All technical aspects of digital photography
Discussion Last answer Replies / Clicks
Raymond Mckay

sharpness

thanks for the help
532 clicks
Ascanio Colonna di Paliano

How to choose a digital camera: megapixels

I'll try to make this as clear as possible.

When choosing what camera to buy, there are several factors that each one of us will use to weigh choices, which vary from person to person.
It is difficult, for amateurs, to understand what they are going to do, and what they are going to need, exactly.
For this reason it is good to bias the choice primarily on four things (in this order):

1) Budjet;
2) Performance;
3) Compatibility with owned equipment;
4) Handling and feeling;

What I want to help you on (for what I can), is judging performance.

Specs are useless (if not used along with other factors) and can easily confuse you.
In simple words: going through a store shelf and checking the pixel-count on the tags by the cameras is not a good way to understand their performance.
This is because the resolution of a camera indicates only the size of the image output, but not it's quality, nor the amount of detail (because images are interpolated, and the final result derives also from the quality of the camera's software).

If you haven't got the change to try the cameras you have to choose from, or check out sample images (preferably printed in a lab), a good way is to look at both the resolution and the sensor size (this way you can get to know how big the sensor's pixels are).

It easy to understand why an 8.2megapixel sensor sized 22.5*15mm (Canon EOS 20D - $1400) will probably fire better pictures than an 8.0megapixels sensor sized 8.8*6.6mm (Canon PowerShot Pro1 - $700), because bigger pixels pick more light and will have to work less to render the image, reducing drastically the so called signal-to-noise ratio.

Another important thing, though tricky, is to check the image processing system generation.
When picking between cameras of the same brand it's easier: the Canon EOS 20D has the latest DiG!C-II processor, whilst the Canon PowerShot Pro1 has the older DiG!C. Obviously the newer will probably render better results.

So assuming you can not give a look at test images, you have these two tools to understand a little more deeply the performance capabilities of digital cameras:

1) sensor size / megapixel count (the higher the result, the better);
2) image processor generation;

___________________________________________

Conclusion:
In general, when figuring how to spend a good lot of money, it's always better to check the actual output images, because there are in fact a lot of little factors that can influence the true performance of a camera, while not listed in the specs (the Canon EOS 20D, for example, has an analogue power system issue that produces visible artifacts when using high ISO settings).

If you can compare output images, then you should carefully look at the following things (amongst others):

1) Image detail:
If you are talking DSLR (with removable lenses), then you must know what the mounted lens is capable of, in order to separate camera and lens performance.
If checking on screen, you must zoom to 200% or 400% and cover an area with nice and small details, so you can judge how sharp the image is (both at the center and borders of the image).
An important thing is to shoot a light-haired pet in front of a dark background: a bad processing system will often give you a heavy red blur around light streaks on a dark background)
If checking on print, you must blow up to an A4 or A3 size. This way you can see both the amount of detail and how well the image is interpolated: ofted when blowing prints you will see an annoying denting on slanted lines.

2) Image contrast:
Again if the camera has removable lenses, you better know what the lens is capable of (best way is to use a fixed 50mm top quality lens, cheap and good).
You should have a few images covering the most common situations, both low and high contrast compositions (like a dark tree in front of a strong-lit sky).
What you have to look for is burned light areas (the least the better) and bleeding around such light areas (which is most given by the lens, but not only).

3) Image tone:
Here the best pick is a portrait, as the skin tones are the most difficult to get right. Best way to check is to print the image (at a lab) and give a look at the print in daylight.
Don't worry, you can often vary the camera's tone (plus it depends on the white balance in each picture).

4) Image saturation:
Here we have two different situations:
1) if you prefer shooting and printing, then a commercial camera will do, as they are set up to fullfill the consumer's wills of achieving colourful pictures.
2) if you intend post-processing your images, you are looking for a camera that outputs neutral, I'd say "dull", images, with low contrast and saturation, so that you have more space to play around later on with Photoshop.
In general the least the contrast/saturation the better, you can always pump-up your images or ask your lab to do it (often labs do it automatically).


This is about it.
With Internet it's quite easy to find pictures made with the cameras we want to compare, giving us the chance to understand how those cameras really work.
Once you have compared the performance you can pass on to other factors of your interest.

Hope this helps,
Ascanio.
09.05.05, 14:03
I'll try to make this as clear as possible.

When choosing what camera to buy, there are several factors that each one of us will use to weigh choices, which vary from person to person.
It is difficult, for amateurs, to understand what they are going to do, and what they are going to need, exactly.
For this reason it is good to bias the choice primarily on four things (in this order):

1) Budjet;
2) Performance;
3) Compatibility with owned equipment;
4) Handling and feeling;

What I want to help you on (for what I can), is judging performance.

Specs are useless (if not used along with other factors) and can easily confuse you.
In simple words: going through a store shelf and checking the pixel-count on the tags by the cameras is not a good way to understand their performance.
This is because the resolution of a camera indicates only the size of the image output, but not it's quality, nor the amount of detail (because images are interpolated, and the final result derives also from the quality of the camera's software).

If you haven't got the change to try the cameras you have to choose from, or check out sample images (preferably printed in a lab), a good way is to look at both the resolution and the sensor size (this way you can get to know how big the sensor's pixels are).

It easy to understand why an 8.2megapixel sensor sized 22.5*15mm (Canon EOS 20D - $1400) will probably fire better pictures than an 8.0megapixels sensor sized 8.8*6.6mm (Canon PowerShot Pro1 - $700), because bigger pixels pick more light and will have to work less to render the image, reducing drastically the so called signal-to-noise ratio.

Another important thing, though tricky, is to check the image processing system generation.
When picking between cameras of the same brand it's easier: the Canon EOS 20D has the latest DiG!C-II processor, whilst the Canon PowerShot Pro1 has the older DiG!C. Obviously the newer will probably render better results.

So assuming you can not give a look at test images, you have these two tools to understand a little more deeply the performance capabilities of digital cameras:

1) sensor size / megapixel count (the higher the result, the better);
2) image processor generation;

___________________________________________

Conclusion:
In general, when figuring how to spend a good lot of money, it's always better to check the actual output images, because there are in fact a lot of little factors that can influence the true performance of a camera, while not listed in the specs (the Canon EOS 20D, for example, has an analogue power system issue that produces visible artifacts when using high ISO settings).

If you can compare output images, then you should carefully look at the following things (amongst others):

1) Image detail:
If you are talking DSLR (with removable lenses), then you must know what the mounted lens is capable of, in order to separate camera and lens performance.
If checking on screen, you must zoom to 200% or 400% and cover an area with nice and small details, so you can judge how sharp the image is (both at the center and borders of the image).
An important thing is to shoot a light-haired pet in front of a dark background: a bad processing system will often give you a heavy red blur around light streaks on a dark background)
If checking on print, you must blow up to an A4 or A3 size. This way you can see both the amount of detail and how well the image is interpolated: ofted when blowing prints you will see an annoying denting on slanted lines.

2) Image contrast:
Again if the camera has removable lenses, you better know what the lens is capable of (best way is to use a fixed 50mm top quality lens, cheap and good).
You should have a few images covering the most common situations, both low and high contrast compositions (like a dark tree in front of a strong-lit sky).
What you have to look for is burned light areas (the least the better) and bleeding around such light areas (which is most given by the lens, but not only).

3) Image tone:
Here the best pick is a portrait, as the skin tones are the most difficult to get right. Best way to check is to print the image (at a lab) and give a look at the print in daylight.
Don't worry, you can often vary the camera's tone (plus it depends on the white balance in each picture).

4) Image saturation:
Here we have two different situations:
1) if you prefer shooting and printing, then a commercial camera will do, as they are set up to fullfill the consumer's wills of achieving colourful pictures.
2) if you intend post-processing your images, you are looking for a camera that outputs neutral, I'd say "dull", images, with low contrast and saturation, so that you have more space to play around later on with Photoshop.
In general the least the contrast/saturation the better, you can always pump-up your images or ask your lab to do it (often labs do it automatically).


This is about it.
With Internet it's quite easy to find pictures made with the cameras we want to compare, giving us the chance to understand how those cameras really work.
Once you have compared the performance you can pass on to other factors of your interest.

Hope this helps,
Ascanio.
485 clicks
To the
top