Resampling

The process of resampling an image can go in two directions - up or down, which happens when an image is sized. If you enlarge a specific image within Fireworks you don't want to just enlarge the size of the individual pixel elements, the mosaic stones. If you did that the result would be an image that is made up of visible squares. For example, having an image that is 100 by 100 pixels enlarged to be 400 by 400 pixels would result in the original pixel gaining the size of a 4 by 4 pixel square.

Here is a visual example when an image is enlarged:

	The circle here is on a 100 by 100 pixel canvas. The resolution is set to 72 ppi, that means 72 pixels make up an inch. This resolution shows the stroke clearly and the human eye is comfortable in recognizing the shape as a circle drawn with a soft crayola like marker.
	Rescaling the image to 400 % would result in a picture that looks like the image on the left. Instead of having a square if 1 pixel with one distinct color value the squares are now 4 pixels by 4 pixels, that means 16 times more of the same color. That is a change which does not escape our eye and we need to do something about it.
	Fireworks uses a mathematical process called resampling. It is a statistical model that adds additional pixels which lie in the range of gradients between neighboring pixels. The process of resampling is adding color information to the new pixels that are created when the image is enlarged. Therefore all of the 15 additional pixels will have an individual and distinct color which was found by interpolating the color values of the neighboring pixels.

Resampling also takes place when an image is scaled down. In this case, there is color information in surrounding pixels that has to be integrated into the color of fewer pixels. Let's think of the reverse process of scaling a 400 by 400 pixel image to 100 by 100 pixels. If we were to just omit three out of four pixels the image would loose many colors and look grainy.

Resampling again interpolates the pixels that are to be discarded and comes up with a new color for the fewer pixels that reflects a mixture of all the surrounding colors. Of course less pixels mean that less detail will be visible.

In both cases, enlarging or reducing, the image will lose quality. Even a sophisticated algorithm can only fill in the additional pixels in certain amounts, if the image is enlarged too much our eye will be able to see that it is an unnatural picture that was manipulated by a machine.

Therefore we can state a few guidelines about working with digital images:

• Always start with an image that has a higher resolution than what you want to use in your document. If you plan to have parts of the image enlarged have the resolution of the source image match your desired maximum resolution.
• Take into consideration that cropping images will make them smaller and if you want to fill the same space resampling will take place.
• Downsampling is a process that cannot be avoided. There is a finite amount of information that can be compressed into a digital format which is limited to 72 pixels per inch.