution. There
after, the other variables like lens quality, Sensor size and sensor technology
play the vital role. A sensor manufacturer can put same number of sensor
pixels on a small sensor or a bigger sensor. So the smaller sensor will
have higher sensor density than the bigger sensor, but specification-wise
they are both same mega pixel sensor! Now take a look into these three photographs
of a birds eye. They are reproduced with three different pixel densities.
How the resolution of the lowest density looks against that of a higher
density? Same is true for a image sensor. But there is a snag with high
sensor density also. Higher the density, higher the noise, lower the light
sensitivity and magnification of lens aberrations! All these factors depend
on the manufacturing technology of the image sensor. So a lower megapixel
sensor can produce much high fidelity in color and resolution than a higher
megapixel sensor.
Dark Count and Dark Noise
All optical sensor suffers from this malady. What happens here is the
sensor produces electric charge on its own with time and elevated temperature.
This charge has nothing to do with the optical image. So even in total
darkness, the sensor produces signal, which appears as noise. It increases
with exposure time and ambient temperature. Professional Digital cameras
employ special circuit and firmware to cancel out some of this Dark Count
/ noise. That is why all the consumer level digital cameras are so noisy
in time exposure and low light situation. Unfortunately, no camera manufacturer
ever provides (at least to my knowledge) a Dark Count specification of
their sensor. So it is best to take few sample time exposure and low light
shot to check it yourself.
Lastly, but not least, look for the quality of the optics. You may not
believe your eyes what a lens quality can do to your photograph without
changing the megapixel resolution of your camera. This test can only be
performed in a Digital SLR like D30. Try to shoot the same subject using
same camera setting using two similar focal length lens
|
|
|
|
|
Quataray
70 210mm
140$
|
Sigma
70 200mm
750$
|
Canon
70 200mm
1350$
|
manufactured by different manufacturer, or even from the same manufacturer!
Though very unfair comparison, but to prove the point, see the table below,
same focal length lenses from different manufacturer targeted for different
consumer levels at the same camera store cost from 140$ to 1350$. Why
almost 10 times price difference? Mainly because of superior optical quality.
The image made with 140$ lens on same D30 will look very soft and unacceptable
when compared with picture taken with 1350$ lens. In this case image sensor
and rest of the electronics of the camera is common. For a fixed lens
camera, you are in no luck of ever replacing the lens. So if the lens
of a higher megapixel camera is of inferior quality, then you are much
better off with a lower megapixel camera with a better lens.
So the moral is, do not overemphasize on megapixel for cameras above
2.1 megapixel rating. Check for quality of image and optics, which will
pay back far more than the raw megapixel number.
How big a print I can make?
The photoprinter does not understand megapixel. It only cares about Dot
Per Inch or DPI. Whereas our computer monitor are generally set for displaying
72 DPI resolution. This resolution is too small for any photo printer,
as a photo realistic print requires at least 300 DPI resolution. If you
lower the DPI, you may get a bigger print, but the print quality will
degrade. It will show pixilation and jaggies at the edge, especially on
the curved or diagonal lines. There are, however, ways to make bigger
quality prints from a lower megapixel source by using special software
tools like Genuine Fractal. We will discuss it later in the FAQ.
A simple unitary method calculation gives you how big photo-realistic
prints you can get from your images. The key is considering each pixel
as a dot. Now let us do a simple sum, what will be a photoquality print
size from a 640x480 size image?:
A photoquality print dot takes 1/300 of a inch
space. |
Therefore 640 dots will take 640/300 = 2.13 inch |
and 480 dots will take 480/300 = 1.6 inch |
Hence 640x480 picture will make a 2.13x1.6 inch
photoquality print! |
Using same calculation, I have computed the maximum photoquality prints
achievable with various popular image sizes.
Some Popular Cameras within this range
|
Mega Pixel Rating
|
Image size
(in pixels)
|
Photoquality print Size
(in inches)
|
Agfa ePhoto 1280, CL30
Canon Powershot A10, A30, A50
Casio QV8000SX
Fuji FinePix A101, 1300, 1400, MX600, MX1700,
Nikon Coolpix 900
Olympus D100, D370, D460Z, D620L
Sony Mavica FD87, FD88
|
1.3
|
1280 X 960
|
4 X 3
|
Agfa ePhoto CL50
Canon Powershot A20, A40, A70, S10, S100,
S300, S330
Casio QV2000UX
Epson PhotoPC 850Z
Fuji FinePix 2300, 2400, 2600, 2800
Nikon Coolpix 800, 950, 2500
Olympus C2000Z, C2040, C2100, D230, E100RS
Sony Mavica CD1000, FD95, FD97,
|
2.2
|
1600 X 1200
|
5 X 4
|
Agfa ePhoto 1280, CL30
Canon Powershot G1, S20, S30, D30
Casio QV3000EX, QV3500EX
Epson PhotoPC 3100Z
Fuji FinePix 4700, 4800, 4900, MX2700,
Nikon Coolpix 880, 990, 885, 995
Olympus C3000Z, C3020Z, C3030Z, C3040Z
Sony Mavica CD300, DSC S70, S75
|
3.3
|
2048 x 1536
|
7 x 5
|
Fuji FinePix 6800, 6900
Minolta Dimage 7
Nikon Coolpix 5000
Olympus E20
Sony DSC F707,
|
5
|
2560 x 1920
|
8.5 x 6.4
|
|
|
|
|
|