|
The marketing race for "more megapixels" would like us
to believe that "more is better". Unfortunately, it's
not that simple. The number of pixels is only one of many factors
affecting image quality and more pixels is not always better. The
quality of a pixel value can be described in terms of geometrical
accuracy, color accuracy, dynamic range, noise, and artifacts. The
quality of a pixel value depends on the number of photodetectors
that were used to determine it, the quality of the lens and sensor
combination, the size of the photodiode(s), the quality of the
camera components, the level of sophistication of the in-camera
imaging processing software, the image file format used to store
it, etc. Different sensor and camera designs make different
compromises.
|
| |
Geometrical Accuracy
Geometrical or spatial accuracy is
related to the number of pixel locations on the sensor
and the ability of the lens to match the sensor resolution. The resolution
topic explains how this is measured at this site. Interpolation
will not improve geometrical accuracy as it cannot create what
was not captured.
|
| |
Color Accuracy
Conventional sensors using a color
filter array have only one photodiode per pixel location and
will display some color inaccuracies around the edges because
the missing pixels in each color channel
are estimated based on demosaicing algorithms. Increasing the
number of pixel locations on the sensor will reduce the
visibility of these artifacts. Foveon
sensors have three photodetectors per pixel location and
create therefore a higher color accuracy by eliminating the
demosaicing artifacts. Unfortunately their sensitivities
are currently lower than conventional sensors and the technology
is only available in a few cameras.
|
| |
Dynamic Range
The size of the pixel location and
the fill
factor determine the size of the photodiode and this has a
big impact on the dynamic
range. Higher quality sensors are more accurate and will be
able to output a larger dynamic
range which can be preserved when storing the pixel values
into a RAW
image file. A variant of the Fujifilm
Super CCD, the Super CCD SR uses two photodiodes per pixel
location with the objective to increase the dynamic range. A
more sensitive photodiode measures the shadows, while a less
sensitive photodiode measures the highlights.
|
| |
Noise
The pixel value consists of two
components:
(1) what you want to see (the actual measurement of the value in
the scene)
(2) what you do not want to see (noise).
The higher (1), and
the lower (2), the better the quality of the pixel. The quality of
the sensor and the size of its pixel locations have a great impact
on noise and how it changes with increasing sensitivity.
|
| |
Artifacts
Besides noise, there are many other
types of artifacts
that determine pixel quality.
|
| |
Conclusion
Unfortunately there is no single
standard objective quality number to compare image quality
across different types of sensors and cameras. For instance, a 3
megapixel Foveon type sensor uses 9 million photodetectors in 3
million pixel locations. The resulting quality is higher than a
3 megapixel but lower than a 9 megapixel conventional image and
it also depends on the ISO level you compare it at. Likewise, a
6 megapixel Fujifilm Super CCD image is based on measurements in
3 million pixel locations. The quality is higher than a 3
megapixel image but lower than a 6 megapixel image. A 6
megapixel digital compact image will be of lower quality than a
6 megapixel digital SLR image with larger pixels. To determine
an "equivalent" resolution is tricky at best.
End of the day, the most important thing is that you are happy
with the quality level that comes out of your camera for the
purpose that you need it for (e.g. website, viewing on computer,
printing, enlargements, publishing, etc.). I strongly recommend
that you look beyond megapixels when purchasing a digital
camera.
|
| |
|
|
